memstead_base/engine/query.rs
1//! Engine read paths — accessors and queries.
2//!
3//! Read-only methods on `Engine`: store / schema / mount accessors,
4//! per-mem path helpers (`gitdir_for` / `worktree_for`), aggregated
5//! views (`communities`, `orphans`, `stubs`, `most_connected`,
6//! `missing_required_outgoing`), per-mem summaries (`health`,
7//! `status`, `context`), search (`list`, `search`,
8//! `search_indexes`), and the bytes-level read wrappers
9//! (`list_entities`, `read_entity`, `read_provenance`). Capability and
10//! cross-mem link gating live here too — they're consulted by
11//! handlers before any mutation reaches the backend.
12
13use std::cell::OnceCell;
14use std::collections::HashMap;
15use std::path::{Path, PathBuf};
16use std::sync::Arc;
17
18use memstead_schema::Schema;
19
20use crate::engine_fallback_type;
21use crate::entity::{Entity, EntityId};
22use crate::graph::{LouvainOutput, community::detect_communities};
23use crate::mem::MemRouterSnapshot;
24use crate::ops::{ContextResult, Direction, NeighborInfo, SearchResult, SearchScope, WarningHint};
25use crate::provenance::Provenance;
26#[cfg(not(target_arch = "wasm32"))]
27use crate::search_index::{MemIndex, build_all};
28use crate::store::Store;
29use crate::workspace::{MountCapability, MountStorage, WorkspaceSettings};
30
31use super::{BackendFactory, Engine, EngineError, MountedBackend};
32
33impl Engine {
34 /// In-memory store populated at construction time from every
35 /// mount's backend. Read-only at this point in the rebuild —
36 /// mutation paths land in a later session.
37 pub fn store(&self) -> &Store {
38 &self.store
39 }
40
41 /// Per-mem schema, keyed by mount's mem name. Each entry is the
42 /// schema resolved from that mount's pin at boot, so the map holds
43 /// genuinely heterogeneous schemas in a multi-schema workspace.
44 pub fn schemas(&self) -> &HashMap<String, Arc<Schema>> {
45 &self.schemas
46 }
47
48 /// Workspace-authored schemas loaded from
49 /// `WorkspaceSettings.schemas_dir` at construction. Distinct from
50 /// [`Self::schemas`] (per-mem, only schemas pinned by a mount):
51 /// this slice carries every workspace-loaded schema regardless of
52 /// whether a mem pins it. Used by `memstead_overview` to enumerate
53 /// schemas referenced by `mem_create_rules.schemas[]` but not
54 /// pinned by any mem — agents see what could be pinned without
55 /// looking up the workspace.toml directly.
56 pub fn workspace_schemas(&self) -> &[Arc<Schema>] {
57 &self.workspace_schemas
58 }
59
60 /// Embedded built-in schemas loaded once at boot. Handlers
61 /// resolving a schema pin by `<name>@<version>` (MCP's `memstead_schema`,
62 /// `memstead_overview` rendering) walk mem-pinned, workspace, and
63 /// built-in catalogues in order — built-ins are the catch-all when
64 /// no mem or workspace dir pins the schema. Workspace schemas
65 /// shadow built-ins on `(name, version)` collision; resolve from
66 /// `workspace_schemas()` first.
67 pub fn builtin_schemas(&self) -> &[Arc<Schema>] {
68 &self.builtin_schemas
69 }
70
71 /// Classify a schema's trust origin — the single authority every read
72 /// surface consults before serving a schema's instruction-prose.
73 ///
74 /// A schema is [`OriginClass::FirstParty`] iff it is an engine built-in
75 /// **or** pinned by a writable mount in this workspace. Built-ins are
76 /// compiled into the binary — unforgeable. A non-built-in schema earns
77 /// first-party status only once the operator *adopts* it by writably
78 /// mounting a mem that pins it: writing into a mem is the act that
79 /// legitimately needs a schema's authoring prose (`system_message`,
80 /// `write_rules`, …), and the mount's writable posture is set by the
81 /// consumer's own config — a publisher cannot forge it.
82 ///
83 /// Everything else is [`OriginClass::ThirdParty`]: a schema present in
84 /// the catalogue but pinned only by read-only mounts (a registry-
85 /// installed read-mem or an adopted foreign folder/clone), or one the
86 /// engine cannot vouch for at all. Its prose is served structural-only
87 /// so a stranger's free-text never reaches a consuming agent as
88 /// instructions. This classifies by the mount graph — never by scanning
89 /// the schema's content, which a publisher controls — and `ThirdParty`
90 /// is the safe default for any ambiguous origin.
91 ///
92 /// Note a read-only mount pinning a *built-in* schema (e.g. a registry
93 /// mem on `default@1.0.0`) resolves to the consumer's own clean copy
94 /// and stays first-party — the de-framing targets only foreign,
95 /// non-built-in schemas that no writable mem has adopted.
96 pub fn schema_origin(&self, schema: &Arc<Schema>) -> crate::render::OriginClass {
97 use crate::render::OriginClass;
98 let (name, version) = schema.id();
99 // Built-in schemas are compiled in — first-party, unforgeable.
100 let is_builtin = self.builtin_schemas.iter().any(|s| {
101 let id = s.id();
102 id.0 == name && id.1 == version
103 });
104 if is_builtin {
105 return OriginClass::FirstParty;
106 }
107 // Adoption signal: some writable mount pins this exact schema, so
108 // the operator authors against it here.
109 let canon = format!("{name}@{version}");
110 let pinned_by_writable = self.mounts().iter().any(|m| {
111 m.schema.as_ref().map(|s| s.to_string()).as_deref() == Some(canon.as_str())
112 && self.mem_router().is_writable(&m.mem)
113 });
114 if pinned_by_writable {
115 OriginClass::FirstParty
116 } else {
117 OriginClass::ThirdParty
118 }
119 }
120
121 /// Classify a mem's *data* trust origin — the authority every read
122 /// surface consults before serving an entity's content (bodies,
123 /// snippets, titles). A writable mount is [`OriginClass::FirstParty`]:
124 /// its content is authored in this workspace. Anything else — a
125 /// read-only mount (a registry-installed read-mem or an adopted
126 /// foreign folder/clone) or an unknown mem — is
127 /// [`OriginClass::ThirdParty`], so the consuming agent/host treats the
128 /// content as quoted, untrusted data.
129 ///
130 /// This reads the deployment's declaration when one exists (see
131 /// [`Self::declare_mem_origin`]), else the mount's already-decided
132 /// writable/read-only posture (fixed at adopt/mount time) — it never
133 /// scans content, and both levers are consumer-side config, so a
134 /// publisher cannot forge first-party. Distinct from
135 /// [`Self::schema_origin`], which governs a schema's
136 /// instruction-prose: the data channel and the instruction channel
137 /// are separate vectors with separate authorities.
138 pub fn mem_origin_class(&self, mem: &str) -> crate::render::OriginClass {
139 if let Some(declared) = self.declared_origins.get(mem) {
140 return *declared;
141 }
142 if self.mem_router().is_writable(mem) {
143 crate::render::OriginClass::FirstParty
144 } else {
145 crate::render::OriginClass::ThirdParty
146 }
147 }
148
149 /// Declare a mem's data-trust origin as a deployment fact — the
150 /// embedding process (a curated hosted read tier, an app that vouches
151 /// for a bundled mem) overrides the writability inference for one mem.
152 /// Composition-layer-only by design: not persisted, not reachable over
153 /// MCP, never derived from mem content — the operator running the
154 /// process is the only authority that can set it, so a served mem the
155 /// deployment does *not* vouch for keeps reporting third-party on
156 /// every surface. (Deliberately absent from the CLI: that surface
157 /// operates a workspace, not a deployment; the CLI counterpart would be
158 /// a workspace-config knob no use case demands yet.)
159 pub fn declare_mem_origin(
160 &mut self,
161 mem: impl Into<String>,
162 origin: crate::render::OriginClass,
163 ) {
164 self.declared_origins.insert(mem.into(), origin);
165 }
166
167 /// Per-file errors collected during load. Non-fatal: the engine
168 /// continues with whatever did parse. Empty when every backend's
169 /// content parses cleanly.
170 pub fn load_errors(&self) -> &[(PathBuf, String)] {
171 &self.load_errors
172 }
173
174 /// Resolve a VISIBLE mem to its folder-storage root on disk.
175 ///
176 /// Unknown or quarantined names refuse `UNKNOWN_MEM` (the same
177 /// visibility gate `search` and the conflicts door apply); a
178 /// visible mount whose storage is not folder-backed returns
179 /// `Ok(None)` so callers branch on backend applicability without
180 /// inventing a typed error. Read-only mounts resolve too — this is
181 /// a read accessor, not a write gate. Generic by design: any
182 /// consumer that needs a folder mem's disk root (per-mem changelog
183 /// readers, export tooling, future doors) gets the same answer.
184 pub fn folder_mem_root(&self, mem: &str) -> Result<Option<PathBuf>, EngineError> {
185 let mount = self
186 .mounts
187 .iter()
188 .find(|m| m.mount.mem == mem)
189 .ok_or_else(|| self.unknown_mem_error(mem))?;
190 if self.quarantine_reason(mem).is_some() {
191 return Err(self.unknown_mem_error(mem));
192 }
193 match &mount.mount.storage {
194 crate::workspace::MountStorage::Folder { path } => Ok(Some(path.clone())),
195 _ => Ok(None),
196 }
197 }
198
199 /// Workspace-level operator policy (mem create/delete rules,
200 /// cross-mem links). Defaults to empty; populated via
201 /// [`Engine::set_settings`] after construction. Surfaced for MCP
202 /// handlers (`memstead_health { include_config: true }`,
203 /// `memstead_overview`'s lifecycle-namespaces section) and other
204 /// consumers that need to read workspace policy.
205 pub fn settings(&self) -> &WorkspaceSettings {
206 &self.settings
207 }
208
209 /// The pipeline configs — the v2 single-record binding store — loaded
210 /// from the workspace at boot: the read-only queryable surface the
211 /// loader exposes. Empty for engines not booted from a workspace root,
212 /// or for a workspace that declares no pipelines. The ingest skill
213 /// and future MCP tools consume this structured form
214 /// rather than re-reading the JSON folders.
215 pub fn pipeline_configs(&self) -> &crate::pipeline_store::BindingConfigs {
216 &self.pipeline_configs
217 }
218
219 /// The pipeline configs serialized as a JSON string — the read
220 /// counterpart of the `add_projection_json` edit entry point.
221 /// Serialization-boundary callers (where serde does not live)
222 /// get the store in one call and deserialize on their side.
223 ///
224 /// Shape: `{ "bindings": [{ mem, name, config }] }` — the v2
225 /// single-record store (`config` carries the whole binding: inline
226 /// `sources`, `operations`, everything). The `mediums` / `facets` /
227 /// `ingests` keys are **gone** with their record kinds. This reads the
228 /// live binding store fresh (like the brief path) rather than the
229 /// in-memory snapshot, so an edit shows back immediately. A missing
230 /// root or a legacy/unreadable store yields the fallback empty object.
231 pub fn pipeline_configs_json(&self) -> String {
232 let empty = || "{\"bindings\":[]}".to_string();
233 let Some(root) = self.workspace_root() else {
234 return empty();
235 };
236 match crate::pipeline_store::load_pipeline_configs(root) {
237 Ok(configs) => serde_json::to_string(&configs).unwrap_or_else(|_| empty()),
238 Err(_) => empty(),
239 }
240 }
241
242 /// Overwrite the in-memory pipeline configs. The workspace-root boot
243 /// paths call this after [`crate::pipeline_store::load_pipeline_configs`];
244 /// exposed so the full boot helper (a separate crate) can populate the
245 /// same surface.
246 pub fn set_pipeline_configs(&mut self, configs: crate::pipeline_store::BindingConfigs) {
247 self.pipeline_configs = configs;
248 }
249
250 /// Build a [`WarningHint::NoteMissing`] when the workspace has
251 /// `[mutations].require_notes = true` and the caller omitted (or
252 /// passed a blank/whitespace-only) `note`; `None` otherwise.
253 ///
254 /// This is the single enforcement point for the `require_notes`
255 /// provenance nudge. Every mutation that accepts a `note` calls it
256 /// on its commit-landing path and pushes the result onto the
257 /// outcome's `warnings`, so both the CLI and the MCP transports
258 /// inherit identical behaviour from the engine response rather than
259 /// each re-deriving the policy at its own boundary (the drift that
260 /// left the policy decorative on the CLI). `tool` becomes the
261 /// warning's `details.tool` — callers pass the engine-level verb
262 /// (`create_entity`, `update_entity`, `relate_entity`,
263 /// `delete_entity`, `rename_entity`, `create_mem`,
264 /// `delete_mem`), matching the commit `Tool:` provenance trailer.
265 /// The mutation still commits — the policy nudges, it never blocks.
266 pub fn note_missing_warning(&self, tool: &str, note: Option<&str>) -> Option<WarningHint> {
267 if !self.settings.mutations.require_notes.unwrap_or(false) {
268 return None;
269 }
270 let has_note = note.map(|n| !n.trim().is_empty()).unwrap_or(false);
271 if has_note {
272 return None;
273 }
274 Some(WarningHint::NoteMissing {
275 tool: tool.to_string(),
276 })
277 }
278
279 /// Backend factory currently installed on this engine. Returned by
280 /// value because [`BackendFactory`] is a function pointer (`Copy`).
281 /// Used by [`crate::mem_management::create_mem`] to materialise
282 /// the backend for a freshly-registered mount; consumers that need
283 /// to instantiate a backend ad-hoc can call this directly.
284 pub fn backend_factory(&self) -> BackendFactory {
285 self.backend_factory
286 }
287
288 /// Git-branch ops bundle currently installed on this engine.
289 /// `None` on lean-flavor engines that don't see mem-repo
290 /// mounts. Returned by value because [`super::GitBranchOps`] is
291 /// `Copy`. `create_mem` reaches for
292 /// the bundle to drive `prune_residue` against an unmounted
293 /// gitdir when the `ForceOverwrite` recovery action is selected.
294 pub fn git_branch_ops(&self) -> Option<super::GitBranchOps> {
295 self.git_branch_ops
296 }
297
298 /// Convenience: look up a parsed entity by id. Returns `None` for
299 /// unknown ids, including stub entries created for unresolved
300 /// inline-link targets — callers that want to distinguish real
301 /// from stub branch on `Entity::stub`.
302 pub fn get_entity(&self, id: &EntityId) -> Option<&Entity> {
303 self.store.get(id)
304 }
305
306 /// The stored provenance anchors for `id`, read from its mem's
307 /// anchors sidecar. Empty for an entity with none, an unknown mem, or
308 /// a backend that does not persist anchors (a pre-anchor archive / any
309 /// sealed read-only mount). Additive read surface (E3a): the
310 /// resolution *model* lives in [`crate::anchor`]
311 /// ([`crate::anchor::resolve_anchor`] / [`crate::anchor::compose_entity_anchors`]);
312 /// the live per-anchor *state* (which requires observing the source
313 /// artifacts through the medium/preparation pipeline) is E3b's concern.
314 /// The anchors sidecar's parse error for `mem`, if it has one.
315 ///
316 /// The anchor readers below degrade a malformed sidecar to "no anchors",
317 /// which keeps a read path alive but makes a corrupt file
318 /// indistinguishable from an empty one. For a *reader* that is the right
319 /// trade; for anything that draws a conclusion from the absence of
320 /// anchors it is not — a fidelity pass would report every artifact
321 /// uncovered and call it a finding, when the truth is that it could not
322 /// read the file. Callers that need that distinction ask here first and
323 /// refuse. `None` means the sidecar is absent (legitimately no anchors
324 /// yet) or parses cleanly; the binding store draws the same distinction
325 /// with its quarantine path.
326 pub fn anchors_sidecar_error(&self, mem: &str) -> Option<String> {
327 let mount = self.mounts.iter().find(|m| m.mount.mem == mem)?;
328 // Three distinct ways to be unreadable, and only one of them is a
329 // parse error. `.ok().flatten()` would collapse the first into
330 // "absent", which is the very confusion this exists to prevent.
331 let bytes = match mount.backend.read_anchors_sidecar() {
332 // A backend error — permission denied, an IO fault. The file may
333 // be perfectly well-formed; we simply could not look at it.
334 Err(e) => return Some(format!("could not read the sidecar: {e}")),
335 // Genuinely absent: a mem with no anchors yet. Not an error.
336 Ok(None) => return None,
337 Ok(Some(b)) => b,
338 };
339 // An empty or whitespace-only file parses as "no anchors" by a
340 // deliberate tolerance in `from_bytes`. That tolerance is right for a
341 // reader and wrong here: a sidecar truncated to zero by an
342 // interrupted write is not a mem that never had anchors.
343 if bytes.iter().all(|b| b.is_ascii_whitespace()) {
344 return Some(
345 "the sidecar file is empty — an interrupted write leaves this state, and it is not the same as having no anchors; remove the file if the mem genuinely has none"
346 .to_string(),
347 );
348 }
349 match crate::anchor::AnchorSidecar::from_bytes(&bytes) {
350 Ok(_) => None,
351 Err(e) => Some(e.to_string()),
352 }
353 }
354
355 pub fn entity_anchors(&self, id: &EntityId) -> Vec<crate::anchor::Anchor> {
356 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == id.mem()) else {
357 return Vec::new();
358 };
359 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
360 return Vec::new();
361 };
362 match crate::anchor::AnchorSidecar::from_bytes(&bytes) {
363 Ok(sc) => sc.get(id.as_ref()).to_vec(),
364 // Deliberate degrade-to-empty for the read path; a caller that
365 // must not confuse "unreadable" with "none" checks
366 // [`Self::anchors_sidecar_error`] first.
367 Err(_) => Vec::new(),
368 }
369 }
370
371 /// The stored anchors for `id`, each paired with its **live** resolution
372 /// state when the engine could observe the source artifact this pass.
373 ///
374 /// Additive over [`Self::entity_anchors`]: the durable data is unchanged;
375 /// `state` is the [`crate::anchor::resolve_anchor`] outcome against an
376 /// observation the engine produces here. A `path`-namespace anchor
377 /// (codebase / filesystem / git) is observed against the working tree at
378 /// the current HEAD; an `entity`-namespace anchor is observed against the
379 /// live graph by [`Self::observe_entity_anchor`]:
380 ///
381 /// - artifact absent ⇒ [`AnchorState::Orphaned`](crate::anchor::AnchorState::Orphaned);
382 /// - artifact present, non-hash class (`authored` / `informed-by`) ⇒
383 /// [`Resolves`](crate::anchor::AnchorState::Resolves);
384 /// - artifact present, hash-bearing class (`anchored` / `derived`) ⇒ the
385 /// prepared-content hash comparison decides:
386 /// [`Resolves`](crate::anchor::AnchorState::Resolves) on a match,
387 /// [`Drifted`](crate::anchor::AnchorState::Drifted) on a stable-medium
388 /// mismatch, [`Recheck`](crate::anchor::AnchorState::Recheck) on an
389 /// unstable medium or when a hash is unavailable on either side (a
390 /// hash-less anchor, a `tree` grain, an unreadable artifact).
391 ///
392 /// For an `entity` grain the same table applies, read off the store
393 /// rather than the filesystem: the entity missing (or present only as a
394 /// stub) is `Orphaned`, and a hash-bearing class compares the canonical
395 /// rendered markdown.
396 ///
397 /// `state` is `None` (unobserved — never a fabricated state) when there is
398 /// no workspace root, when the grain is `url` (the engine never fetches;
399 /// a url anchor's hash is the registry's prepared form of the content
400 /// its observer supplied at write time), when an `entity` anchor's
401 /// source declares a preparation the registry does not know (the form
402 /// cannot be computed), or when an `entity` anchor's mem is **not
403 /// mounted** — an unmounted mem is not a mem of deleted entities, and
404 /// saying so would route a deletion proposal to prune.
405 pub fn entity_anchors_resolved(&self, id: &EntityId) -> Vec<ResolvedAnchor> {
406 let anchors = self.entity_anchors(id);
407 let source_roots = self.anchor_source_roots(id.mem());
408 anchors
409 .into_iter()
410 .map(|anchor| {
411 let observed = self.observe_anchor(&anchor, &source_roots);
412 let (state, observed_hash) = match observed {
413 Some((state, hash)) => (Some(state), hash),
414 None => (None, None),
415 };
416 ResolvedAnchor {
417 anchor,
418 state,
419 observed_hash,
420 }
421 })
422 .collect()
423 }
424
425 /// Per-anchor observation — THE one resolution mechanism, shared by
426 /// binding-backed verify (`mem_anchors_resolved`, which the ingest
427 /// render/report/prune/findings paths consume), the per-entity read
428 /// (`entity_anchors_resolved`), and the standalone
429 /// `verify_mem_anchors` operation. Path-shaped grains
430 /// (`span`/`file`/`tree`) observe under the **decision-29 candidate
431 /// priority** (backlog-sweep plan 03a): the anchor's artifact path is
432 /// SOURCE-relative first — when its `source` name resolves through
433 /// `source_roots` to a declared pointer, the pointer-joined path is
434 /// authoritative — and workspace-relative only as the fallback, tried
435 /// when the source-join does not resolve. A path resolving under both
436 /// joins is decided by that priority, deterministically. An anchor
437 /// without a `source` (a hand-authored mem, a binding-less write)
438 /// observes workspace-relative exactly as before. A `url`
439 /// grain has no engine-side observation and returns `None` (the report
440 /// vocabulary's `unresolvable`) — the engine never fetches; its hash is
441 /// recorded from observation-supplied content at write time — as does a
442 /// workspace-root-less engine. An `entity`
443 /// grain is not a filesystem path but is not unobservable either — it
444 /// resolves against the live graph via
445 /// [`Self::observe_entity_anchor`], under the preparation its source
446 /// declares (touchpoint A of [`crate::preparation`]: the registry
447 /// decides the prepared form the artifact hashes as). This replaces the retired `single_path_medium_root` gate,
448 /// whose single-source assumption nulled every anchor of a mem with
449 /// zero or several bindings — the honest per-anchor answer supersedes
450 /// the all-or-nothing mem-level one.
451 fn observe_anchor(
452 &self,
453 anchor: &crate::anchor::Anchor,
454 source_roots: &std::collections::BTreeMap<String, AnchorSourceJoin>,
455 ) -> Option<(crate::anchor::AnchorState, Option<String>)> {
456 let join = anchor
457 .source
458 .as_deref()
459 .and_then(|name| source_roots.get(name));
460 // An `entity`-grain anchor points into a mem's graph, not a file
461 // tree. It has always returned `None` here — "unobserved this pass" —
462 // which meant it could never be drifted, never be orphaned, and always
463 // blocked prune. That is the bail the S1b pilot demonstrated: a
464 // deliberately stale anchor over a changed source entity went unflagged
465 // while the capability matrix claimed full parity.
466 if anchor.grain == crate::anchor::AnchorGrain::Entity {
467 return self.observe_entity_anchor(anchor, join.and_then(|j| j.preparation.as_deref()));
468 }
469 let root = self.workspace_root.as_deref()?;
470 observe_path_anchor(root, anchor, join)
471 }
472
473 /// Observe an `entity`-grain anchor against the live graph — the entity-
474 /// namespace counterpart of [`observe_path_anchor`], and the mechanism
475 /// that makes a graph-medium binding's drift real.
476 ///
477 /// The artifact is an entity id. Present/absent comes from the store, so
478 /// this works uniformly across backends — a git-branch mem has no
479 /// working-tree file to stat, which is exactly why observation cannot go
480 /// through the filesystem here. The compared form is the preparation
481 /// registry's **prepared form** for `preparation` (the anchor's source's
482 /// declared preparation, [`crate::preparation::entity_prepared_hash`]):
483 /// the **canonical rendered markdown** when the source declares none —
484 /// byte-for-byte today's form — or the load-bearing serialization under
485 /// `entity-load-bearing`; hashed with the same `prepared_content_hash`
486 /// the path arm uses, so an anchor's recorded hash means the same thing
487 /// in both namespaces. An identifier the registry does not know cannot
488 /// be prepared: the anchor is reported unobserved (`None`), never hashed
489 /// under a fabricated form.
490 ///
491 /// A stub is treated as absent: a stub is the engine's placeholder for an
492 /// unresolved reference, not the entity the anchor claims to pin. Scoring
493 /// it as present would let a dangling anchor resolve clean.
494 fn observe_entity_anchor(
495 &self,
496 anchor: &crate::anchor::Anchor,
497 preparation: Option<&str>,
498 ) -> Option<(crate::anchor::AnchorState, Option<String>)> {
499 let id = EntityId::canonical(&anchor.artifact);
500
501 // The mem the anchor points into must be MOUNTED before a store miss
502 // can mean anything. If it is not, every entity in it is missing from
503 // the store, and reporting `Orphaned` would say "the source deleted
504 // these" about entities sitting untouched on disk. `None` — genuinely
505 // unobserved — is the honest answer.
506 //
507 // This guard lives here, at the one observation site, and not at the
508 // callers. An earlier fix put it in `run_verify` alone; `prune` reaches
509 // anchor resolution by its own path (`mem_anchors_resolved`), so the
510 // sync brief went on proposing the deletion of every destination
511 // entity — a data-loss suggestion routed to the graph's only
512 // maintenance writer, from a mem merely being unmounted. A guard that
513 // protects one caller is not a guard on the behaviour.
514 if !self.mounts.iter().any(|m| m.mount.mem == id.mem()) {
515 return None;
516 }
517
518 let entity = self.store.get(&id).filter(|e| !e.stub);
519 let Some(entity) = entity else {
520 return Some((
521 crate::anchor::resolve_anchor(anchor, &crate::anchor::ArtifactObservation::Absent),
522 None,
523 ));
524 };
525 let current_hash = if anchor.class.is_hash_bearing() {
526 let type_def = self
527 .schema_for(id.mem())
528 .and_then(|schema| schema.get_type(&entity.entity_type));
529 Some(crate::preparation::entity_prepared_hash(
530 entity,
531 type_def.as_deref(),
532 preparation,
533 )?)
534 } else {
535 None
536 };
537 let observation = crate::anchor::ArtifactObservation::Present {
538 current_hash: current_hash.clone(),
539 };
540 Some((
541 crate::anchor::resolve_anchor(anchor, &observation),
542 current_hash,
543 ))
544 }
545
546 /// The `source name → join` map for `mem`'s bindings: the filesystem
547 /// roots that anchors written in the source dialect join onto (decision
548 /// 26: anchor artifact paths are source-relative first) and the
549 /// preparation each source declares (touchpoint A: what the registry
550 /// prepares the artifact as before hashing). Empty when the workspace
551 /// has no root, the pipeline store does not load, or `mem` has no
552 /// bindings — resolution then degrades to the workspace-relative dialect
553 /// alone with no preparation, which is exactly the hand-authored-mem
554 /// posture.
555 pub(crate) fn anchor_source_roots(
556 &self,
557 mem: &str,
558 ) -> std::collections::BTreeMap<String, AnchorSourceJoin> {
559 let mut roots = std::collections::BTreeMap::new();
560 let Some(root) = self.workspace_root.as_deref() else {
561 return roots;
562 };
563 let Ok(configs) = crate::pipeline_store::load_pipeline_configs(root) else {
564 return roots;
565 };
566 for record in configs.bindings.iter().filter(|r| r.mem == mem) {
567 for source in &record.config.sources {
568 roots
569 .entry(source.name.clone())
570 .or_insert_with(|| AnchorSourceJoin {
571 pointer: source.pointer.clone(),
572 preparation: source.preparation.clone(),
573 source: source.clone(),
574 deny_paths: record.config.deny_paths.clone(),
575 });
576 }
577 }
578 roots
579 }
580
581 /// Reverse anchor lookup: every `(entity_id, anchor)` across all mems
582 /// whose anchor references `artifact_path`. This is the query the
583 /// rebuilt check-realization hook consumes — given the file an agent
584 /// just edited, which entities anchored to it. A `span`/`file`/`tree`
585 /// anchor references the path when its base path (locator suffix
586 /// `@commit` / `#span` stripped) equals the path, or — for a `tree`
587 /// grain — when the path lies under the tree. Path-shaped grains only;
588 /// `url` / `entity` anchors are matched by exact base equality.
589 pub fn anchors_referencing_artifact(
590 &self,
591 artifact_path: &str,
592 ) -> Vec<(EntityId, crate::anchor::Anchor)> {
593 let mut out = Vec::new();
594 for mount in &self.mounts {
595 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
596 continue;
597 };
598 let Ok(sc) = crate::anchor::AnchorSidecar::from_bytes(&bytes) else {
599 continue;
600 };
601 // Source-dialect anchors (decision 26) reference the same
602 // artifact under its pointer-joined workspace form — match both.
603 let source_roots = self.anchor_source_roots(&mount.mount.mem);
604 for (eid, anchors) in &sc.entities {
605 for a in anchors {
606 let joined = a
607 .source
608 .as_deref()
609 .and_then(|name| source_roots.get(name))
610 .map(|join| join_pointer(&join.pointer, anchor_base_path(&a.artifact)));
611 if anchor_references_path(a, artifact_path)
612 || joined.is_some_and(|j| {
613 path_references(
614 &j,
615 a.grain == crate::anchor::AnchorGrain::Tree,
616 artifact_path,
617 )
618 })
619 {
620 out.push((EntityId(eid.clone()), a.clone()));
621 }
622 }
623 }
624 }
625 out
626 }
627
628 /// Every `(entity_id, resolved anchor)` in `mem`, read from its anchors
629 /// sidecar once and each paired with its **live** resolution state (the
630 /// same observation [`Self::entity_anchors_resolved`] produces per entity,
631 /// computed here mem-wide in a single sidecar read). Empty for an unknown
632 /// mem, a backend that persists no anchors, or a mem with none.
633 ///
634 /// Additive read surface: the durable data is unchanged; `state` is the
635 /// [`crate::anchor::resolve_anchor`] outcome against an observation the
636 /// engine produces — the working tree for a `path`-namespace anchor, the
637 /// live graph for an `entity` one — or `None` when unobserved (never
638 /// fabricated). The verify pipeline consumes it to adjudicate a mem's
639 /// anchors against the source; audit/health can reuse it.
640 pub fn mem_anchors_resolved(&self, mem: &str) -> Vec<(EntityId, ResolvedAnchor)> {
641 let Some(mount) = self.mounts.iter().find(|m| m.mount.mem == mem) else {
642 return Vec::new();
643 };
644 let Ok(Some(bytes)) = mount.backend.read_anchors_sidecar() else {
645 return Vec::new();
646 };
647 let Ok(sc) = crate::anchor::AnchorSidecar::from_bytes(&bytes) else {
648 return Vec::new();
649 };
650 let mut out = Vec::new();
651 let source_roots = self.anchor_source_roots(mem);
652 for (eid, anchors) in &sc.entities {
653 for anchor in anchors {
654 let observed = self.observe_anchor(anchor, &source_roots);
655 let (state, observed_hash) = match observed {
656 Some((state, hash)) => (Some(state), hash),
657 None => (None, None),
658 };
659 out.push((
660 EntityId(eid.clone()),
661 ResolvedAnchor {
662 anchor: anchor.clone(),
663 state,
664 observed_hash,
665 },
666 ));
667 }
668 }
669 out
670 }
671
672 /// Standalone anchor verification — "do my sources still say what I
673 /// recorded?" for one mem, regardless of how it was built. Walks the
674 /// mem's anchor sidecar through the shared per-anchor mechanism
675 /// ([`Self::observe_anchor`] via [`Self::mem_anchors_resolved`]) and
676 /// classifies every anchor into the report vocabulary: `resolved`
677 /// (source present, hash matches or non-hash class), `drifted`
678 /// (present, hash differs, stability `stable`), `recheck` (hash
679 /// differs under `unstable`, or a hash is missing on either side),
680 /// `unresolvable` (source absent, or a grain/medium the mechanism
681 /// does not reach — never fabricated into drift). Read-only on mem
682 /// content: pure sidecar read + filesystem observation, no commit on
683 /// any backend. A mem with no anchors returns an empty report.
684 pub fn verify_mem_anchors(&self, mem: &str) -> Result<MemAnchorVerification, EngineError> {
685 if !self.mem_router.is_visible(mem) {
686 return Err(self.unknown_mem_error(mem));
687 }
688 let mut report = MemAnchorVerification {
689 mem: mem.to_string(),
690 ..Default::default()
691 };
692 for (eid, resolved) in self.mem_anchors_resolved(mem) {
693 let state = match resolved.state {
694 Some(crate::anchor::AnchorState::Resolves) => {
695 report.resolved += 1;
696 "resolved"
697 }
698 Some(crate::anchor::AnchorState::Drifted) => {
699 report.drifted += 1;
700 "drifted"
701 }
702 Some(crate::anchor::AnchorState::Recheck) => {
703 report.recheck += 1;
704 "recheck"
705 }
706 Some(crate::anchor::AnchorState::Orphaned) | None => {
707 report.unresolvable += 1;
708 "unresolvable"
709 }
710 };
711 report.anchors.push(VerifiedAnchor {
712 entity_id: eid.to_string(),
713 artifact: resolved.anchor.artifact.clone(),
714 grain: resolved.anchor.grain.as_wire().to_string(),
715 class: resolved.anchor.class.as_wire().to_string(),
716 state: state.to_string(),
717 observed_hash: resolved.observed_hash,
718 });
719 }
720 Ok(report)
721 }
722
723 /// Mem names the engine knows about, in declaration order.
724 /// Cheap; useful for callers that need to enumerate before
725 /// dispatching by mem.
726 pub fn mem_names(&self) -> Vec<&str> {
727 self.mounts.iter().map(|m| m.mount.mem.as_str()).collect()
728 }
729
730 /// Derivation-staleness report for one mem (agent-trust plan 12):
731 /// every EXPLICIT edge whose rel-type the mem's schema declares
732 /// `derivation: true`, compared against its recorded baseline.
733 /// Baseline differs from the target's current hash → `stale`;
734 /// no baseline recorded (edge predates the declaration, or was
735 /// load-derived) → `unbaselined`, distinctly — never fabricated
736 /// as fresh or stale. Fresh edges are not reported. A mem whose
737 /// schema declares no derivation rel-types returns the empty
738 /// report; an unreadable sidecar reads as empty (every edge
739 /// unbaselined) rather than an error.
740 pub fn derivation_report(
741 &self,
742 mem: &str,
743 ) -> Result<Vec<crate::ops::health::DerivationFinding>, EngineError> {
744 if !self.mem_router.is_visible(mem) {
745 return Err(self.unknown_mem_error(mem));
746 }
747 let Some(schema) = self.schemas.get(mem) else {
748 return Ok(Vec::new());
749 };
750 let declared: std::collections::HashSet<&str> = schema
751 .manifest
752 .relationships
753 .definitions
754 .iter()
755 .filter(|d| d.derivation)
756 .map(|d| d.name.as_str())
757 .collect();
758 if declared.is_empty() {
759 return Ok(Vec::new());
760 }
761 let sidecar = self
762 .mounts
763 .iter()
764 .find(|m| m.mount.mem == mem)
765 .and_then(|m| {
766 m.backend
767 .read_entity(Path::new(crate::derivation::DERIVATION_SIDECAR_PATH))
768 .ok()
769 .flatten()
770 })
771 .and_then(|bytes| crate::derivation::DerivationSidecar::from_bytes(&bytes).ok())
772 .unwrap_or_default();
773
774 let mut out = Vec::new();
775 let mut sources: Vec<&crate::entity::Entity> = self
776 .store
777 .all_entities()
778 .filter(|e| !e.stub && e.id.mem() == mem)
779 .collect();
780 sources.sort_by(|a, b| a.id.as_ref().cmp(b.id.as_ref()));
781 for entity in sources {
782 for edge in self.store.outgoing(&entity.id) {
783 if !declared.contains(edge.rel_type.as_str())
784 || edge.source != crate::store::EdgeSource::Explicit
785 {
786 continue;
787 }
788 let current = self
789 .store
790 .get(&edge.target)
791 .map(|t| t.content_hash.clone())
792 .unwrap_or_default();
793 match sidecar.get(entity.id.as_ref(), &edge.rel_type, edge.target.as_ref()) {
794 None => out.push(crate::ops::health::DerivationFinding {
795 source: entity.id.clone(),
796 rel_type: edge.rel_type.clone(),
797 target: edge.target.clone(),
798 state: "unbaselined".to_string(),
799 baseline: None,
800 current,
801 }),
802 Some(baseline) if baseline != current => {
803 out.push(crate::ops::health::DerivationFinding {
804 source: entity.id.clone(),
805 rel_type: edge.rel_type.clone(),
806 target: edge.target.clone(),
807 state: "stale".to_string(),
808 baseline: Some(baseline.to_string()),
809 current,
810 })
811 }
812 Some(_) => {}
813 }
814 }
815 }
816 Ok(out)
817 }
818
819 /// Public-shape mount record for `mem`, or `None` for an unknown
820 /// mem.
821 ///
822 /// Surfaces the operator-facing
823 /// [`crate::workspace::Mount`] (mem name, schema pin, storage
824 /// reference, capability, lifecycle, cross_linkable) so MCP / CLI
825 /// handlers can branch on backend-specific shapes via
826 /// [`crate::workspace::MountStorage`] when they need accessors
827 /// that don't make sense on every backend (e.g. gitdir / branch
828 /// for `memstead_health { include_config: true }`'s git-class
829 /// payload). Backends that want the equivalent of full's
830 /// `engine.gitdir_for(mem)` match
831 /// `engine.mount(mem).map(|m| &m.storage)` against
832 /// `MountStorage::GitBranch { gitdir, branch }` and walk
833 /// directly — keeps the engine surface backend-neutral.
834 ///
835 /// Counterpart to [`Self::mem_names`] which lists every mount.
836 pub fn mount(&self, mem: &str) -> Option<&crate::workspace::Mount> {
837 self.mounts
838 .iter()
839 .find(|m| m.mount.mem == mem)
840 .map(|m| &m.mount)
841 }
842
843 /// Orphan count attributed to each mem's pinned schema, over the
844 /// given `orphan_ids` (the caller pre-filters them by any mem scope).
845 /// Lets a health surface show that ingest-mem isolates (orphans by
846 /// design) and code-mem debt land in different schema buckets rather
847 /// than one blended, misleading total. Mems with no settled pin
848 /// bucket under the empty string.
849 pub fn orphans_by_schema(
850 &self,
851 orphan_ids: &[EntityId],
852 ) -> std::collections::BTreeMap<String, usize> {
853 let mut by_schema = std::collections::BTreeMap::new();
854 for id in orphan_ids {
855 let schema = self
856 .store()
857 .get(id)
858 .and_then(|e| self.mount(&e.mem))
859 .and_then(|m| m.schema.as_ref().map(|s| s.as_display()))
860 .unwrap_or_default();
861 *by_schema.entry(schema).or_insert(0) += 1;
862 }
863 by_schema
864 }
865
866 /// Community count attributed to each schema across `mems`: a cluster
867 /// counts toward every schema whose mems it touches, so these figures
868 /// can sum above the global community count — the same "touches"
869 /// semantic as the mem-scoped count. Per-schema dedup keeps a cluster
870 /// touching two mems of one schema from being counted twice.
871 pub fn communities_by_schema(
872 &self,
873 mems: &[String],
874 ) -> std::collections::BTreeMap<String, usize> {
875 let louvain = self.communities();
876 let mut buckets: std::collections::BTreeMap<String, std::collections::BTreeSet<String>> =
877 std::collections::BTreeMap::new();
878 for name in mems {
879 let schema = self
880 .mount(name)
881 .and_then(|m| m.schema.as_ref().map(|s| s.as_display()))
882 .unwrap_or_default();
883 let clusters = crate::graph::community::clusters_in_mem(self.store(), louvain, name);
884 buckets.entry(schema).or_default().extend(clusters);
885 }
886 buckets
887 .into_iter()
888 .map(|(schema, set)| (schema, set.len()))
889 .collect()
890 }
891
892 /// All mounts the engine knows about, in declaration order.
893 /// Counterpart to [`Self::mem_names`] when the caller needs
894 /// the full mount shape (e.g. to enumerate by storage variant).
895 pub fn mounts(&self) -> Vec<&crate::workspace::Mount> {
896 self.mounts.iter().map(|m| &m.mount).collect()
897 }
898
899 /// Names of mems whose mount declares
900 /// [`crate::workspace::MountCapability::Write`], in declaration
901 /// order. Convenience over `mounts().iter().filter(...).map(...)`
902 /// for handlers that gate by writable status (`memstead_health`,
903 /// `memstead_overview`'s mem roster, the lifecycle tools'
904 /// candidate list). Read-only mounts (archive backends) are
905 /// excluded.
906 pub fn writable_mem_names(&self) -> Vec<&str> {
907 self.mounts
908 .iter()
909 .filter(|m| m.mount.capability == MountCapability::Write)
910 .map(|m| m.mount.mem.as_str())
911 .collect()
912 }
913
914 /// The default writable mem — the target a mutation lands in when
915 /// it omits `mem`. `None` when no writable mem is mounted.
916 ///
917 /// Defined as the **first writable mount in declaration order**, i.e.
918 /// the seed / earliest-created writable mem. This is a *stable*
919 /// designation, not a function of the current name set: new mems
920 /// register via `register_writable_mem`, which pushes onto the end
921 /// of the mount list (and `mounts.json` preserves that order across
922 /// reboots), so creating an additional mem never moves the default
923 /// — even one whose name sorts ahead alphabetically. Deleting the
924 /// current default promotes the next-earliest writable mem; that is
925 /// the only thing that shifts it. Both the MCP `resolve_mem` and the
926 /// CLI's omitted-`--mem` path resolve through here so the two
927 /// surfaces always agree (the
928 /// pre-fix MCP path read `writable_mems().iter().next()` off an
929 /// unordered `HashSet`, which silently retargeted writes when a second
930 /// mem appeared).
931 pub fn default_writable_mem(&self) -> Option<&str> {
932 self.mounts
933 .iter()
934 .find(|m| m.mount.capability == MountCapability::Write)
935 .map(|m| m.mount.mem.as_str())
936 }
937
938 /// On-disk folder path for a folder-backed mount, or `None` for
939 /// any other backend (git-branch, archive) or unknown mem.
940 /// Convenience over `engine.mount(mem).map(|m| &m.storage)` +
941 /// matching on `MountStorage::Folder { path }`. Used by
942 /// handlers that need a filesystem path for a folder mem
943 /// (e.g. `memstead_health { include_config: true }`'s
944 /// `mems[].vcs.worktree` field for folder mounts).
945 pub fn folder_path_for_mem(&self, mem: &str) -> Option<&Path> {
946 match self.mount(mem).map(|m| &m.storage) {
947 Some(crate::workspace::MountStorage::Folder { path }) => Some(path.as_path()),
948 _ => None,
949 }
950 }
951
952 /// Runtime snapshot of writable / visible mems. Handlers that
953 /// need the writable roster (`memstead_health`'s `writable_mems` /
954 /// `read_mems`), per-mem origin tag (`include_config:
955 /// true`'s `mems[].origin`), or visibility check
956 /// (`memstead_overview`'s mem list, the lifecycle tools' collision
957 /// guard) consume the router here. Returned by reference — the
958 /// `Arc` is held on the engine; callers that need a clonable
959 /// handle can `Arc::clone` the engine's field directly when that
960 /// surface arrives.
961 pub fn mem_router(&self) -> &MemRouterSnapshot {
962 &self.mem_router
963 }
964
965 /// Resolve the gitdir for a writable mem. Used by `memstead_health
966 /// { include_config: true }` to surface per-mem `vcs.gitdir`
967 /// so outer-repo bookkeeping clients can `git -C <gitdir>` per
968 /// mem without hardcoding the layout.
969 ///
970 /// - `EngineError::UnknownMem` when the name does not resolve.
971 /// - `EngineError::Mem` when the mount's storage is not
972 /// git-branch-backed (folder, archive — they have no gitdir).
973 pub fn gitdir_for(&self, mem_name: &str) -> Result<PathBuf, EngineError> {
974 let m = self
975 .mount(mem_name)
976 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
977 match &m.storage {
978 MountStorage::GitBranch { gitdir, .. } => Ok(gitdir.clone()),
979 MountStorage::Folder { .. } | MountStorage::Archive { .. } | MountStorage::InMemory => {
980 Err(EngineError::Mem(format!(
981 "mem '{mem_name}' has no resolved gitdir"
982 )))
983 }
984 }
985 }
986
987 /// Resolve the worktree for a writable mem. Used by
988 /// `memstead_health { include_config: true }` to surface per-mem
989 /// `vcs.worktree`.
990 ///
991 /// - `EngineError::UnknownMem` when the name does not resolve.
992 /// - `EngineError::Mem` when the mount's backend has no
993 /// worktree concept (git-branch with no working tree, archive).
994 ///
995 /// Folder mounts surface their on-disk path. Git-branch mounts
996 /// follow the `dir: Some(...)` composition pattern: when the
997 /// workspace root contains a folder named after the mem with a
998 /// `.memstead/config.json` marker, that folder is the worktree
999 /// (disk-shape composition). Otherwise — pure mem-repo-backed
1000 /// — return Err.
1001 pub fn worktree_for(&self, mem_name: &str) -> Result<PathBuf, EngineError> {
1002 let m = self
1003 .mount(mem_name)
1004 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1005 match &m.storage {
1006 MountStorage::Folder { path } => Ok(path.clone()),
1007 MountStorage::GitBranch { .. } => {
1008 if let Some(root) = self.workspace_root.as_deref() {
1009 let candidate = root.join(mem_name);
1010 if candidate
1011 .join(crate::mem::MEM_META_DIR)
1012 .join("config.json")
1013 .is_file()
1014 {
1015 return Ok(candidate.canonicalize().unwrap_or(candidate));
1016 }
1017 }
1018 Err(EngineError::Mem(format!(
1019 "mem '{mem_name}' has no working tree (mem-repo-backed)"
1020 )))
1021 }
1022 MountStorage::Archive { .. } => Err(EngineError::Mem(format!(
1023 "mem '{mem_name}' is archive-backed and has no worktree"
1024 ))),
1025 MountStorage::InMemory => Err(EngineError::Mem(format!(
1026 "mem '{mem_name}' is in-memory and has no worktree"
1027 ))),
1028 }
1029 }
1030
1031 /// Per-mem `.memstead/config.json` payload, when available. Used
1032 /// by `memstead_health { include_config: true }` to surface the
1033 /// opaque `write_guidance` map and the catch-all `extra` fields
1034 /// per mem.
1035 ///
1036 /// Folder-backed mounts return `Some(&MemConfig)` when
1037 /// `<path>/.memstead/config.json` parsed cleanly at construction.
1038 /// Git-branch and archive backends return `None` until the
1039 /// read-from-storage-backend path lifts (the V1 unified engine
1040 /// loads configs only from folder layouts; the file lives
1041 /// inside the gitdir / archive for the other backends and
1042 /// needs a backend-level read primitive).
1043 ///
1044 /// Unknown mem names return `None` (no error variant — the
1045 /// accessor is intentionally lenient because memstead_health emits
1046 /// an empty detail block per missing config rather than
1047 /// aborting the call).
1048 pub fn mem_config_for(&self, mem: &str) -> Option<&memstead_schema::config::MemConfig> {
1049 self.mounts
1050 .iter()
1051 .find(|m| m.mount.mem == mem)
1052 .and_then(|m| m.mem_config.as_ref())
1053 }
1054
1055 /// The authoring-provenance payload an installed mem carries, read
1056 /// from the archive's `.memstead/provenance.json` at construction.
1057 /// `None` when the mem carries none (a pre-provenance archive, a
1058 /// runtime-created mem, or a backend that does not surface one) —
1059 /// the read path reports provenance as absent. Unknown mem names
1060 /// return `None`.
1061 pub fn archive_provenance_for(&self, mem: &str) -> Option<&memstead_schema::ArchiveProvenance> {
1062 self.mounts
1063 .iter()
1064 .find(|m| m.mount.mem == mem)
1065 .and_then(|m| m.archive_provenance.as_ref())
1066 }
1067
1068 /// Iterate `(mem_name, &MemConfig)` for every mount whose
1069 /// mem-config payload loaded at construction. Used by callers
1070 /// that walk every writable mount's config (`memstead health`'s
1071 /// per-mem dump, the workspace-dump CLI). The yielded `&str` is
1072 /// the authoritative mem leaf from the mount record.
1073 ///
1074 /// Folder-backed mounts yield when their `.memstead/config.json`
1075 /// parsed cleanly. Git-branch and archive backends are silent in
1076 /// V1 (the same deferred-read-from-storage gap that
1077 /// [`Self::mem_config_for`] documents).
1078 pub fn mem_configs_named(
1079 &self,
1080 ) -> impl Iterator<Item = (&str, &memstead_schema::config::MemConfig)> {
1081 self.mounts
1082 .iter()
1083 .filter_map(|m| m.mem_config.as_ref().map(|c| (m.mount.mem.as_str(), c)))
1084 }
1085
1086 /// Resolved `Arc<Schema>` for a writable mem by name. `None`
1087 /// when the name is not a registered mount.
1088 ///
1089 /// Cheap — `Arc::clone` over the per-mem schema map. Resolved
1090 /// schemas are stored in `HashMap<String, Arc<Schema>>` so the
1091 /// lookup is a single hash hit + clone.
1092 pub fn schema_for(&self, mem: &str) -> Option<std::sync::Arc<memstead_schema::Schema>> {
1093 self.schemas.get(mem).cloned()
1094 }
1095
1096 /// Cached current branch-tip cursor (typically a 40-char hex
1097 /// SHA for git-branch backends; `None` for fresh mems or
1098 /// backends that don't track a head — folder / archive).
1099 ///
1100 /// The value is the per-mount `last_known_head`, seeded at
1101 /// construction by `backend.current_head()` and refreshed by
1102 /// [`Self::reload_if_stale`] / mutation paths after a
1103 /// successful commit.
1104 ///
1105 /// - `EngineError::UnknownMem` when the name does not resolve.
1106 pub fn mem_head_sha(&self, mem_name: &str) -> Result<Option<String>, EngineError> {
1107 let m = self
1108 .mounts
1109 .iter()
1110 .find(|m| m.mount.mem == mem_name)
1111 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1112 Ok(m.last_known_head.clone())
1113 }
1114
1115 /// Whether a sibling writer has advanced this mem's backend past
1116 /// the engine's cached `last_known_head` — a read-only drift probe
1117 /// that does **not** reload (unlike [`Self::reload_if_stale`]). One
1118 /// `backend.current_head()` read compared against the cached cursor;
1119 /// the comparison clears once the engine re-reads (a `reload` /
1120 /// `reload_if_stale` refreshes `last_known_head` to the live tip).
1121 ///
1122 /// Only git-branch backends track a head, so folder / archive /
1123 /// in-memory mounts always report `false`. A backend that errors on
1124 /// the probe (transient refdb hiccup) reports `false` rather than
1125 /// surfacing the error — drift is advisory, and the next real
1126 /// operation's reload path is the authoritative sync.
1127 ///
1128 /// - `EngineError::UnknownMem` when the name does not resolve.
1129 pub fn mem_drifted(&self, mem_name: &str) -> Result<bool, EngineError> {
1130 let m = self
1131 .mounts
1132 .iter()
1133 .find(|m| m.mount.mem == mem_name)
1134 .ok_or_else(|| self.unknown_mem_error(mem_name))?;
1135 let live = m.backend.current_head().ok().flatten();
1136 Ok(live != m.last_known_head)
1137 }
1138
1139 /// Workspace root the engine booted from, when one is known.
1140 /// `None` for engines built directly from a mount list (tests,
1141 /// ad-hoc consumers). Set by [`Self::from_workspace_root`] and
1142 /// the full counterpart.
1143 pub fn workspace_root(&self) -> Option<&Path> {
1144 self.workspace_root.as_deref()
1145 }
1146
1147 /// Typed warnings surfaced during mem load — drift findings
1148 /// the loader pipeline collects per entity. Empty for V1; the
1149 /// accessor surfaces them so handlers can merge into health
1150 /// summaries uniformly.
1151 pub fn load_warnings(&self) -> &[WarningHint] {
1152 &self.load_warnings
1153 }
1154
1155 /// The quarantine roster: mems that failed their mem-level boot
1156 /// step and serve nothing until repaired + reloaded. Empty on a
1157 /// fully healthy workspace. Surfaced on overview and health.
1158 pub fn quarantined_mems(&self) -> &[crate::engine::QuarantinedMem] {
1159 &self.quarantined
1160 }
1161
1162 /// The quarantine entry for `mem`, when it is quarantined.
1163 pub fn quarantine_reason(&self, mem: &str) -> Option<&crate::engine::QuarantinedMem> {
1164 self.quarantined.iter().find(|q| q.mount.mem == mem)
1165 }
1166
1167 /// Mems whose lazy entity load is still DEFERRED — on the mount
1168 /// roster with a resolved schema pin, but with no entities in the
1169 /// store yet. Read surfaces that render per-mem counts or
1170 /// distributions consult this: a count over a deferred mem's slice
1171 /// of the store is a count over nothing, and rendering it as a
1172 /// bare zero is the silent absence the lazy-mount contract forbids.
1173 /// Either trigger the load ([`Self::ensure_mems_loaded`]) or render
1174 /// the load state explicitly.
1175 pub fn deferred_mems(&self) -> Vec<&str> {
1176 self.mounts
1177 .iter()
1178 .filter(|m| m.deferred)
1179 .map(|m| m.mount.mem.as_str())
1180 .collect()
1181 }
1182
1183 /// Whether `mem` is a lazy mount whose entity load has not run yet.
1184 pub fn mem_is_deferred(&self, mem: &str) -> bool {
1185 self.mounts.iter().any(|m| m.deferred && m.mount.mem == mem)
1186 }
1187
1188 /// The typed error for a mem name that did not resolve to a
1189 /// serving mount: `MEM_QUARANTINED` (carrying the underlying boot
1190 /// failure and its repair command) when the mem is on the
1191 /// quarantine roster, `UNKNOWN_MEM` otherwise. Every lookup site
1192 /// that fails to find a mem routes here so a quarantined mem is
1193 /// never misreported as unknown — honest absence, with the reason.
1194 pub fn unknown_mem_error(&self, mem: &str) -> EngineError {
1195 match self.quarantine_reason(mem) {
1196 Some(q) => EngineError::MemQuarantined {
1197 mem: mem.to_string(),
1198 reason_code: q.reason_code.clone(),
1199 reason_message: q.reason_message.clone(),
1200 },
1201 None => EngineError::UnknownMem(mem.to_string()),
1202 }
1203 }
1204
1205 /// The workspace-level boot diagnosis a diagnostic-shell engine
1206 /// carries (`None` on ordinarily booted engines).
1207 pub fn boot_diagnosis(&self) -> Option<(&str, &str)> {
1208 self.boot_diagnosis
1209 .as_ref()
1210 .map(|(c, m)| (c.as_str(), m.as_str()))
1211 }
1212
1213 /// Build a mem-less diagnostic-shell engine for a workspace whose
1214 /// boot failed at the WORKSPACE level (nothing loadable — e.g. an
1215 /// unparseable store). It serves no mems and no entities; its one
1216 /// job is answering overview/health with the typed boot diagnosis
1217 /// so a session can always ask WHY the graph is gone — the MCP
1218 /// server serves this instead of exiting into `-32000 Connection
1219 /// closed` (degrade, never disappear).
1220 pub fn diagnostic_shell(reason_code: String, reason_message: String) -> Engine {
1221 let mut engine =
1222 Engine::from_mounts(Vec::new()).expect("an empty mount list always constructs");
1223 engine.boot_diagnosis = Some((reason_code, reason_message));
1224 engine
1225 }
1226
1227 /// Append boot-path quarantine entries recorded outside
1228 /// `from_mounts_inner` (backend-instantiation failures happen
1229 /// before the mount list reaches the engine constructor). Boot
1230 /// paths only — quarantine is a boot judgment, never a runtime
1231 /// mutation.
1232 pub fn extend_quarantine(&mut self, entries: Vec<crate::engine::QuarantinedMem>) {
1233 self.quarantined.extend(entries);
1234 }
1235
1236 // ---------------------------------------------------------------
1237 // Read-side delegates onto the kernel ops/graph functions.
1238 //
1239 // The mem-router engine exposed each of these directly so the
1240 // MCP layer could call them without reaching into the store. The
1241 // unified engine mirrors that surface so the MCP migration is a
1242 // straight rename rather than a re-architecture.
1243 //
1244 // Multi-mem cache strategy: per-mem community detection and
1245 // per-mem search indexes are unnecessary at this layer — the
1246 // engine-wide store already carries every mount's edges; Louvain
1247 // and tantivy run once across the union. `mem_schemas` for
1248 // health/search is the engine's existing `schemas` field as-is.
1249 // ---------------------------------------------------------------
1250
1251 /// Lazy community-detection cache. First call runs Louvain
1252 /// against the current store using one pinned schema for
1253 /// `community.{resolution, seed}` and the per-rel weights.
1254 /// Subsequent calls return the cached result. Mutations invalidate
1255 /// the cache via [`Self::invalidate_communities`].
1256 ///
1257 /// One detection run per engine. The partition is workspace-global,
1258 /// so it needs a single source for the Louvain parameters; that
1259 /// source is the schema of the lexicographically-first mem name —
1260 /// a stable key, so the partition is deterministic across processes
1261 /// even when mounts pin heterogeneous schemas. For a single-schema
1262 /// workspace every mem's schema is identical, so the choice of
1263 /// key is immaterial there.
1264 pub fn communities(&self) -> &LouvainOutput {
1265 // Generation sanity: a stored memo must match the live store
1266 // generation — the mutation hooks clear stale memos before any
1267 // read can arrive here. A mismatch would mean a mutation path
1268 // skipped its invalidation call (the exact silent-staleness
1269 // bug the generation exists to catch).
1270 if let Some((memo_key, _)) = self.community_memo.get() {
1271 debug_assert_eq!(
1272 *memo_key,
1273 self.derived_key(),
1274 "community memo key lags the engine — a mutation path missed invalidate_communities"
1275 );
1276 }
1277 &self
1278 .community_memo
1279 .get_or_init(|| (self.derived_key(), self.compute_communities()))
1280 .1
1281 }
1282
1283 /// The current validity key for derived-structure memos: store
1284 /// generation plus schemas epoch (see [`super::DerivedKey`]).
1285 pub fn derived_key(&self) -> super::DerivedKey {
1286 super::DerivedKey {
1287 store_generation: self.store.generation(),
1288 schemas_epoch: self.schemas_epoch,
1289 }
1290 }
1291
1292 fn compute_communities(&self) -> LouvainOutput {
1293 {
1294 // Select the parameter schema by a stable key (smallest
1295 // mem name) rather than unordered-map iteration, so the
1296 // partition does not vary between processes. Fall back to
1297 // the builtin default for the empty-mounts case (caller
1298 // still gets a valid empty Louvain result against an empty
1299 // store).
1300 let schema = self
1301 .schemas
1302 .iter()
1303 .min_by(|a, b| a.0.cmp(b.0))
1304 .map(|(_, s)| s.clone())
1305 .unwrap_or_else(Schema::builtin_default);
1306 let manifest = &schema.manifest;
1307 let resolution = manifest.community.resolution;
1308 let seed = manifest.community.seed;
1309 let schema_for_weights = schema.clone();
1310 detect_communities(&self.store, resolution, seed, move |rel_type| {
1311 schema_for_weights
1312 .manifest
1313 .relationships
1314 .definitions
1315 .iter()
1316 .find(|d| d.name == rel_type)
1317 .map(|d| d.default_weight as f64)
1318 .unwrap_or(1.0)
1319 })
1320 }
1321 }
1322
1323 /// Drop the cached community detection result and the grounded
1324 /// labelling memo — unless the store still sits at the generation
1325 /// each memo was computed from (flywheel W8/01). The keep case is
1326 /// exactly the batch-rollback path: the restored snapshot restored
1327 /// the generation with it, so the memo describes the live state
1328 /// and recomputing it would be pure waste. Every real mutation
1329 /// bumps the generation first, so those clears behave as before.
1330 /// Coupling the labelling reset here means every site that already
1331 /// invalidates communities (all mutation paths, drift reload,
1332 /// quarantine attach/detach, apply-commit) invalidates the
1333 /// labelling too, so a stale label can never outlive the state
1334 /// change that moved it.
1335 pub fn invalidate_communities(&mut self) {
1336 let key = self.derived_key();
1337 if !matches!(self.community_memo.get(), Some((k, _)) if *k == key) {
1338 self.community_memo = OnceCell::new();
1339 }
1340 if !matches!(self.labelling_memo.get(), Some((k, _)) if *k == key) {
1341 self.labelling_memo = OnceCell::new();
1342 }
1343 }
1344
1345 /// The grounded labelling of one mem — `None` when its pinned
1346 /// schema declares no `relationships.labelling`. Computed on
1347 /// first access for every declaring mem, memoised until the next
1348 /// invalidation; generation-keyed like `community_memo`.
1349 pub fn mem_labelling(&self, mem: &str) -> Option<&crate::ops::labelling::MemLabelling> {
1350 // Generation sanity, same contract as `communities()`: a
1351 // stored memo must match the live derived key — a mismatch
1352 // means a mutation path missed invalidate_communities.
1353 if let Some((memo_key, _)) = self.labelling_memo.get() {
1354 debug_assert_eq!(
1355 *memo_key,
1356 self.derived_key(),
1357 "labelling memo key lags the engine — a mutation path missed invalidate_communities"
1358 );
1359 }
1360 let (_, map) = self.labelling_memo.get_or_init(|| {
1361 let mut out = std::collections::HashMap::new();
1362 for (mem_name, schema) in &self.schemas {
1363 if let Some(lab) = crate::ops::labelling::labelling_of(schema) {
1364 out.insert(
1365 mem_name.clone(),
1366 crate::ops::labelling::compute_mem_labelling(
1367 &self.store,
1368 mem_name,
1369 &lab.attack,
1370 ),
1371 );
1372 }
1373 }
1374 (self.derived_key(), out)
1375 });
1376 map.get(mem)
1377 }
1378
1379 /// One entity's served labelling view — `None` when the entity is
1380 /// a stub or its mem's schema declares no labelling; serving
1381 /// surfaces then keep their byte-identical payloads. The shape
1382 /// block is present exactly when the declaration carries a
1383 /// `support` walk.
1384 pub fn computed_labelling(
1385 &self,
1386 entity: &Entity,
1387 ) -> Option<crate::ops::labelling::LabellingView> {
1388 use crate::ops::labelling::{Label, compute_shape, labelling_of};
1389 if entity.stub {
1390 return None;
1391 }
1392 let schema = self.schemas.get(entity.mem.as_str())?;
1393 let lab = labelling_of(schema)?;
1394 let mem_lab = self.mem_labelling(entity.mem.as_str())?;
1395 let label = *mem_lab.labels.get(entity.id.0.as_str())?;
1396 let defeated_by = if label == Label::Defeated {
1397 mem_lab.accepted_attackers_of(entity.id.0.as_str())
1398 } else {
1399 Vec::new()
1400 };
1401 let undecided_by = if label == Label::Undecided {
1402 mem_lab.undecided_attackers_of(entity.id.0.as_str())
1403 } else {
1404 Vec::new()
1405 };
1406 let shape = lab.support.as_ref().map(|walk| {
1407 let label_of = |id: &EntityId| -> Option<Label> {
1408 self.mem_labelling(id.mem())
1409 .and_then(|ml| ml.labels.get(id.0.as_str()).copied())
1410 };
1411 compute_shape(&self.store, &entity.id, walk, &label_of)
1412 });
1413 Some(crate::ops::labelling::LabellingView {
1414 label,
1415 defeated_by,
1416 undecided_by,
1417 shape,
1418 })
1419 }
1420
1421 /// The `labelling` health axis payload — per declaring mem:
1422 /// counts per label, the defeated list with its accepted
1423 /// attackers, the undecided list with its open attacker set, and
1424 /// the excluded cross-mem attack-edge count. One composer shared
1425 /// by the CLI health command and both MCP flavours.
1426 pub fn health_labelling_axis(&self, mem_filter: Option<&str>) -> serde_json::Value {
1427 use crate::ops::labelling::Label;
1428 let mut mems = serde_json::Map::new();
1429 let mut mem_names: Vec<&String> = self.schemas.keys().collect();
1430 mem_names.sort();
1431 for mem in mem_names {
1432 if let Some(v) = mem_filter
1433 && mem != v
1434 {
1435 continue;
1436 }
1437 let Some(ml) = self.mem_labelling(mem) else {
1438 continue;
1439 };
1440 let mut accepted = 0usize;
1441 let mut defeated: Vec<serde_json::Value> = Vec::new();
1442 let mut undecided: Vec<serde_json::Value> = Vec::new();
1443 for (id, label) in &ml.labels {
1444 match label {
1445 Label::Accepted => accepted += 1,
1446 Label::Defeated => defeated.push(serde_json::json!({
1447 "id": id,
1448 "defeated_by": ml.accepted_attackers_of(id),
1449 })),
1450 Label::Undecided => undecided.push(serde_json::json!({
1451 "id": id,
1452 "undecided_by": ml.undecided_attackers_of(id),
1453 })),
1454 }
1455 }
1456 mems.insert(
1457 mem.clone(),
1458 serde_json::json!({
1459 "counts": {
1460 "accepted": accepted,
1461 "defeated": defeated.len(),
1462 "undecided": undecided.len(),
1463 },
1464 "defeated": defeated,
1465 "undecided": undecided,
1466 "cross_mem_edges_excluded": ml.cross_mem_edges_excluded,
1467 }),
1468 );
1469 }
1470 serde_json::Value::Object(mems)
1471 }
1472
1473 /// Real entities with no incoming or outgoing edges — leaf-declared
1474 /// types exempt (their edge-less entities are terminal by
1475 /// construction; see [`Self::leaf_population`]).
1476 pub fn orphans(&self) -> Vec<EntityId> {
1477 crate::graph::query::find_orphans_with_schemas(&self.store, &self.schemas)
1478 }
1479
1480 /// Count of real entities per leaf-declared type, keyed
1481 /// `<schema_ref>:<type>` — the visible population the orphan
1482 /// exemption covers.
1483 pub fn leaf_population(&self) -> std::collections::BTreeMap<String, usize> {
1484 crate::graph::query::leaf_population(&self.store, &self.schemas)
1485 }
1486
1487 /// Stub entities with their referencer ids.
1488 pub fn stubs(&self) -> Vec<(EntityId, Vec<EntityId>)> {
1489 crate::graph::query::find_stubs(&self.store)
1490 }
1491
1492 /// Top `limit` entities by total degree.
1493 pub fn most_connected(&self, limit: usize) -> Vec<crate::graph::query::Connectivity> {
1494 crate::graph::query::most_connected(&self.store, limit)
1495 }
1496
1497 /// Entities whose type's `required_outgoing` blocks are not yet
1498 /// satisfied. `mem_filter = None` scans every mem; `Some(v)`
1499 /// scans only that mem.
1500 pub fn missing_required_outgoing(
1501 &self,
1502 mem_filter: Option<&str>,
1503 ) -> Vec<crate::ops::health::MissingRequiredOutgoingReport> {
1504 crate::ops::health::collect_missing_required_outgoing(
1505 &self.store,
1506 mem_filter,
1507 &self.schemas,
1508 )
1509 }
1510
1511 /// Standing violations of declared `constraints` (the health
1512 /// `constraints` include) — every non-stub entity whose type
1513 /// declares constraints its current state violates, in
1514 /// deterministic `(mem, id)` order.
1515 pub fn constraint_findings(
1516 &self,
1517 mem_filter: Option<&str>,
1518 ) -> Vec<crate::ops::health::ConstraintFindingReport> {
1519 crate::ops::health::collect_constraint_findings(&self.store, mem_filter, &self.schemas)
1520 }
1521
1522 /// The evaluated aggregate signals for one entity — `None` when
1523 /// the mem has no schema, the type is unknown or a stub, or the
1524 /// type declares no signals; serving surfaces then keep their
1525 /// byte-identical payloads.
1526 pub fn computed_signals(
1527 &self,
1528 entity: &Entity,
1529 ) -> Option<Vec<crate::ops::signals::ComputedSignal>> {
1530 if entity.stub {
1531 return None;
1532 }
1533 let schema = self.schemas.get(entity.mem.as_str())?;
1534 let td = schema.types.get(entity.entity_type.as_str())?;
1535 if td.signals.is_empty() {
1536 return None;
1537 }
1538 Some(crate::ops::signals::compute_signals(
1539 &self.store,
1540 td,
1541 &entity.id,
1542 ))
1543 }
1544
1545 /// Every entity carrying at least one signal above `none` — the
1546 /// include-gated `signals` health axis.
1547 pub fn signal_reports(
1548 &self,
1549 mem_filter: Option<&str>,
1550 ) -> Vec<crate::ops::health::SignalReport> {
1551 crate::ops::health::collect_signal_reports(&self.store, mem_filter, &self.schemas)
1552 }
1553
1554 /// The `signals` health axis payload — the entity roster plus
1555 /// per-level counts. One composer shared by the CLI health
1556 /// command and both MCP flavours so the axis cannot drift
1557 /// between surfaces.
1558 pub fn health_signals_axis(&self, mem_filter: Option<&str>) -> serde_json::Value {
1559 use memstead_schema::SignalLevel;
1560 let reports = self.signal_reports(mem_filter);
1561 let mut notice = 0usize;
1562 let mut warn = 0usize;
1563 for r in &reports {
1564 for s in &r.signals {
1565 match s.level {
1566 Some(SignalLevel::Notice) => notice += 1,
1567 Some(SignalLevel::Warn) => warn += 1,
1568 None => {}
1569 }
1570 }
1571 }
1572 serde_json::json!({
1573 "entities": reports,
1574 "counts": { "notice": notice, "warn": warn },
1575 })
1576 }
1577
1578 /// Defective section-format declarations the loaded schemas carry
1579 /// (lenient boot recorded them; install would have refused).
1580 pub fn schema_format_defects(&self) -> Vec<crate::ops::health::SchemaFormatDefect> {
1581 crate::ops::health::collect_schema_format_defects(&self.schemas)
1582 }
1583
1584 /// Conformance-axis integrity findings for one mem — which
1585 /// entities a write would refuse under the effective schema, and
1586 /// why. `target_schema = None` lints against the mem's current
1587 /// pin; `Some(ref)` lints against that schema instead (resolved
1588 /// among mem-pinned, workspace, and built-in schemas).
1589 pub fn conformance_findings(
1590 &self,
1591 mem: &str,
1592 target_schema: Option<&memstead_schema::SchemaRef>,
1593 ) -> Result<Vec<crate::ops::integrity::IntegrityFinding>, EngineError> {
1594 let pinned = self
1595 .schemas
1596 .get(mem)
1597 .ok_or_else(|| self.unknown_mem_error(mem))?;
1598 let effective: Arc<Schema> = match target_schema {
1599 None => pinned.clone(),
1600 Some(target) => self.resolve_schema_by_ref(target).ok_or_else(|| {
1601 let consulted: Vec<_> = self
1602 .workspace_schemas
1603 .iter()
1604 .chain(self.builtin_schemas.iter())
1605 .cloned()
1606 .collect();
1607 EngineError::SchemaNotFound {
1608 mem: mem.to_string(),
1609 pin: target.as_display(),
1610 sources: crate::engine::error::SchemaSourceDiagnostic::for_failed_pin(
1611 &target.name,
1612 &target.version,
1613 &consulted,
1614 ),
1615 install_hint: None,
1616 }
1617 .with_schema_install_probe(self.workspace_root())
1618 })?,
1619 };
1620 Ok(crate::ops::integrity::conformance_findings(
1621 &self.store,
1622 mem,
1623 &effective,
1624 &self.schemas,
1625 ))
1626 }
1627
1628 /// Resolve an exact `name@version` ref against every schema this
1629 /// engine can see: mem-pinned, workspace-authored, built-in.
1630 /// `None` when no loaded schema matches.
1631 pub(crate) fn resolve_schema_by_ref(
1632 &self,
1633 target: &memstead_schema::SchemaRef,
1634 ) -> Option<Arc<Schema>> {
1635 self.schemas
1636 .values()
1637 .chain(self.workspace_schemas.iter())
1638 .chain(self.builtin_schemas.iter())
1639 .find(|s| {
1640 let (name, version) = s.id();
1641 name == target.name && version == target.version
1642 })
1643 .cloned()
1644 }
1645
1646 /// The mem's `Mount.schema` expectation assertion, when set.
1647 /// `None` for unknown mems *and* for mems whose mount carries no
1648 /// assertion (the authoritative pin then lives in the backend
1649 /// config; the resolved active schema, not this, is the effective pin).
1650 pub fn schema_pin(&self, mem: &str) -> Option<memstead_schema::SchemaRef> {
1651 self.mounts
1652 .iter()
1653 .find(|m| m.mount.mem == mem)
1654 .and_then(|m| m.mount.schema.clone())
1655 }
1656
1657 /// The mem's in-flight migration target, when dual-pin state is
1658 /// active. `None` for settled or unknown mems.
1659 pub fn migration_target(&self, mem: &str) -> Option<memstead_schema::SchemaRef> {
1660 self.mounts
1661 .iter()
1662 .find(|m| m.mount.mem == mem)
1663 .and_then(|m| m.mount.migration_target.clone())
1664 }
1665
1666 /// Consistency-axis integrity findings for one mem — the
1667 /// pre-existing graph-coherence categories (dangling links, stubs)
1668 /// projected into the `{ id, axis, code, detail }` finding shape.
1669 pub fn consistency_findings(
1670 &self,
1671 mem: &str,
1672 ) -> Result<Vec<crate::ops::integrity::IntegrityFinding>, EngineError> {
1673 if !self.schemas.contains_key(mem) {
1674 return Err(self.unknown_mem_error(mem));
1675 }
1676 Ok(crate::ops::integrity::consistency_findings(
1677 &self.store,
1678 mem,
1679 ))
1680 }
1681
1682 /// Engine-wide health summary across every mount.
1683 pub fn health(&self) -> crate::ops::HealthSummary {
1684 self.health_inner(None)
1685 }
1686
1687 /// Health summary scoped to one visible mem. The scans and
1688 /// structural counts narrow to that mem; workspace-level facts
1689 /// (quarantine roster, boot diagnosis, workspace-scoped warnings)
1690 /// stay global — an agent scoping to one mem must still see them.
1691 /// A name that is quarantined or not on the visible roster refuses
1692 /// `UNKNOWN_MEM` — the same gate `search` applies to its `mem`
1693 /// filter, so "no such mem" and "healthy mem, nothing to report"
1694 /// can never be confused. `None` is the engine-wide sweep.
1695 pub fn health_scoped(
1696 &self,
1697 mem: Option<&str>,
1698 ) -> Result<crate::ops::HealthSummary, crate::EngineError> {
1699 if let Some(name) = mem
1700 && (self.quarantine_reason(name).is_some() || !self.mem_router.is_visible(name))
1701 {
1702 return Err(self.unknown_mem_error(name));
1703 }
1704 Ok(self.health_inner(mem))
1705 }
1706
1707 fn health_inner(&self, mem: Option<&str>) -> crate::ops::HealthSummary {
1708 let fallback = engine_fallback_type();
1709 let mut summary =
1710 crate::ops::health::compute_health(&self.store, fallback.as_ref(), &self.schemas, mem);
1711 // Merge in load-time drift warnings so every caller of
1712 // Engine::health — MCP handler, Swift FFI, direct CLI —
1713 // sees the SuspiciousNestedPrefix / DuplicateSectionHeading
1714 // findings without reaching into private engine state. The
1715 // MCP handler further appends request-scoped warnings on
1716 // top. Mirrors full's merge.
1717 if !self.load_warnings.is_empty() {
1718 let mut merged = self.load_warnings.clone();
1719 merged.append(&mut summary.warnings);
1720 summary.warnings = merged;
1721 }
1722 // Quarantine roster — a boot-honesty fact, present whenever
1723 // non-empty, never behind an include gate. Empty (and omitted
1724 // from the wire) on a healthy workspace.
1725 summary.quarantined = self
1726 .quarantined
1727 .iter()
1728 .map(|q| crate::ops::QuarantinedMemReport {
1729 mem: q.mount.mem.clone(),
1730 reason_code: q.reason_code.clone(),
1731 reason_message: q.reason_message.clone(),
1732 })
1733 .collect();
1734 // Per-file load failures ride the report unconditionally, like
1735 // the quarantine roster — each entry's message names the remedy
1736 // (the merge-conflict refusal names `memstead conflicts
1737 // resolve`), and a remedy only a library accessor carries is a
1738 // capability nobody finds at the moment it is needed.
1739 summary.load_errors = self
1740 .load_errors
1741 .iter()
1742 .map(|(path, msg)| crate::ops::LoadErrorReport {
1743 file: path.display().to_string(),
1744 error: msg.clone(),
1745 })
1746 .collect();
1747 summary.boot_diagnosis = self
1748 .boot_diagnosis
1749 .as_ref()
1750 .map(|(code, message)| serde_json::json!({ "code": code, "message": message }));
1751 // Surface OUTER_REPO_NOT_IGNORING_MEM_REPO when the
1752 // workspace is embedded inside a git repository whose
1753 // .gitignore does not list `mem-repo/`. Skipped when
1754 // workspace_root is unset (engine built ad-hoc from a mount
1755 // list).
1756 if let Some(root) = self.workspace_root.as_deref()
1757 && let Some(outer) = crate::workspace_root::find_enclosing_git_repo(root)
1758 && !crate::workspace_root::outer_repo_ignores_mem_repo(&outer, root)
1759 {
1760 summary
1761 .warnings
1762 .push(WarningHint::OuterRepoNotIgnoringMemRepo {
1763 outer_repo_root: outer.display().to_string(),
1764 workspace_root: root.display().to_string(),
1765 });
1766 }
1767 // Authoring-drift axis: for every pinned schema whose sealed
1768 // copy carries an install-provenance stamp, report a MISSING
1769 // authoring package (stamped path gone) or a DIVERGED one
1770 // (present but no longer parsed-equivalent to the seal).
1771 // Unstamped schemas — sealed pre-stamp, built-ins, archive
1772 // installs — produce no finding. Read-only on both copies.
1773 summary.warnings.extend(self.authoring_drift_findings());
1774 // Rot axis for the pins the drift axis skips: an UNSTAMPED
1775 // sealed package whose content no longer passes current-
1776 // language authoring validation gets its own low-tier hint —
1777 // the holding runs fine on the tolerant seal, but the package
1778 // (and the unlocatable authoring source it came from) is no
1779 // longer installable, and nothing else would say so before the
1780 // next install attempt. A parsing unstamped package stays
1781 // silent; stamped pins are the drift axis's business.
1782 summary.warnings.extend(self.unstamped_rot_findings());
1783 // Under a mem scope, mem-attributable warnings narrow to the
1784 // scoped mem; workspace- and request-scoped warnings return
1785 // `None` from `source_mem()` and stay visible regardless.
1786 // Mirrors the full flavour's compose filter.
1787 if let Some(v) = mem {
1788 summary
1789 .warnings
1790 .retain(|w| w.source_mem().is_none_or(|wv| wv == v));
1791 }
1792 summary
1793 }
1794
1795 /// Compute the authoring-drift findings for every stamped pinned
1796 /// schema. See the call site in [`Self::health`] for the axis
1797 /// contract; returns an empty list when no workspace root is set
1798 /// (ad-hoc mount-list engines have no authoring tree to check).
1799 fn authoring_drift_findings(&self) -> Vec<WarningHint> {
1800 let Some(root) = self.workspace_root.as_deref() else {
1801 return Vec::new();
1802 };
1803 // Group pinning mems by (name, version) — BTreeMap for a
1804 // deterministic finding order.
1805 let mut pins: std::collections::BTreeMap<(String, String), Vec<String>> =
1806 std::collections::BTreeMap::new();
1807 for (mem, schema) in &self.schemas {
1808 let (name, version) = schema.id();
1809 pins.entry((name.to_string(), version.to_string()))
1810 .or_default()
1811 .push(mem.clone());
1812 }
1813 let mut out = Vec::new();
1814 for ((name, version), mut mems) in pins {
1815 mems.sort();
1816 let Some(stamped_path) = self.read_install_provenance(root, &name, &version) else {
1817 continue;
1818 };
1819 let schema_ref = format!("{name}@{version}");
1820 let authoring = std::path::Path::new(&stamped_path);
1821 if !authoring.is_dir() {
1822 out.push(WarningHint::SchemaAuthoringSourceMissing {
1823 schema_ref,
1824 stamped_path,
1825 mems,
1826 });
1827 continue;
1828 }
1829 let sealed = self
1830 .schemas
1831 .get(&mems[0])
1832 .expect("mems collected from self.schemas keys")
1833 .clone();
1834 match memstead_schema::load_schema_from_dir(authoring) {
1835 Err(e) => out.push(WarningHint::SchemaAuthoringSourceDiverged {
1836 schema_ref,
1837 stamped_path,
1838 mems,
1839 detail: format!("the authoring package no longer loads: {e}"),
1840 }),
1841 Ok(authored) => {
1842 if schema_parsed_fingerprint(&authored) != schema_parsed_fingerprint(&sealed) {
1843 out.push(WarningHint::SchemaAuthoringSourceDiverged {
1844 schema_ref,
1845 stamped_path,
1846 mems,
1847 detail: "the parsed authoring package differs from the sealed copy \
1848 the engine runs on"
1849 .to_string(),
1850 });
1851 }
1852 }
1853 }
1854 }
1855 out
1856 }
1857
1858 /// Compute the rot findings for every UNSTAMPED pinned schema: read
1859 /// the sealed package's content back (folder seal directory, or the
1860 /// `__MEMSTEAD:schemas/` ref via the ops bundle) and run the
1861 /// authoring-tier check over it. A pin with a stamp is skipped (the
1862 /// divergence axis owns it); a pin with no readable sealed package
1863 /// — built-ins resolving from the embedded catalogue — is skipped
1864 /// too (nothing on disk can rot). See the call site in
1865 /// [`Self::health`] for the axis contract.
1866 fn unstamped_rot_findings(&self) -> Vec<WarningHint> {
1867 let Some(root) = self.workspace_root.as_deref() else {
1868 return Vec::new();
1869 };
1870 let mut pins: std::collections::BTreeMap<(String, String), Vec<String>> =
1871 std::collections::BTreeMap::new();
1872 for (mem, schema) in &self.schemas {
1873 let (name, version) = schema.id();
1874 pins.entry((name.to_string(), version.to_string()))
1875 .or_default()
1876 .push(mem.clone());
1877 }
1878 let mut out = Vec::new();
1879 for ((name, version), mut mems) in pins {
1880 mems.sort();
1881 if self
1882 .read_install_provenance(root, &name, &version)
1883 .is_some()
1884 {
1885 continue; // stamped — the divergence axis checks it
1886 }
1887 let schema_ref = format!("{name}@{version}");
1888 // Folder seal: the sealed package is a real directory the
1889 // authoring loader can probe directly.
1890 let sealed_dir = root.join(".memstead").join("schemas").join(&schema_ref);
1891 let detail: Option<String> = if sealed_dir.join("schema.yaml").is_file() {
1892 memstead_schema::load_schema_from_dir(&sealed_dir)
1893 .err()
1894 .map(|e| e.to_string())
1895 } else if let Some((manifest, types)) = self.read_sealed_package_yamls(&name, &version)
1896 {
1897 memstead_schema::loader::check_package_reauthorable(&manifest, &types)
1898 .err()
1899 .map(|e| e.to_string())
1900 } else {
1901 None // no sealed copy anywhere — embedded builtin
1902 };
1903 if let Some(detail) = detail {
1904 out.push(WarningHint::SchemaUnstampedSourceRot {
1905 schema_ref,
1906 mems,
1907 detail,
1908 });
1909 }
1910 }
1911 out
1912 }
1913
1914 /// Read a sealed package's `schema.yaml` + `types/*.yaml` back from
1915 /// the `__MEMSTEAD:schemas/` ref, reconstructing the type-file names
1916 /// from the pinned parsed schema's type roster (seal-time authoring
1917 /// enforces stem == declared type name). `None` when the ops bundle
1918 /// or the package is absent — the embedded-builtin state.
1919 fn read_sealed_package_yamls(
1920 &self,
1921 name: &str,
1922 version: &str,
1923 ) -> Option<(String, Vec<(String, String)>)> {
1924 let ops = self.git_branch_ops()?;
1925 let root = self.workspace_root.as_deref()?;
1926 let gitdir = self
1927 .mounts
1928 .iter()
1929 .find_map(|m| match &m.mount.storage {
1930 crate::workspace::MountStorage::GitBranch { gitdir, .. } => Some(gitdir.clone()),
1931 _ => None,
1932 })
1933 .or_else(|| {
1934 let g = root.join("mem-repo").join(".git");
1935 g.is_dir().then_some(g)
1936 })?;
1937 let read = |rel: &str| -> Option<String> {
1938 (ops.read_schema_file)(&gitdir, name, version, rel)
1939 .ok()
1940 .flatten()
1941 .and_then(|bytes| String::from_utf8(bytes).ok())
1942 };
1943 let manifest = read("schema.yaml")?;
1944 let schema = self
1945 .schemas
1946 .values()
1947 .find(|s| {
1948 let (n, v) = s.id();
1949 n == name && v.to_string() == version
1950 })?
1951 .clone();
1952 let mut type_names: Vec<String> = schema.types.keys().cloned().collect();
1953 type_names.sort();
1954 let mut types = Vec::new();
1955 for t in type_names {
1956 if let Some(body) = read(&format!("types/{t}.yaml")) {
1957 types.push((t, body));
1958 }
1959 }
1960 Some((manifest, types))
1961 }
1962
1963 /// Read the install-provenance stamp for a sealed schema package,
1964 /// checking the folder location first
1965 /// (`.memstead/schemas/<name>@<version>/`) and falling back to the
1966 /// `__MEMSTEAD:schemas/` ref via the git-branch ops bundle when
1967 /// wired. `None` when no stamp exists anywhere — the normal state
1968 /// for pre-stamp seals, built-ins, and archive installs.
1969 fn read_install_provenance(&self, root: &Path, name: &str, version: &str) -> Option<String> {
1970 let folder_stamp = root
1971 .join(".memstead")
1972 .join("schemas")
1973 .join(format!("{name}@{version}"))
1974 .join(memstead_schema::INSTALL_PROVENANCE_FILE);
1975 let bytes = if folder_stamp.is_file() {
1976 std::fs::read(&folder_stamp).ok()
1977 } else {
1978 let ops = self.git_branch_ops()?;
1979 let gitdir = self
1980 .mounts
1981 .iter()
1982 .find_map(|m| match &m.mount.storage {
1983 crate::workspace::MountStorage::GitBranch { gitdir, .. } => {
1984 Some(gitdir.clone())
1985 }
1986 _ => None,
1987 })
1988 .or_else(|| {
1989 let g = root.join("mem-repo").join(".git");
1990 g.is_dir().then_some(g)
1991 })?;
1992 (ops.read_schema_file)(
1993 &gitdir,
1994 name,
1995 version,
1996 memstead_schema::INSTALL_PROVENANCE_FILE,
1997 )
1998 .ok()
1999 .flatten()
2000 }?;
2001 let v: serde_json::Value = serde_json::from_slice(&bytes).ok()?;
2002 v.get("authoring_path")?.as_str().map(String::from)
2003 }
2004
2005 /// Engine-wide [`crate::ops::Status`] across every mount — the graph
2006 /// counts behind `memstead status` (renamed from `stats` with the
2007 /// command, D11; fields unchanged).
2008 pub fn status(&self) -> crate::ops::Status {
2009 let mut types_in_use: Vec<String> = self
2010 .store
2011 .all_entities()
2012 .filter(|e| !e.stub && !e.entity_type.is_empty())
2013 .map(|e| e.entity_type.clone())
2014 .collect();
2015 types_in_use.sort();
2016 types_in_use.dedup();
2017
2018 let mut edge_types: HashMap<String, usize> = HashMap::new();
2019 for id in self.store.all_ids() {
2020 for edge in self.store.outgoing(id) {
2021 *edge_types.entry(edge.rel_type.clone()).or_insert(0) += 1;
2022 }
2023 }
2024
2025 crate::ops::Status {
2026 entity_count: self.store.all_entities().filter(|e| !e.stub).count(),
2027 edge_count: self.store.edge_count(),
2028 edge_types,
2029 community_count: self.communities().count,
2030 mem_count: self.mounts.len(),
2031 types_in_use,
2032 }
2033 }
2034
2035 /// Build a [`ContextResult`] for `id`: the community cluster id
2036 /// (or `None` when the entity is a stub or not present), plus the
2037 /// outgoing + incoming neighbour lists.
2038 pub fn context(&self, id: &EntityId) -> Option<ContextResult> {
2039 let entity = self.store.get(id)?;
2040 let community = self
2041 .communities()
2042 .entity_cluster_map
2043 .get(id.as_ref())
2044 .cloned();
2045 let mut neighbors = Vec::new();
2046 for edge in self.store.outgoing(id) {
2047 if let Some(target) = self.store.get(&edge.target) {
2048 neighbors.push(NeighborInfo {
2049 id: target.id.clone(),
2050 title: target.title.clone(),
2051 relationship: edge.rel_type.clone(),
2052 direction: Direction::Outgoing,
2053 });
2054 }
2055 }
2056 for edge in self.store.incoming(id) {
2057 if let Some(source) = self.store.get(&edge.from) {
2058 neighbors.push(NeighborInfo {
2059 id: source.id.clone(),
2060 title: source.title.clone(),
2061 relationship: edge.rel_type.clone(),
2062 direction: Direction::Incoming,
2063 });
2064 }
2065 }
2066 Some(ContextResult {
2067 entity_id: entity.id.clone(),
2068 community,
2069 neighbors,
2070 })
2071 }
2072
2073 /// Lazily-built per-mem search index map. The map carries one
2074 /// entry per writable mem. Build cost scales with entity count;
2075 /// expect hundreds-of-ms for thousand-entity workspaces. Not
2076 /// available on `wasm32` targets — search lives behind the bridge
2077 /// (see [`Self::search`] for the typed refuse).
2078 #[cfg(not(target_arch = "wasm32"))]
2079 pub fn search_indexes(&self) -> &HashMap<String, MemIndex> {
2080 if let Some((memo_key, _)) = self.search_indexes_memo.get() {
2081 debug_assert_eq!(
2082 *memo_key,
2083 self.derived_key(),
2084 "search memo key lags the engine — a mutation path missed invalidate_search_indexes"
2085 );
2086 }
2087 &self
2088 .search_indexes_memo
2089 .get_or_init(|| (self.derived_key(), build_all(&self.store, &self.schemas)))
2090 .1
2091 }
2092
2093 /// Drop the cached per-mem search index map. No-op on `wasm32`
2094 /// where no index exists; the method stays present so mutation
2095 /// hooks can call it unconditionally.
2096 /// Incrementally maintain the search-index memo for a known
2097 /// touched-id set (flywheel W8/01, criterion 1): replace or remove
2098 /// exactly the touched documents in place and advance the memo's
2099 /// key, instead of dropping the whole map. Semantics:
2100 ///
2101 /// - Memo empty → nothing to maintain; the next read builds fresh.
2102 /// - Memo current (key matches) → no-op (rollback already landed).
2103 /// - Schemas epoch moved → the index FIELD SET may have changed:
2104 /// the named, scoped fallback — drop the memo for a full
2105 /// rebuild. Never a silent widening: this is the one case the
2106 /// plan names (schema-shape change).
2107 /// - Otherwise: per touched id, a real non-stub entity in the
2108 /// store is re-indexed (delete-then-add on the id term), and an
2109 /// absent or stub entry is removed — stubs stay excluded exactly
2110 /// as the bulk build excludes them. Touched mems' writers
2111 /// commit; any tantivy error falls back to dropping the memo
2112 /// (warn-logged), never to serving a stale index.
2113 #[cfg(not(target_arch = "wasm32"))]
2114 pub(crate) fn maintain_search_indexes(&mut self, touched: &[crate::EntityId]) {
2115 let current = super::DerivedKey {
2116 store_generation: self.store.generation(),
2117 schemas_epoch: self.schemas_epoch,
2118 };
2119 let Some((memo_key, indexes)) = self.search_indexes_memo.get_mut() else {
2120 return;
2121 };
2122 if *memo_key == current {
2123 return;
2124 }
2125 if memo_key.schemas_epoch != current.schemas_epoch {
2126 self.search_indexes_memo = OnceCell::new();
2127 return;
2128 }
2129 let mut touched_mems: std::collections::HashSet<&str> = std::collections::HashSet::new();
2130 for id in touched {
2131 let Some(idx) = indexes.get_mut(id.mem()) else {
2132 continue;
2133 };
2134 let result = match self.store.get(id) {
2135 Some(entity) if !entity.stub => idx.index_entity(entity),
2136 _ => idx.remove_entity(id),
2137 };
2138 if let Err(e) = result {
2139 tracing::warn!(
2140 id = id.as_ref(),
2141 error = %e,
2142 "incremental index maintenance failed; dropping the memo for a full rebuild"
2143 );
2144 self.search_indexes_memo = OnceCell::new();
2145 return;
2146 }
2147 touched_mems.insert(id.mem());
2148 }
2149 for (mem, idx) in indexes.iter_mut() {
2150 if !touched_mems.contains(mem.as_str()) {
2151 continue;
2152 }
2153 if let Err(e) = idx.commit() {
2154 tracing::warn!(
2155 mem = mem.as_str(),
2156 error = %e,
2157 "incremental index commit failed; dropping the memo for a full rebuild"
2158 );
2159 self.search_indexes_memo = OnceCell::new();
2160 return;
2161 }
2162 }
2163 *memo_key = current;
2164 }
2165
2166 /// No-op shim on `wasm32`, mirroring `invalidate_search_indexes`
2167 /// so mutation paths call it unconditionally.
2168 #[cfg(target_arch = "wasm32")]
2169 pub(crate) fn maintain_search_indexes(&mut self, _touched: &[crate::EntityId]) {}
2170
2171 /// Unconditionally drop the search-index memo, regardless of its
2172 /// generation. The forced variant exists for embedders that want
2173 /// to release the index's memory in a long-lived process (or force
2174 /// a from-scratch rebuild for verification); the generation-checked
2175 /// [`Self::invalidate_search_indexes`] stays the mutation-path
2176 /// hook.
2177 pub fn drop_search_indexes(&mut self) {
2178 #[cfg(not(target_arch = "wasm32"))]
2179 {
2180 self.search_indexes_memo = OnceCell::new();
2181 }
2182 }
2183
2184 pub fn invalidate_search_indexes(&mut self) {
2185 #[cfg(not(target_arch = "wasm32"))]
2186 {
2187 // Same generation check as `invalidate_communities`: keep
2188 // the memo when the store still sits at its generation
2189 // (the batch-rollback case), clear otherwise.
2190 if let Some((memo_key, _)) = self.search_indexes_memo.get()
2191 && *memo_key == self.derived_key()
2192 {
2193 return;
2194 }
2195 self.search_indexes_memo = OnceCell::new();
2196 }
2197 }
2198
2199 /// Filter the in-memory store by metadata only (no text match).
2200 #[cfg(not(target_arch = "wasm32"))]
2201 pub fn list(&self, scope: &SearchScope) -> crate::ops::ListResult {
2202 let fallback = engine_fallback_type();
2203 crate::ops::search::list(&self.store, scope, fallback.as_ref(), &self.schemas)
2204 }
2205
2206 /// Run a search against the lazily-built index map. Returns
2207 /// [`EngineError::SearchUnavailable`] on `wasm32` targets — browser
2208 /// consumers route search to the bridge; the local
2209 /// engine never builds a tantivy index in WASM. Native targets get
2210 /// the same shape as before, wrapped in `Ok`.
2211 pub fn search(&self, scope: &SearchScope) -> Result<SearchResult, EngineError> {
2212 // A mem filter naming a quarantined OR nonexistent mem refuses
2213 // typed `UNKNOWN_MEM`, matching every other mem-naming surface. A
2214 // success with 0 hits (the old nonexistent-mem behaviour, with a
2215 // missing-index warning) is indistinguishable from a true empty
2216 // result — the one thing a typed surface must never be.
2217 if let Some(mem) = scope.mem.as_deref()
2218 && (self.quarantine_reason(mem).is_some() || !self.mem_router.is_visible(mem))
2219 {
2220 return Err(self.unknown_mem_error(mem));
2221 }
2222 #[cfg(target_arch = "wasm32")]
2223 {
2224 let _ = scope;
2225 return Err(EngineError::SearchUnavailable);
2226 }
2227 #[cfg(not(target_arch = "wasm32"))]
2228 {
2229 let fallback = engine_fallback_type();
2230 Ok(crate::ops::search::search(
2231 &self.store,
2232 scope,
2233 fallback.as_ref(),
2234 self.search_indexes(),
2235 &self.schemas,
2236 ))
2237 }
2238 }
2239
2240 /// All mem-relative entity paths under `mem`. Delegates to
2241 /// the backend's `list_entities`. Order is backend-defined.
2242 pub fn list_entities(&self, mem: &str) -> Result<Vec<PathBuf>, EngineError> {
2243 let m = self.find_mount(mem)?;
2244 m.backend.list_entities().map_err(EngineError::Backend)
2245 }
2246
2247 /// Raw bytes for a single entity (`Ok(None)` if absent).
2248 pub fn read_entity(&self, mem: &str, rel_path: &Path) -> Result<Option<Vec<u8>>, EngineError> {
2249 let m = self.find_mount(mem)?;
2250 m.backend
2251 .read_entity(rel_path)
2252 .map_err(EngineError::Backend)
2253 }
2254
2255 /// Provenance entries for `mem` since `cursor`. Cursor shape is
2256 /// backend-specific (RFC-3339 timestamp for folder, commit SHA for
2257 /// git-branch); `None` means "from the beginning".
2258 pub fn read_provenance(
2259 &self,
2260 mem: &str,
2261 cursor: Option<&str>,
2262 ) -> Result<Vec<Provenance>, EngineError> {
2263 let m = self.find_mount(mem)?;
2264 m.backend
2265 .read_provenance(cursor)
2266 .map_err(EngineError::Backend)
2267 }
2268
2269 /// Capability declared on the mount for `mem`. Surfaced for
2270 /// callers that need to gate before dispatching a write — the
2271 /// engine itself does not yet enforce capability (mutation paths
2272 /// land in a later session).
2273 pub fn capability(&self, mem: &str) -> Result<crate::workspace::MountCapability, EngineError> {
2274 let m = self.find_mount(mem)?;
2275 Ok(m.mount.capability)
2276 }
2277
2278 /// Returns `true` when `from`'s source mem is mounted with
2279 /// [`crate::workspace::MountCapability::ReadOnly`]. Returns
2280 /// `false` for Write-Mems and for mems whose mount is absent
2281 /// from the router (no mount → no ReadOnly assertion can be
2282 /// made; the absence is treated as not-ReadOnly so consumers
2283 /// don't trip on transient lookup misses).
2284 ///
2285 /// Plan body §"Single edge source in the store" specifies this
2286 /// helper as the derived-on-demand alternative to adding a new
2287 /// field on [`crate::store::Edge`]. Strict-invariant validators
2288 /// and surfaces that want to highlight cross-mount references
2289 /// call this rather than pattern-matching on a per-edge marker.
2290 /// The information is fully derivable from the current mount
2291 /// roster, so no new state needs to live on the edge itself.
2292 pub fn edge_is_from_readonly(&self, from: &EntityId) -> bool {
2293 match self.capability(from.mem()) {
2294 Ok(crate::workspace::MountCapability::ReadOnly) => true,
2295 Ok(crate::workspace::MountCapability::Write) | Err(_) => false,
2296 }
2297 }
2298
2299 /// Whether a cross-mem edge from `from_mem` to `to_mem` is
2300 /// permitted under the current [`crate::WorkspaceSettings`]
2301 /// cross-mem link policy.
2302 ///
2303 /// Resolution rules (matches full's `mem_router` semantics):
2304 /// 1. Same-mem edge (`from_mem == to_mem`) → always
2305 /// allowed; the policy gates *cross*-mem edges only.
2306 /// 2. Explicit `cross_mem_links[from_mem]`:
2307 /// - `"*"` (wildcard) → allowed regardless of target.
2308 /// - `["a", ...]` (allowlist) → allowed iff `to_mem` is in
2309 /// the list.
2310 /// 3. Per-create-rule `default_cross_links` synthesis — if
2311 /// rule (1) didn't grant permission and `from_mem` matches
2312 /// a `[[mem_management.create]]` rule whose
2313 /// `default_cross_links` is set, the synthesised value
2314 /// contributes:
2315 /// - `"*"` → allowed regardless of target.
2316 /// - `["a", ...]` → allowed iff `to_mem` is in the list.
2317 /// 4. Otherwise → denied (default-deny posture).
2318 ///
2319 /// The synthesis layer compiles a [`crate::mem_management::CreateRuleSet`]
2320 /// lazily on first call and caches it; [`Self::set_settings`]
2321 /// invalidates the cache. Compilation failure (malformed glob
2322 /// in a rule) logs a warning and the synthesis layer is silently
2323 /// skipped — the resolver still returns `true` from explicit
2324 /// policy alone, so a half-broken config doesn't lock out edges
2325 /// the operator did intend to allow. Operators who want hard
2326 /// validation pre-compile via
2327 /// [`crate::mem_management::CreateRuleSet::new`] before
2328 /// calling [`Self::set_settings`].
2329 ///
2330 /// The MCP `memstead_relate` handler's cross-mem gate consumes
2331 /// this method directly.
2332 pub fn cross_mem_link_allowed(&self, from_mem: &str, to_mem: &str) -> bool {
2333 use memstead_schema::workspace_config::CrossLinkValue;
2334 if from_mem == to_mem {
2335 return true;
2336 }
2337
2338 // Step 1: explicit cross_mem_links policy.
2339 if let Some(value) = self.settings.cross_mem_links.get(from_mem) {
2340 match value {
2341 CrossLinkValue::Wildcard => return true,
2342 CrossLinkValue::List(targets) => {
2343 if targets.iter().any(|t| t == to_mem) {
2344 return true;
2345 }
2346 // Fall through to synthesis check — a List that
2347 // doesn't include the target may still allow it
2348 // via per-rule default_cross_links union.
2349 }
2350 }
2351 }
2352
2353 // Step 2: per-create-rule default_cross_links synthesis.
2354 let rule_set = self.create_rule_set_memo.get_or_init(|| {
2355 crate::mem_management::CreateRuleSet::new(
2356 self.settings.mem_create_rules.clone(),
2357 )
2358 .unwrap_or_else(|err| {
2359 tracing::warn!(
2360 error = %err,
2361 "cross_mem_link_allowed: failed to compile mem_create_rules — synthesis disabled (resolver falls back to explicit-policy-only)"
2362 );
2363 crate::mem_management::CreateRuleSet::default()
2364 })
2365 });
2366
2367 // Compose the same `<mem_path>/<name>` candidate the create-rule
2368 // composer matched against. The rule globs are keyed on the composed
2369 // lifecycle path (e.g. `memstead/project`, compiled with
2370 // `literal_separator`), not the bare leaf name — matching
2371 // `from_mem` alone silently misses, so synthesis denied a link
2372 // that `memstead_overview` rendered as rule-granted (the
2373 // leaf-vs-composed-path divergence). Flat-layout mems (no
2374 // hierarchical path) keep the bare leaf, matching their bare rule.
2375 let candidate = match self.mount(from_mem).and_then(|m| m.mem_path()) {
2376 Some(path) => format!("{path}/{from_mem}"),
2377 None => from_mem.to_string(),
2378 };
2379 if let Some(matched) = rule_set.first_match(std::path::Path::new(&candidate))
2380 && let Some(synth) = matched.default_cross_links.as_ref()
2381 {
2382 return match synth {
2383 CrossLinkValue::Wildcard => true,
2384 CrossLinkValue::List(targets) => targets.iter().any(|t| t == to_mem),
2385 };
2386 }
2387
2388 false
2389 }
2390
2391 pub(super) fn find_mount(&self, mem: &str) -> Result<&MountedBackend, EngineError> {
2392 self.mounts
2393 .iter()
2394 .find(|m| m.mount.mem == mem)
2395 .ok_or_else(|| self.unknown_mem_error(mem))
2396 }
2397}
2398
2399/// The base path of an anchor artifact ref — the locator suffixes a
2400/// medium may append (`@<commit>`, `#<span>`) stripped so the reverse
2401/// lookup compares paths, not versioned/located refs.
2402pub(crate) fn anchor_base_path(artifact: &str) -> &str {
2403 let cut = artifact.find(['@', '#']).unwrap_or(artifact.len());
2404 &artifact[..cut]
2405}
2406
2407/// One mem's standalone anchor-verification report — the counts plus
2408/// the per-anchor rows, in sidecar order.
2409#[derive(Debug, Clone, Default, serde::Serialize)]
2410pub struct MemAnchorVerification {
2411 pub mem: String,
2412 /// Source present, hash matches (or a non-hash class whose source
2413 /// exists).
2414 pub resolved: usize,
2415 /// Source present, hash differs, stability `stable` — real drift.
2416 pub drifted: usize,
2417 /// Hash differs under `unstable` stability, or a hash is missing on
2418 /// either side — flagged for re-examination, never called drift.
2419 pub recheck: usize,
2420 /// Source absent, or a grain/medium the mechanism does not reach.
2421 pub unresolvable: usize,
2422 pub anchors: Vec<VerifiedAnchor>,
2423}
2424
2425/// One anchor's verification row.
2426#[derive(Debug, Clone, serde::Serialize)]
2427pub struct VerifiedAnchor {
2428 pub entity_id: String,
2429 pub artifact: String,
2430 pub grain: String,
2431 pub class: String,
2432 /// `resolved` | `drifted` | `recheck` | `unresolvable`.
2433 pub state: String,
2434 #[serde(skip_serializing_if = "Option::is_none")]
2435 pub observed_hash: Option<String>,
2436}
2437
2438/// Whether `anchor` references `path`. `tree`-grain anchors match `path`
2439/// itself and anything beneath the tree; every other grain matches by
2440/// exact base-path equality.
2441/// A stored anchor paired with its live resolution state, when observable.
2442/// See [`Engine::entity_anchors_resolved`] for how `state` is produced and
2443/// when it is `None` (unobserved, never fabricated).
2444#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize)]
2445pub struct ResolvedAnchor {
2446 /// The durable anchor record (flattened on the wire so the resolved shape
2447 /// is the stored anchor plus a `state` field).
2448 #[serde(flatten)]
2449 pub anchor: crate::anchor::Anchor,
2450 /// The live resolution state, or `None` when the engine could not observe
2451 /// the source artifact this pass: a `url` grain, no workspace root, an
2452 /// ambiguous or absent path medium, or an `entity` grain whose mem is not
2453 /// mounted. That last case is load-bearing — an unmounted mem is not a mem
2454 /// of deleted entities, so it must read as unobserved rather than
2455 /// `Orphaned`, which prune would act on.
2456 #[serde(skip_serializing_if = "Option::is_none")]
2457 pub state: Option<crate::anchor::AnchorState>,
2458 /// The prepared-content hash the observation computed this pass —
2459 /// present only for a hash-bearing (`anchored` / `derived`) anchor whose
2460 /// artifact could be read: a `file` / `span` anchor resolving to a
2461 /// readable file, or an `entity` anchor whose mem is mounted (hashed over
2462 /// the canonical rendered markdown, so the value means the same thing in
2463 /// both namespaces). The verify pass's backfill leg records it onto a
2464 /// hash-less anchor. Engine-internal
2465 /// observation detail, deliberately not serialized: the wire shape stays
2466 /// the stored anchor plus `state`.
2467 #[serde(skip)]
2468 pub observed_hash: Option<String>,
2469}
2470
2471/// What an anchor's `source` name resolves to in its mem's bindings: the
2472/// declared pointer (the filesystem root a source-dialect artifact path
2473/// joins onto, decision 26) and the declared preparation (what the
2474/// preparation registry prepares the artifact as before hashing —
2475/// touchpoint A of [`crate::preparation`]).
2476#[derive(Debug, Clone, PartialEq, Eq)]
2477pub(crate) struct AnchorSourceJoin {
2478 /// The source's declared `pointer`.
2479 pub(crate) pointer: String,
2480 /// The source's declared `preparation`, if any.
2481 pub(crate) preparation: Option<String>,
2482 /// The declaring source itself — its scope is what a `tree` anchor's
2483 /// prepared form enumerates under a code-map preparation.
2484 pub(crate) source: crate::pipeline::Source,
2485 /// The binding's `deny_paths`, applied on top of the source scope.
2486 pub(crate) deny_paths: Vec<String>,
2487}
2488
2489/// Observe a single path-namespace anchor against `root` (its medium's
2490/// filesystem root) and resolve its live state plus — for a present
2491/// hash-bearing (`anchored` / `derived`) `file` / `span` anchor — the
2492/// artifact's **prepared-content hash**
2493/// ([`crate::anchor::prepared_content_hash`]). `None` when the anchor's
2494/// grain does not reference a filesystem path.
2495///
2496/// The computed hash is what lets [`crate::anchor::resolve_anchor`]
2497/// adjudicate `drifted` vs `resolves` deterministically against the recorded
2498/// hash. The prepared form is the registry's rule for the anchor's
2499/// source's preparation ([`crate::preparation::path_prepared_hash`]): a
2500/// `span` anchor hashes its whole containing file (the span locator selects
2501/// within it; the file is the hashed unit), except under a **delivery
2502/// preparation**, where a `<path>#<key>` span names one delivery unit (the
2503/// unit's own text is the hashed unit, and a key the file no longer yields
2504/// is an absent artifact); under a **code-map** preparation a file or span
2505/// hashes the interface digest, and a `tree` hashes the code map of every
2506/// scoped file under it. A `tree` grain under no code map has no prepared
2507/// form and observes no hash; a read failure likewise observes no hash —
2508/// those resolve `recheck`, never a fabricated `drifted`. Non-hash classes
2509/// (`authored` / `informed-by`) skip the read entirely, so an anchor-less or
2510/// hash-free mem pays no observation cost.
2511fn observe_path_anchor(
2512 root: &Path,
2513 anchor: &crate::anchor::Anchor,
2514 join: Option<&AnchorSourceJoin>,
2515) -> Option<(crate::anchor::AnchorState, Option<String>)> {
2516 use crate::anchor::AnchorGrain;
2517 match anchor.grain {
2518 AnchorGrain::Span | AnchorGrain::File | AnchorGrain::Tree => {}
2519 AnchorGrain::Url | AnchorGrain::Entity => return None,
2520 }
2521 let source_pointer = join.map(|j| j.pointer.as_str());
2522 let preparation = join.and_then(|j| j.preparation.as_deref());
2523 let base = anchor_base_path(&anchor.artifact);
2524 // Decision 29: the source-join is authoritative — an artifact path is
2525 // source-relative first (joined onto the declaring source's pointer,
2526 // which may deliberately leave the workspace root for out-of-root
2527 // pointers); the workspace-relative form is tried only when the
2528 // source-join does not resolve. A path resolving under both joins is
2529 // decided by this priority, deterministically.
2530 let path = source_pointer
2531 .map(|pointer| root.join(join_pointer(pointer, base)))
2532 .filter(|joined| joined.exists())
2533 .unwrap_or_else(|| root.join(base));
2534 if !path.exists() {
2535 return Some((
2536 crate::anchor::resolve_anchor(anchor, &crate::anchor::ArtifactObservation::Absent),
2537 None,
2538 ));
2539 }
2540 let current_hash = if !anchor.class.is_hash_bearing() {
2541 None
2542 } else if matches!(anchor.grain, AnchorGrain::File | AnchorGrain::Span) && path.is_file() {
2543 match std::fs::read(&path).ok().map(|bytes| {
2544 crate::preparation::path_prepared_hash(
2545 preparation,
2546 &anchor.artifact,
2547 anchor.grain,
2548 &bytes,
2549 )
2550 }) {
2551 Some(crate::preparation::PathPrepared::Hash(h)) => Some(h),
2552 Some(crate::preparation::PathPrepared::UnitAbsent) => {
2553 return Some((
2554 crate::anchor::resolve_anchor(
2555 anchor,
2556 &crate::anchor::ArtifactObservation::Absent,
2557 ),
2558 None,
2559 ));
2560 }
2561 Some(crate::preparation::PathPrepared::NoHash) | None => None,
2562 }
2563 } else if anchor.grain == AnchorGrain::Tree
2564 && path.is_dir()
2565 && preparation == Some(crate::preparation::CODE_MAP)
2566 && let Some(join) = join
2567 {
2568 // The tree's code map: every scoped file under the tree, by the
2569 // declaring source's own scope and the binding's deny paths. The
2570 // path the anchor names is workspace-relative or source-relative;
2571 // the enumeration is workspace-relative, so compare the resolved
2572 // absolute paths.
2573 let files: Vec<(String, String)> =
2574 crate::ingest::cursor::enumerate_facet_files(&join.source, &join.deny_paths, root)
2575 .into_iter()
2576 .filter(|f| root.join(f).starts_with(&path))
2577 .filter_map(|f| {
2578 std::fs::read(root.join(&f))
2579 .ok()
2580 .map(|bytes| (f, String::from_utf8_lossy(&bytes).into_owned()))
2581 })
2582 .collect();
2583 Some(crate::anchor::prepared_content_hash(
2584 crate::preparation::code_map_tree_digest(&files).as_bytes(),
2585 ))
2586 } else {
2587 None
2588 };
2589 let observation = crate::anchor::ArtifactObservation::Present {
2590 current_hash: current_hash.clone(),
2591 };
2592 Some((
2593 crate::anchor::resolve_anchor(anchor, &observation),
2594 current_hash,
2595 ))
2596}
2597
2598/// Deterministic fingerprint of a PARSED schema for the
2599/// authoring-drift equivalence check. Compares semantic content, never
2600/// raw bytes: YAML comments (the CLI-injected editor-header lines) and
2601/// whitespace vanish at parse time, and `Schema.types` — a `HashMap`
2602/// with nondeterministic iteration order — is rendered sorted by type
2603/// name so two loads of equivalent packages always fingerprint alike.
2604fn schema_parsed_fingerprint(schema: &memstead_schema::Schema) -> String {
2605 let mut keys: Vec<&String> = schema.types.keys().collect();
2606 keys.sort();
2607 let types: Vec<String> = keys
2608 .iter()
2609 .map(|k| format!("{k}={:?}", schema.types[k.as_str()]))
2610 .collect();
2611 format!(
2612 "{:?}|{}|{}",
2613 schema.manifest,
2614 schema.version,
2615 types.join(";")
2616 )
2617}
2618
2619fn anchor_references_path(anchor: &crate::anchor::Anchor, path: &str) -> bool {
2620 let base = anchor_base_path(&anchor.artifact);
2621 path_references(base, anchor.grain == crate::anchor::AnchorGrain::Tree, path)
2622}
2623
2624/// Whether `base` (a file path, or a tree root when `is_tree`) references
2625/// `path` — exact match, or containment for a tree.
2626fn path_references(base: &str, is_tree: bool, path: &str) -> bool {
2627 if base == path {
2628 return true;
2629 }
2630 if is_tree {
2631 let prefix = base.strip_suffix('/').unwrap_or(base);
2632 return path.starts_with(&format!("{prefix}/"));
2633 }
2634 false
2635}
2636
2637/// Join a source pointer and a source-relative artifact path into the
2638/// pointer-joined (workspace-relative) form — the decision-26 dialect
2639/// bridge. Plain string concatenation with a separator: the pointer is
2640/// workspace-relative (and may climb out via `..`), the artifact is
2641/// source-relative; no canonicalization here, the filesystem resolves it.
2642pub(crate) fn join_pointer(pointer: &str, base: &str) -> String {
2643 let pointer = pointer.trim_end_matches('/');
2644 if pointer.is_empty() || pointer == "." {
2645 base.to_string()
2646 } else {
2647 format!("{pointer}/{base}")
2648 }
2649}
2650
2651#[cfg(test)]
2652mod tests {
2653 use std::path::Path;
2654
2655 use tempfile::TempDir;
2656
2657 use crate::backend::{BackendError, MemBackend};
2658 use crate::engine::test_helpers::*;
2659 use crate::engine::{Engine, EngineError, RelateEntityArgs};
2660 use crate::entity::EntityId;
2661 use crate::ops::{Direction, SearchScope, WarningHint};
2662 use crate::provenance::Provenance;
2663 use crate::storage::{ArchiveBackend, FilesystemMemWriter, MemWriter};
2664
2665 use crate::vcs::CommitContext;
2666 use crate::workspace::{Mount, MountCapability, MountLifecycle, MountStorage};
2667
2668 /// `schema_origin` is the trust-classification authority: a built-in
2669 /// (or workspace-authored) schema is first-party; a schema whose
2670 /// `(name, version)` is in neither catalogue is third-party — the safe
2671 /// default for an origin the engine cannot vouch for.
2672 #[test]
2673 fn schema_origin_classifies_builtin_first_party_and_unknown_third_party() {
2674 use std::sync::Arc;
2675
2676 use crate::render::OriginClass;
2677
2678 let tmp = TempDir::new().unwrap();
2679 let engine = Engine::from_mounts(vec![(
2680 folder_mount("specs", tmp.path().to_path_buf()),
2681 Box::new(FilesystemMemWriter::new(tmp.path().to_path_buf())) as Box<dyn MemBackend>,
2682 )])
2683 .unwrap();
2684
2685 // A built-in schema (the catalogue the engine resolved against).
2686 let builtin = engine.builtin_schemas()[0].clone();
2687 assert_eq!(
2688 engine.schema_origin(&builtin),
2689 OriginClass::FirstParty,
2690 "a built-in schema is first-party"
2691 );
2692
2693 // A schema whose version is in no catalogue — a stand-in for a
2694 // schema that entered from outside the workspace. Same name, a
2695 // version the engine never loaded.
2696 let foreign = Arc::new(memstead_schema::Schema {
2697 manifest: builtin.manifest.clone(),
2698 version: semver::Version::new(99, 0, 0),
2699 types: builtin.types.clone(),
2700 });
2701 assert_eq!(
2702 engine.schema_origin(&foreign),
2703 OriginClass::ThirdParty,
2704 "a schema in neither catalogue classifies third-party (safe default)"
2705 );
2706 }
2707
2708 /// `mem_origin_class` classifies a writable mount first-party (its
2709 /// content is authored in this workspace) and a read-only mount
2710 /// third-party (registry-installed read-mem or adopted foreign
2711 /// folder/clone — quoted, untrusted data). An unknown mem is
2712 /// third-party (the safe default).
2713 #[test]
2714 fn mem_origin_class_writable_first_party_readonly_third_party() {
2715 use crate::render::OriginClass;
2716
2717 let tmp = TempDir::new().unwrap();
2718 // Writable folder mem.
2719 let writable_dir = tmp.path().join("writable");
2720 std::fs::create_dir_all(&writable_dir).unwrap();
2721 let writer = FilesystemMemWriter::new(writable_dir.clone());
2722
2723 // Read-only archive mem.
2724 let body = "---\ntype: spec\n---\n# Ext\n\n## Identity\n\nFrom an archive.\n";
2725 let archive_path = build_archive(tmp.path(), "ext", &[("ext.md", body.as_bytes())]);
2726
2727 let engine = Engine::from_mounts(vec![
2728 (
2729 folder_mount("local", writable_dir),
2730 Box::new(writer) as Box<dyn MemBackend>,
2731 ),
2732 (
2733 archive_mount("external", archive_path.clone()),
2734 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
2735 ),
2736 ])
2737 .unwrap();
2738
2739 assert_eq!(
2740 engine.mem_origin_class("local"),
2741 OriginClass::FirstParty,
2742 "a writable mount is first-party"
2743 );
2744 assert_eq!(
2745 engine.mem_origin_class("external"),
2746 OriginClass::ThirdParty,
2747 "a read-only mount is third-party"
2748 );
2749 assert_eq!(
2750 engine.mem_origin_class("no-such-mem"),
2751 OriginClass::ThirdParty,
2752 "an unknown mem is third-party (safe default)"
2753 );
2754 }
2755
2756 /// `declare_mem_origin` lets the embedding deployment vouch for one
2757 /// read-only mount as first-party (the curated hosted read tier),
2758 /// overriding the writability inference for that mem only — sibling
2759 /// read-only mounts keep the safe third-party default.
2760 #[test]
2761 fn declared_origin_overrides_inference_per_mem() {
2762 use crate::render::OriginClass;
2763
2764 let tmp = TempDir::new().unwrap();
2765 let body = "---\ntype: spec\n---\n# Ext\n\n## Identity\n\nFrom an archive.\n";
2766 let vouched_path = build_archive(tmp.path(), "vouched", &[("v.md", body.as_bytes())]);
2767 let other_path = build_archive(tmp.path(), "other", &[("o.md", body.as_bytes())]);
2768
2769 let mut engine = Engine::from_mounts(vec![
2770 (
2771 archive_mount("vouched", vouched_path.clone()),
2772 Box::new(ArchiveBackend::new(vouched_path)) as Box<dyn MemBackend>,
2773 ),
2774 (
2775 archive_mount("other", other_path.clone()),
2776 Box::new(ArchiveBackend::new(other_path)) as Box<dyn MemBackend>,
2777 ),
2778 ])
2779 .unwrap();
2780
2781 engine.declare_mem_origin("vouched", OriginClass::FirstParty);
2782
2783 assert_eq!(
2784 engine.mem_origin_class("vouched"),
2785 OriginClass::FirstParty,
2786 "the deployment's declaration wins over the read-only inference"
2787 );
2788 assert_eq!(
2789 engine.mem_origin_class("other"),
2790 OriginClass::ThirdParty,
2791 "an undeclared sibling mount keeps the safe default"
2792 );
2793 }
2794
2795 /// The adopt-gate: a non-built-in schema is first-party only once a
2796 /// writable mount pins it (the operator authors against it here).
2797 /// Pinned only by a read-only mount — a registry read-mem or an
2798 /// adopted foreign folder/clone — it stays third-party, so
2799 /// `memstead_schema` serves it structural-only.
2800 #[test]
2801 fn schema_origin_third_party_until_pinned_by_a_writable_mount() {
2802 use memstead_schema::SchemaRef;
2803
2804 use crate::render::OriginClass;
2805
2806 let manifest = r#"name: trust-test
2807version: 0.1.0
2808description: adopt-gate test schema
2809when_to_use: tests
2810types:
2811 - doc
2812relationships:
2813 mode: strict
2814 definitions:
2815 - name: _default
2816 description: fallback
2817 default_weight: 1.0
2818community:
2819 resolution: 1.0
2820 seed: 42
2821"#;
2822 let pin = SchemaRef::new("trust-test", semver::Version::new(0, 1, 0));
2823
2824 let mk_engine = |cap: MountCapability| -> Engine {
2825 let tmp = TempDir::new().unwrap();
2826 let schemas_dir = tmp.path().join("schemas");
2827 std::fs::create_dir_all(&schemas_dir).unwrap();
2828 write_schema_files_with_default_type(&schemas_dir, "trust-test", manifest, &["doc"]);
2829 let mem_dir = tmp.path().join("mem");
2830 std::fs::create_dir_all(&mem_dir).unwrap();
2831 let mount = Mount {
2832 mem: "v".to_string(),
2833 schema: Some(pin.clone()),
2834 storage: MountStorage::Folder {
2835 path: mem_dir.clone(),
2836 },
2837 capability: cap,
2838 lifecycle: MountLifecycle::Eager,
2839 cross_linkable: true,
2840 migration_target: None,
2841 };
2842 let backend = Box::new(FilesystemMemWriter::new(mem_dir)) as Box<dyn MemBackend>;
2843 // Keep `tmp` alive for the engine's lifetime by leaking it —
2844 // the test process is short-lived and the folder must outlast
2845 // the closure.
2846 std::mem::forget(tmp);
2847 Engine::from_mounts_with_schemas_dir(vec![(mount, backend)], Some(&schemas_dir))
2848 .unwrap()
2849 };
2850
2851 // Read-only mount: the foreign schema is never adopted → third-party.
2852 let ro = mk_engine(MountCapability::ReadOnly);
2853 let schema = ro.schemas().get("v").expect("schema resolved").clone();
2854 assert_eq!(
2855 ro.schema_origin(&schema),
2856 OriginClass::ThirdParty,
2857 "a non-built-in schema pinned only by a read-only mount is third-party"
2858 );
2859
2860 // Writable mount pinning the same schema: adopted → first-party.
2861 let rw = mk_engine(MountCapability::Write);
2862 let schema = rw.schemas().get("v").expect("schema resolved").clone();
2863 assert_eq!(
2864 rw.schema_origin(&schema),
2865 OriginClass::FirstParty,
2866 "a writable mount pinning the schema adopts it → first-party"
2867 );
2868 }
2869
2870 /// Consumer read path: an installed (archive-backed) mem that ships
2871 /// a `.memstead/provenance.json` payload surfaces per-entity authoring
2872 /// provenance through `archive_provenance_for`. A noted entity carries
2873 /// its rationale; an entity authored without a note is absent from the
2874 /// payload and reads as provenance-absent (no fabricated value); the
2875 /// `history` disposition records that full history is not shipped.
2876 #[test]
2877 fn archive_provenance_surfaces_per_entity_and_reports_absence() {
2878 use memstead_schema::History;
2879
2880 let tmp = TempDir::new().unwrap();
2881 let config = br#"{"format":3,"name":"seed","version":"0.1.0","schema":"default@1.0.0"}"#;
2882 let alpha = b"---\ntype: spec\n---\n# Alpha\n\n## Identity\n\na\n\n## Purpose\n\np\n";
2883 let beta = b"---\ntype: spec\n---\n# Beta\n\n## Identity\n\nb\n\n## Purpose\n\np\n";
2884 // alpha noted; beta deliberately absent from the payload.
2885 let provenance = br#"{"format":1,"history":"summarised","entities":{"alpha":{"rationale":"why alpha exists","kind":"create","timestamp":"2026-06-24T00:00:00Z","actor":"agent"}}}"#;
2886 let archive = build_archive(
2887 tmp.path(),
2888 "seed",
2889 &[
2890 (".memstead/config.json", config),
2891 ("alpha.md", alpha),
2892 ("beta.md", beta),
2893 (".memstead/provenance.json", provenance),
2894 ],
2895 );
2896 let engine = Engine::from_mounts(vec![(
2897 archive_mount("seed", archive.clone()),
2898 Box::new(ArchiveBackend::new(archive)) as Box<dyn MemBackend>,
2899 )])
2900 .unwrap();
2901
2902 let prov = engine
2903 .archive_provenance_for("seed")
2904 .expect("provenance payload read from the archive");
2905 assert_eq!(
2906 prov.history,
2907 History::Summarised,
2908 "history-not-shipped is observable"
2909 );
2910 assert_eq!(
2911 prov.entity("alpha").and_then(|r| r.rationale.as_deref()),
2912 Some("why alpha exists"),
2913 "noted entity surfaces its rationale"
2914 );
2915 assert!(
2916 prov.entity("beta").is_none(),
2917 "unnoted entity is absent (reported absent, not fabricated)"
2918 );
2919 }
2920
2921 /// A pre-provenance archive (no `.memstead/provenance.json`) reads as
2922 /// provenance uniformly absent — the additive contract: a newer engine
2923 /// installing an old archive reports no provenance, never an error.
2924 #[test]
2925 fn archive_without_provenance_reports_absent() {
2926 let tmp = TempDir::new().unwrap();
2927 let config = br#"{"format":3,"name":"seed","version":"0.1.0","schema":"default@1.0.0"}"#;
2928 let alpha = b"---\ntype: spec\n---\n# Alpha\n\n## Identity\n\na\n\n## Purpose\n\np\n";
2929 let archive = build_archive(
2930 tmp.path(),
2931 "seed",
2932 &[(".memstead/config.json", config), ("alpha.md", alpha)],
2933 );
2934 let engine = Engine::from_mounts(vec![(
2935 archive_mount("seed", archive.clone()),
2936 Box::new(ArchiveBackend::new(archive)) as Box<dyn MemBackend>,
2937 )])
2938 .unwrap();
2939 assert!(
2940 engine.archive_provenance_for("seed").is_none(),
2941 "an archive without a provenance payload reports provenance absent"
2942 );
2943 }
2944
2945 #[test]
2946 fn folder_mount_routes_reads_to_filesystem_backend() {
2947 let tmp = TempDir::new().unwrap();
2948 let mem_dir = tmp.path().to_path_buf();
2949 let writer = FilesystemMemWriter::new(mem_dir.clone());
2950 // MemWriter and MemBackend share method names; the
2951 // module-top `use` brings both into scope. Seed via fully-
2952 // qualified MemWriter calls so dot-syntax stays unambiguous.
2953 <FilesystemMemWriter as MemWriter>::write_entity(&writer, Path::new("a.md"), b"alpha")
2954 .unwrap();
2955 <FilesystemMemWriter as MemWriter>::commit(&writer, "seed", &CommitContext::internal())
2956 .unwrap();
2957
2958 let engine = Engine::from_mounts(vec![(
2959 folder_mount("specs", mem_dir),
2960 Box::new(writer) as Box<dyn MemBackend>,
2961 )])
2962 .unwrap();
2963
2964 let mut paths: Vec<String> = engine
2965 .list_entities("specs")
2966 .unwrap()
2967 .into_iter()
2968 .map(|p| p.to_string_lossy().into_owned())
2969 .collect();
2970 paths.sort();
2971 assert_eq!(paths, vec!["a.md".to_string()]);
2972
2973 assert_eq!(
2974 engine.read_entity("specs", Path::new("a.md")).unwrap(),
2975 Some(b"alpha".to_vec())
2976 );
2977 }
2978
2979 #[test]
2980 fn heterogeneous_mounts_route_to_correct_backend() {
2981 let tmp = TempDir::new().unwrap();
2982
2983 // Folder mem.
2984 let folder_dir = tmp.path().join("folder-mem");
2985 std::fs::create_dir_all(&folder_dir).unwrap();
2986 let folder_writer = FilesystemMemWriter::new(folder_dir.clone());
2987 <FilesystemMemWriter as MemWriter>::write_entity(
2988 &folder_writer,
2989 Path::new("local.md"),
2990 b"local",
2991 )
2992 .unwrap();
2993 <FilesystemMemWriter as MemWriter>::commit(
2994 &folder_writer,
2995 "seed",
2996 &CommitContext::internal(),
2997 )
2998 .unwrap();
2999
3000 // Archive mem.
3001 let archive_path = build_archive(
3002 tmp.path(),
3003 "external",
3004 &[("ext.md", b"external"), ("dir/nested.md", b"nested")],
3005 );
3006
3007 let engine = Engine::from_mounts(vec![
3008 (
3009 folder_mount("local", folder_dir),
3010 Box::new(folder_writer) as Box<dyn MemBackend>,
3011 ),
3012 (
3013 archive_mount("external", archive_path.clone()),
3014 Box::new(ArchiveBackend::new(archive_path)),
3015 ),
3016 ])
3017 .unwrap();
3018
3019 // Routes correctly by mem name.
3020 assert_eq!(engine.mem_names(), vec!["local", "external"]);
3021 assert_eq!(
3022 engine.read_entity("local", Path::new("local.md")).unwrap(),
3023 Some(b"local".to_vec())
3024 );
3025 assert_eq!(
3026 engine.read_entity("external", Path::new("ext.md")).unwrap(),
3027 Some(b"external".to_vec())
3028 );
3029 assert_eq!(
3030 engine
3031 .read_entity("external", Path::new("dir/nested.md"))
3032 .unwrap(),
3033 Some(b"nested".to_vec())
3034 );
3035 // Cross-routing: reading a path from the wrong mem → None
3036 // (the backend doesn't have it), not an error.
3037 assert_eq!(
3038 engine.read_entity("local", Path::new("ext.md")).unwrap(),
3039 None
3040 );
3041 assert_eq!(
3042 engine
3043 .read_entity("external", Path::new("local.md"))
3044 .unwrap(),
3045 None
3046 );
3047 }
3048
3049 #[test]
3050 fn edge_is_from_readonly_classifies_every_edge_by_source_mount_capability() {
3051 // `engine.edge_is_from_readonly` is the derived-on-demand
3052 // alternative to adding a per-edge marker: construct a mixed
3053 // workspace (one Write-Mem + one ReadOnly archive with
3054 // cross-mem wiki-links) and walk every edge in the store,
3055 // asserting each edge's source-mount capability.
3056 let tmp = TempDir::new().unwrap();
3057
3058 // Write folder mem `local` with a spec-shaped entity that
3059 // declares an explicit cross-mem relation into the archive
3060 // (under the alias model edges originate from `## Relationships`).
3061 let folder_dir = tmp.path().join("local-mem");
3062 std::fs::create_dir_all(&folder_dir).unwrap();
3063 let folder_writer = FilesystemMemWriter::new(folder_dir.clone());
3064 let local_md = b"---\ntype: spec\n---\n# Note\n\n## Identity\n\nsee [[external:archived]] for prior context.\n\n## Relationships\n\n- **REFERENCES**: [[external:archived]]\n";
3065 <FilesystemMemWriter as MemWriter>::write_entity(
3066 &folder_writer,
3067 Path::new("note.md"),
3068 local_md,
3069 )
3070 .unwrap();
3071 <FilesystemMemWriter as MemWriter>::commit(
3072 &folder_writer,
3073 "seed",
3074 &CommitContext::internal(),
3075 )
3076 .unwrap();
3077
3078 // ReadOnly archive mem `external` with a spec-shaped entity
3079 // declaring an explicit cross-mem relation back to the local
3080 // note.
3081 let archive_md = b"---\ntype: spec\n---\n# Archived\n\n## Identity\n\nrefers back to [[local:note]] for the current revision.\n\n## Relationships\n\n- **REFERENCES**: [[local:note]]\n";
3082 let archive_path = build_archive(tmp.path(), "external", &[("archived.md", archive_md)]);
3083
3084 let engine = Engine::from_mounts(vec![
3085 (
3086 folder_mount("local", folder_dir),
3087 Box::new(folder_writer) as Box<dyn MemBackend>,
3088 ),
3089 (
3090 archive_mount("external", archive_path.clone()),
3091 Box::new(ArchiveBackend::new(archive_path)),
3092 ),
3093 ])
3094 .unwrap();
3095
3096 // Sanity: both entities are real, both mems are mounted.
3097 let local_id = EntityId::new("local", "note");
3098 let archived_id = EntityId::new("external", "archived");
3099 assert!(engine.get_entity(&local_id).is_some());
3100 assert!(engine.get_entity(&archived_id).is_some());
3101 assert!(matches!(
3102 engine.capability("local").unwrap(),
3103 MountCapability::Write
3104 ));
3105 assert!(matches!(
3106 engine.capability("external").unwrap(),
3107 MountCapability::ReadOnly
3108 ));
3109
3110 // Walk every edge in the store. For each (from, edge) pair,
3111 // `edge_is_from_readonly(from)` must return true iff the
3112 // source mount's capability is ReadOnly. The fixture's two
3113 // wiki-links produce one edge from each mem — both halves
3114 // exercise both branches of the helper.
3115 let mut seen_write_edge = false;
3116 let mut seen_readonly_edge = false;
3117 for from in engine.store().all_ids().cloned().collect::<Vec<_>>() {
3118 for _edge in engine.store().outgoing(&from) {
3119 let is_ro = engine.edge_is_from_readonly(&from);
3120 match engine.capability(from.mem()).unwrap() {
3121 MountCapability::Write => {
3122 assert!(
3123 !is_ro,
3124 "edge from write mem {} reported as ReadOnly",
3125 from.mem()
3126 );
3127 seen_write_edge = true;
3128 }
3129 MountCapability::ReadOnly => {
3130 assert!(
3131 is_ro,
3132 "edge from readonly mem {} reported as Write",
3133 from.mem()
3134 );
3135 seen_readonly_edge = true;
3136 }
3137 }
3138 }
3139 }
3140 assert!(
3141 seen_write_edge,
3142 "fixture must produce at least one edge from a write mem"
3143 );
3144 assert!(
3145 seen_readonly_edge,
3146 "fixture must produce at least one edge from a readonly mem"
3147 );
3148
3149 // Helper also reports `false` for mems absent from the
3150 // router — no mount → no ReadOnly assertion can be made.
3151 let phantom = EntityId::new("missing-mem", "phantom");
3152 assert!(
3153 !engine.edge_is_from_readonly(&phantom),
3154 "absent mount must not be reported as ReadOnly"
3155 );
3156 }
3157
3158 // ---- Engine::changes_since wrapper ------------------------------
3159
3160 #[test]
3161 fn cross_mem_link_allowed_same_mem_always_true() {
3162 // Self-edges (from == to) bypass the cross-mem policy
3163 // entirely — the policy gates *cross*-mem edges only.
3164 let tmp = TempDir::new().unwrap();
3165 let engine = build_demo_engine(&tmp);
3166 assert!(engine.cross_mem_link_allowed("specs", "specs"));
3167 // Even when the mem doesn't exist (not enrolled in
3168 // settings.cross_mem_links), same-mem returns true —
3169 // the engine doesn't validate mem existence here, just the
3170 // policy.
3171 assert!(engine.cross_mem_link_allowed("anywhere", "anywhere"));
3172 }
3173
3174 #[test]
3175 fn cross_mem_link_allowed_absent_denies_by_default() {
3176 // No entry in cross_mem_links for `from_mem` → denied.
3177 // Default-deny is the V1 posture; operators opt in.
3178 let tmp = TempDir::new().unwrap();
3179 let engine = build_demo_engine(&tmp);
3180 assert!(!engine.cross_mem_link_allowed("specs", "engine"));
3181 assert!(!engine.cross_mem_link_allowed("missing", "anywhere"));
3182 }
3183
3184 #[test]
3185 fn cross_mem_link_allowed_wildcard_admits_any_target() {
3186 use memstead_schema::workspace_config::CrossLinkValue;
3187 let tmp = TempDir::new().unwrap();
3188 let mut engine = build_demo_engine(&tmp);
3189 let mut settings = crate::workspace::WorkspaceSettings::default();
3190 settings
3191 .cross_mem_links
3192 .insert("specs".to_string(), CrossLinkValue::Wildcard);
3193 engine.set_settings(settings);
3194 assert!(engine.cross_mem_link_allowed("specs", "engine"));
3195 assert!(engine.cross_mem_link_allowed("specs", "macos"));
3196 assert!(engine.cross_mem_link_allowed("specs", "any-other"));
3197 // Reverse direction is independent — no policy entry for
3198 // engine→specs means denied.
3199 assert!(!engine.cross_mem_link_allowed("engine", "specs"));
3200 }
3201
3202 #[test]
3203 fn cross_mem_link_allowed_allowlist_enforces_membership() {
3204 use memstead_schema::workspace_config::CrossLinkValue;
3205 let tmp = TempDir::new().unwrap();
3206 let mut engine = build_demo_engine(&tmp);
3207 let mut settings = crate::workspace::WorkspaceSettings::default();
3208 settings.cross_mem_links.insert(
3209 "specs".to_string(),
3210 CrossLinkValue::List(vec!["engine".to_string(), "macos".to_string()]),
3211 );
3212 engine.set_settings(settings);
3213 assert!(engine.cross_mem_link_allowed("specs", "engine"));
3214 assert!(engine.cross_mem_link_allowed("specs", "macos"));
3215 assert!(!engine.cross_mem_link_allowed("specs", "external"));
3216 }
3217
3218 #[test]
3219 fn cross_mem_link_allowed_synthesises_from_matching_create_rule_wildcard() {
3220 // No explicit cross_mem_links entry, but a create rule
3221 // matches `from_mem` and carries default_cross_links = "*".
3222 // Synthesis grants permission to any target.
3223 use memstead_schema::workspace_config::CrossLinkValue;
3224 let tmp = TempDir::new().unwrap();
3225 let mut engine = build_demo_engine(&tmp);
3226 let mut settings = crate::workspace::WorkspaceSettings::default();
3227 settings
3228 .mem_create_rules
3229 .push(crate::workspace::CreateRuleSetting {
3230 pattern: "exec-*".to_string(),
3231 schemas: vec!["default".to_string()],
3232 default_cross_links: Some(CrossLinkValue::Wildcard),
3233 });
3234 engine.set_settings(settings);
3235 // No explicit policy; synthesis grants permission for any
3236 // target because the rule's value is Wildcard.
3237 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
3238 assert!(engine.cross_mem_link_allowed("exec-foo", "engine"));
3239 // Mem that doesn't match any rule → still denied.
3240 assert!(!engine.cross_mem_link_allowed("orphan", "specs"));
3241 }
3242
3243 /// #42: synthesis matches a hierarchical mem by composing the same
3244 /// `<mem_path>/<name>` candidate the create-rule glob is keyed on,
3245 /// not the bare leaf. Before the fix, `from_mem = "project"` could
3246 /// never match a `memstead/*` rule (the leaf-vs-composed-path
3247 /// divergence), so enforcement denied a link `memstead_overview`
3248 /// rendered as rule-granted.
3249 #[test]
3250 fn cross_mem_link_allowed_synthesises_for_hierarchical_mem() {
3251 use memstead_schema::workspace_config::CrossLinkValue;
3252 let tmp = TempDir::new().unwrap();
3253 let mem_dir = tmp.path().to_path_buf();
3254 // Mount `project` with a hierarchical branch so its `mem_path()`
3255 // is "memstead" and the composed candidate is "memstead/project".
3256 // The Folder backend handles loading; only the Mount's storage
3257 // feeds `mem_path()`.
3258 let mount = Mount {
3259 mem: "project".into(),
3260 schema: Some(pin("default")),
3261 storage: MountStorage::GitBranch {
3262 gitdir: mem_dir.join(".git"),
3263 branch: "memstead/project".into(),
3264 },
3265 capability: MountCapability::Write,
3266 lifecycle: MountLifecycle::Eager,
3267 cross_linkable: true,
3268 migration_target: None,
3269 };
3270 let writer = FilesystemMemWriter::new(mem_dir.clone());
3271 let mut engine =
3272 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
3273 let mut settings = crate::workspace::WorkspaceSettings::default();
3274 settings
3275 .mem_create_rules
3276 .push(crate::workspace::CreateRuleSetting {
3277 pattern: "memstead/*".to_string(),
3278 schemas: vec!["default".to_string()],
3279 default_cross_links: Some(CrossLinkValue::List(vec!["engine".to_string()])),
3280 });
3281 engine.set_settings(settings);
3282 assert!(
3283 engine.cross_mem_link_allowed("project", "engine"),
3284 "synthesis must match via the composed `memstead/project` candidate"
3285 );
3286 assert!(
3287 !engine.cross_mem_link_allowed("project", "macos"),
3288 "a target outside the rule's default_cross_links is still denied"
3289 );
3290 }
3291
3292 #[test]
3293 fn cross_mem_link_allowed_synthesises_from_matching_create_rule_list() {
3294 // Create rule's default_cross_links is a list — synthesis
3295 // grants permission to listed targets only.
3296 use memstead_schema::workspace_config::CrossLinkValue;
3297 let tmp = TempDir::new().unwrap();
3298 let mut engine = build_demo_engine(&tmp);
3299 let mut settings = crate::workspace::WorkspaceSettings::default();
3300 settings
3301 .mem_create_rules
3302 .push(crate::workspace::CreateRuleSetting {
3303 pattern: "exec-*".to_string(),
3304 schemas: vec!["default".to_string()],
3305 default_cross_links: Some(CrossLinkValue::List(vec!["specs".to_string()])),
3306 });
3307 engine.set_settings(settings);
3308 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
3309 // Target not in the synthesised list → denied.
3310 assert!(!engine.cross_mem_link_allowed("exec-foo", "engine"));
3311 }
3312
3313 #[test]
3314 fn cross_mem_link_allowed_explicit_policy_wins_over_synthesis() {
3315 // Explicit cross_mem_links wildcard fires first; the
3316 // synthesis layer is never consulted (and would deny).
3317 use memstead_schema::workspace_config::CrossLinkValue;
3318 let tmp = TempDir::new().unwrap();
3319 let mut engine = build_demo_engine(&tmp);
3320 let mut settings = crate::workspace::WorkspaceSettings::default();
3321 settings
3322 .cross_mem_links
3323 .insert("exec-foo".to_string(), CrossLinkValue::Wildcard);
3324 // The synthesis layer would deny `exec-foo → engine` (no
3325 // matching rule), but explicit policy returns true first.
3326 engine.set_settings(settings);
3327 assert!(engine.cross_mem_link_allowed("exec-foo", "engine"));
3328 }
3329
3330 #[test]
3331 fn cross_mem_link_allowed_synthesis_unions_into_explicit_list() {
3332 // Explicit list = ["specs"]; create rule synthesises = ["macos"].
3333 // Effective allowed targets: union ({specs, macos}).
3334 use memstead_schema::workspace_config::CrossLinkValue;
3335 let tmp = TempDir::new().unwrap();
3336 let mut engine = build_demo_engine(&tmp);
3337 let mut settings = crate::workspace::WorkspaceSettings::default();
3338 settings.cross_mem_links.insert(
3339 "exec-foo".to_string(),
3340 CrossLinkValue::List(vec!["specs".to_string()]),
3341 );
3342 settings
3343 .mem_create_rules
3344 .push(crate::workspace::CreateRuleSetting {
3345 pattern: "exec-*".to_string(),
3346 schemas: vec!["default".to_string()],
3347 default_cross_links: Some(CrossLinkValue::List(vec!["macos".to_string()])),
3348 });
3349 engine.set_settings(settings);
3350 // Explicit allowlist contains specs → allowed.
3351 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
3352 // Synthesis layer adds macos → allowed.
3353 assert!(engine.cross_mem_link_allowed("exec-foo", "macos"));
3354 // Neither layer allows engine → denied.
3355 assert!(!engine.cross_mem_link_allowed("exec-foo", "engine"));
3356 }
3357
3358 #[test]
3359 fn cross_mem_link_allowed_set_settings_invalidates_compiled_rule_cache() {
3360 // After set_settings, a fresh policy must be reflected on the
3361 // next call — the lazy memo can't return stale rules.
3362 use memstead_schema::workspace_config::CrossLinkValue;
3363 let tmp = TempDir::new().unwrap();
3364 let mut engine = build_demo_engine(&tmp);
3365
3366 // First settings: a rule allows exec-* → specs via synthesis.
3367 let mut s1 = crate::workspace::WorkspaceSettings::default();
3368 s1.mem_create_rules
3369 .push(crate::workspace::CreateRuleSetting {
3370 pattern: "exec-*".to_string(),
3371 schemas: vec!["default".to_string()],
3372 default_cross_links: Some(CrossLinkValue::List(vec!["specs".to_string()])),
3373 });
3374 engine.set_settings(s1);
3375 assert!(engine.cross_mem_link_allowed("exec-foo", "specs"));
3376
3377 // Replace settings: the rule no longer carries
3378 // default_cross_links. Cache must invalidate so the next
3379 // call sees the new policy.
3380 let mut s2 = crate::workspace::WorkspaceSettings::default();
3381 s2.mem_create_rules
3382 .push(crate::workspace::CreateRuleSetting {
3383 pattern: "exec-*".to_string(),
3384 schemas: vec!["default".to_string()],
3385 default_cross_links: None,
3386 });
3387 engine.set_settings(s2);
3388 assert!(!engine.cross_mem_link_allowed("exec-foo", "specs"));
3389 }
3390
3391 #[test]
3392 fn cross_mem_link_allowed_malformed_glob_falls_back_to_explicit_policy() {
3393 // Malformed pattern in a create rule causes CreateRuleSet
3394 // compilation to fail; the resolver logs and disables
3395 // synthesis, but explicit cross_mem_links still works.
3396 use memstead_schema::workspace_config::CrossLinkValue;
3397 let tmp = TempDir::new().unwrap();
3398 let mut engine = build_demo_engine(&tmp);
3399 let mut settings = crate::workspace::WorkspaceSettings::default();
3400 settings
3401 .mem_create_rules
3402 .push(crate::workspace::CreateRuleSetting {
3403 pattern: "[unclosed".to_string(),
3404 schemas: vec!["default".to_string()],
3405 default_cross_links: Some(CrossLinkValue::Wildcard),
3406 });
3407 // Explicit policy still works.
3408 settings
3409 .cross_mem_links
3410 .insert("specs".to_string(), CrossLinkValue::Wildcard);
3411 engine.set_settings(settings);
3412 // Explicit policy: specs → engine allowed.
3413 assert!(engine.cross_mem_link_allowed("specs", "engine"));
3414 // Synthesis disabled (compilation failed); rule's would-be
3415 // wildcard doesn't apply.
3416 assert!(!engine.cross_mem_link_allowed("orphan", "anything"));
3417 }
3418
3419 #[test]
3420 fn cross_mem_link_allowed_empty_list_denies_all_cross_mem_targets() {
3421 // [cross_mem_links] specs = [] is the explicit
3422 // "intentionally locked down" shape — same effect as
3423 // default-deny but operator-acknowledged.
3424 use memstead_schema::workspace_config::CrossLinkValue;
3425 let tmp = TempDir::new().unwrap();
3426 let mut engine = build_demo_engine(&tmp);
3427 let mut settings = crate::workspace::WorkspaceSettings::default();
3428 settings
3429 .cross_mem_links
3430 .insert("specs".to_string(), CrossLinkValue::List(Vec::new()));
3431 engine.set_settings(settings);
3432 // Same-mem still passes — policy only gates cross-mem.
3433 assert!(engine.cross_mem_link_allowed("specs", "specs"));
3434 // Cross-mem denied to every target.
3435 assert!(!engine.cross_mem_link_allowed("specs", "engine"));
3436 assert!(!engine.cross_mem_link_allowed("specs", "anything"));
3437 }
3438
3439 #[test]
3440 fn from_mounts_load_warnings_merge_into_health_summary() {
3441 let tmp = TempDir::new().unwrap();
3442 let mem_dir = tmp.path().to_path_buf();
3443 let body = "---\ntype: spec\n---\n# Dup2\n\n## Identity\n\na.\n\n## Identity\n\nb.\n";
3444 std::fs::write(mem_dir.join("dup2.md"), body).unwrap();
3445
3446 let writer = FilesystemMemWriter::new(mem_dir.clone());
3447 let engine = Engine::from_mounts(vec![(
3448 folder_mount("specs", mem_dir),
3449 Box::new(writer) as Box<dyn MemBackend>,
3450 )])
3451 .unwrap();
3452
3453 let summary = engine.health();
3454 assert!(
3455 summary
3456 .warnings
3457 .iter()
3458 .any(|w| matches!(w, WarningHint::DuplicateSectionHeading { .. })),
3459 "health() must merge load_warnings into summary.warnings: {:?}",
3460 summary.warnings,
3461 );
3462 }
3463
3464 #[test]
3465 fn workspace_root_accessor_is_none_for_engine_built_from_mounts() {
3466 let tmp = TempDir::new().unwrap();
3467 let mem_dir = tmp.path().to_path_buf();
3468 let writer = FilesystemMemWriter::new(mem_dir.clone());
3469 // Newest default generation so the clean-boot assertion below
3470 // isn't tripped by the SCHEMA_GENERATIONS_BEHIND hint.
3471 let mut mount = folder_mount("specs", mem_dir);
3472 mount.schema = Some("default@1.3.0".parse().unwrap());
3473 let engine =
3474 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
3475 assert!(
3476 engine.workspace_root().is_none(),
3477 "from_mounts has no workspace path",
3478 );
3479 // An entity-less mount reports itself empty; nothing else.
3480 assert!(
3481 engine
3482 .load_warnings()
3483 .iter()
3484 .all(|w| w.code() == "MOUNT_UNBACKED"),
3485 "{:?}",
3486 engine.load_warnings()
3487 );
3488 }
3489
3490 #[test]
3491 fn health_omits_outer_repo_warning_when_workspace_root_unset() {
3492 let tmp = TempDir::new().unwrap();
3493 let mem_dir = tmp.path().to_path_buf();
3494 let writer = FilesystemMemWriter::new(mem_dir.clone());
3495 let engine = Engine::from_mounts(vec![(
3496 folder_mount("specs", mem_dir),
3497 Box::new(writer) as Box<dyn MemBackend>,
3498 )])
3499 .unwrap();
3500 let health = engine.health();
3501 assert!(
3502 !health
3503 .warnings
3504 .iter()
3505 .any(|w| matches!(w, WarningHint::OuterRepoNotIgnoringMemRepo { .. })),
3506 "outer-repo check must skip when workspace_root is None",
3507 );
3508 }
3509
3510 #[test]
3511 fn writable_mem_names_filters_by_capability() {
3512 let tmp = TempDir::new().unwrap();
3513 let mem_dir = tmp.path().to_path_buf();
3514 let writer = FilesystemMemWriter::new(mem_dir.clone());
3515 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3516
3517 let engine = Engine::from_mounts(vec![
3518 (
3519 folder_mount("writable", mem_dir),
3520 Box::new(writer) as Box<dyn MemBackend>,
3521 ),
3522 (
3523 archive_mount("sealed", archive_path.clone()),
3524 Box::new(ArchiveBackend::new(archive_path)),
3525 ),
3526 ])
3527 .unwrap();
3528
3529 // Only the writable mount surfaces; the archive (read-only)
3530 // is filtered out.
3531 let names = engine.writable_mem_names();
3532 assert_eq!(names, vec!["writable"]);
3533 }
3534
3535 /// The default writable mem is
3536 /// the FIRST writable mount in declaration order — the stable seed,
3537 /// not the alphabetically-first name. `test` is declared first;
3538 /// `other` sorts ahead alphabetically but is declared second, so it
3539 /// is NOT the default. This is the invariant that stops a second
3540 /// mem from silently retargeting omitted-`mem` writes.
3541 #[test]
3542 fn default_writable_mem_is_declaration_first_not_alphabetical() {
3543 let tmp = TempDir::new().unwrap();
3544 let test_dir = tmp.path().join("test");
3545 let other_dir = tmp.path().join("other");
3546 std::fs::create_dir_all(&test_dir).unwrap();
3547 std::fs::create_dir_all(&other_dir).unwrap();
3548
3549 let engine = Engine::from_mounts(vec![
3550 (
3551 folder_mount("test", test_dir.clone()),
3552 Box::new(FilesystemMemWriter::new(test_dir)) as Box<dyn MemBackend>,
3553 ),
3554 (
3555 folder_mount("other", other_dir.clone()),
3556 Box::new(FilesystemMemWriter::new(other_dir)) as Box<dyn MemBackend>,
3557 ),
3558 ])
3559 .unwrap();
3560
3561 assert_eq!(
3562 engine.default_writable_mem(),
3563 Some("test"),
3564 "default must be the declaration-first writable mem, not the alphabetically-first",
3565 );
3566 }
3567
3568 /// Reverse declaration order to prove the default tracks declaration
3569 /// order rather than a fixed name: with `other` declared first it
3570 /// becomes the default. Together with the test above this pins the
3571 /// lean as mount order, not name sort.
3572 #[test]
3573 fn default_writable_mem_follows_declaration_order() {
3574 let tmp = TempDir::new().unwrap();
3575 let other_dir = tmp.path().join("other");
3576 let test_dir = tmp.path().join("test");
3577 std::fs::create_dir_all(&other_dir).unwrap();
3578 std::fs::create_dir_all(&test_dir).unwrap();
3579
3580 let engine = Engine::from_mounts(vec![
3581 (
3582 folder_mount("other", other_dir.clone()),
3583 Box::new(FilesystemMemWriter::new(other_dir)) as Box<dyn MemBackend>,
3584 ),
3585 (
3586 folder_mount("test", test_dir.clone()),
3587 Box::new(FilesystemMemWriter::new(test_dir)) as Box<dyn MemBackend>,
3588 ),
3589 ])
3590 .unwrap();
3591
3592 assert_eq!(engine.default_writable_mem(), Some("other"));
3593 }
3594
3595 /// A read-only-only workspace has no default writable mem.
3596 #[test]
3597 fn default_writable_mem_none_without_writable_mount() {
3598 let tmp = TempDir::new().unwrap();
3599 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3600 let engine = Engine::from_mounts(vec![(
3601 archive_mount("sealed", archive_path.clone()),
3602 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
3603 )])
3604 .unwrap();
3605 assert_eq!(engine.default_writable_mem(), None);
3606 }
3607
3608 #[test]
3609 fn folder_path_for_mem_returns_path_for_folder_mounts_only() {
3610 let tmp = TempDir::new().unwrap();
3611 let mem_dir = tmp.path().join("specs");
3612 std::fs::create_dir_all(&mem_dir).unwrap();
3613 let writer = FilesystemMemWriter::new(mem_dir.clone());
3614 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3615
3616 let engine = Engine::from_mounts(vec![
3617 (
3618 folder_mount("specs", mem_dir.clone()),
3619 Box::new(writer) as Box<dyn MemBackend>,
3620 ),
3621 (
3622 archive_mount("sealed", archive_path.clone()),
3623 Box::new(ArchiveBackend::new(archive_path)),
3624 ),
3625 ])
3626 .unwrap();
3627
3628 // Folder mount returns its path.
3629 assert_eq!(engine.folder_path_for_mem("specs"), Some(mem_dir.as_path()),);
3630 // Archive mount returns None — caller branches on storage type.
3631 assert_eq!(engine.folder_path_for_mem("sealed"), None);
3632 // Unknown mem returns None — same as Engine::mount.
3633 assert_eq!(engine.folder_path_for_mem("missing"), None);
3634 }
3635
3636 #[test]
3637 fn mount_accessor_returns_public_mount_shape() {
3638 // Build a heterogeneous engine and verify Engine::mount /
3639 // Engine::mounts surface the operator-facing Mount records.
3640 // Handlers branch on MountStorage variants through this
3641 // accessor (replacing full's gitdir_for / worktree_for /
3642 // mem_head_sha / mem_config_for direct-engine
3643 // accessors).
3644 let tmp = TempDir::new().unwrap();
3645 let mem_dir = tmp.path().to_path_buf();
3646 let writer = FilesystemMemWriter::new(mem_dir.clone());
3647 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3648
3649 let engine = Engine::from_mounts(vec![
3650 (
3651 folder_mount("writable", mem_dir.clone()),
3652 Box::new(writer) as Box<dyn MemBackend>,
3653 ),
3654 (
3655 archive_mount("sealed", archive_path.clone()),
3656 Box::new(ArchiveBackend::new(archive_path.clone())),
3657 ),
3658 ])
3659 .unwrap();
3660
3661 // Known mems: each returns a Mount whose storage variant
3662 // matches what the caller passed at construction.
3663 let folder = engine.mount("writable").expect("known mem");
3664 assert!(matches!(folder.storage, MountStorage::Folder { .. }));
3665 assert_eq!(folder.capability, MountCapability::Write);
3666
3667 let archive = engine.mount("sealed").expect("known mem");
3668 match &archive.storage {
3669 MountStorage::Archive { path } => assert_eq!(path, &archive_path),
3670 other => panic!("expected Archive storage, got {other:?}"),
3671 }
3672 assert_eq!(archive.capability, MountCapability::ReadOnly);
3673
3674 // Unknown mem — None, no panic, no error.
3675 assert!(engine.mount("missing").is_none());
3676
3677 // Engine::mounts enumerates every mount in declaration order.
3678 let mounts = engine.mounts();
3679 assert_eq!(mounts.len(), 2);
3680 assert_eq!(mounts[0].mem, "writable");
3681 assert_eq!(mounts[1].mem, "sealed");
3682 }
3683
3684 #[test]
3685 fn mem_router_writable_set_matches_writable_mount_capability() {
3686 // Build an engine with one writable folder mount and one
3687 // read-only archive mount; the router's writable set must
3688 // equal the writable mount's name only.
3689 let tmp = TempDir::new().unwrap();
3690 let mem_dir = tmp.path().join("specs");
3691 std::fs::create_dir_all(&mem_dir).unwrap();
3692 let writer = FilesystemMemWriter::new(mem_dir.clone());
3693 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3694
3695 let engine = Engine::from_mounts(vec![
3696 (
3697 folder_mount("specs", mem_dir.clone()),
3698 Box::new(writer) as Box<dyn MemBackend>,
3699 ),
3700 (
3701 archive_mount("ext", archive_path.clone()),
3702 Box::new(ArchiveBackend::new(archive_path)),
3703 ),
3704 ])
3705 .unwrap();
3706
3707 let router = engine.mem_router();
3708 assert!(router.is_writable("specs"));
3709 assert!(!router.is_writable("ext"));
3710 assert!(router.is_visible("specs"));
3711 assert!(router.is_visible("ext"));
3712 let writable: std::collections::HashSet<&String> = router.writable_mems().iter().collect();
3713 assert_eq!(writable.len(), 1);
3714 assert!(writable.contains(&"specs".to_string()));
3715 }
3716
3717 #[test]
3718 fn mem_router_origin_is_explicit_toml_for_workspace_mounts() {
3719 // Every mount built via `from_mounts` lands as
3720 // `MemOrigin::ExplicitToml` — the file-adapter origin.
3721 // `RuntimeCreated` is reserved for `memstead_mem_create`
3722 // runtime registrations once that handler migrates onto
3723 // the unified engine.
3724 let tmp = TempDir::new().unwrap();
3725 let mem_dir = tmp.path().join("specs");
3726 std::fs::create_dir_all(&mem_dir).unwrap();
3727 let writer = FilesystemMemWriter::new(mem_dir.clone());
3728
3729 let engine = Engine::from_mounts(vec![(
3730 folder_mount("specs", mem_dir),
3731 Box::new(writer) as Box<dyn MemBackend>,
3732 )])
3733 .unwrap();
3734
3735 let origin = engine
3736 .mem_router()
3737 .origin_for_mem("specs")
3738 .expect("known mem");
3739 assert_eq!(origin.kind(), "explicit");
3740 }
3741
3742 #[test]
3743 fn mem_router_dir_for_writable_folder_mount_matches_storage_path() {
3744 // Folder-backed writable mounts surface the storage path
3745 // via `dir_for_mem`. Handlers consuming the router for
3746 // per-mem path resolution rely on this.
3747 let tmp = TempDir::new().unwrap();
3748 let mem_dir = tmp.path().join("specs");
3749 std::fs::create_dir_all(&mem_dir).unwrap();
3750 let writer = FilesystemMemWriter::new(mem_dir.clone());
3751
3752 let engine = Engine::from_mounts(vec![(
3753 folder_mount("specs", mem_dir.clone()),
3754 Box::new(writer) as Box<dyn MemBackend>,
3755 )])
3756 .unwrap();
3757
3758 assert_eq!(
3759 engine.mem_router().dir_for_mem("specs"),
3760 Some(mem_dir.as_path()),
3761 );
3762 assert_eq!(engine.mem_router().dir_for_mem("unknown"), None);
3763 }
3764
3765 #[test]
3766 fn mem_router_archive_path_for_read_only_archive_mount() {
3767 // Read-only archive mounts register via `add_read_only` so
3768 // `archive_path_for_mem` resolves the archive's on-disk
3769 // location.
3770 let tmp = TempDir::new().unwrap();
3771 let mem_dir = tmp.path().join("specs");
3772 std::fs::create_dir_all(&mem_dir).unwrap();
3773 let writer = FilesystemMemWriter::new(mem_dir.clone());
3774 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3775
3776 let engine = Engine::from_mounts(vec![
3777 (
3778 folder_mount("specs", mem_dir),
3779 Box::new(writer) as Box<dyn MemBackend>,
3780 ),
3781 (
3782 archive_mount("ext", archive_path.clone()),
3783 Box::new(ArchiveBackend::new(archive_path.clone())),
3784 ),
3785 ])
3786 .unwrap();
3787
3788 let router = engine.mem_router();
3789 assert_eq!(
3790 router.archive_path_for_mem("ext"),
3791 Some(archive_path.as_path()),
3792 );
3793 // Writable folder mount has no archive path.
3794 assert_eq!(router.archive_path_for_mem("specs"), None);
3795 }
3796
3797 #[test]
3798 fn read_mem_config_via_backend_trait_folder_reads_bytes() {
3799 // Direct trait call against FilesystemMemWriter. Verifies
3800 // the backend-side primitive returns the raw bytes the
3801 // engine then parses.
3802 let tmp = TempDir::new().unwrap();
3803 let mem_dir = tmp.path().to_path_buf();
3804 std::fs::create_dir_all(mem_dir.join(".memstead")).unwrap();
3805 let body = br#"{
3806 "format": 1,
3807 "schema": "default@1.0.0",
3808 "writeGuidance": { "tone": "neutral" }
3809 }"#;
3810 std::fs::write(mem_dir.join(".memstead").join("config.json"), body).unwrap();
3811
3812 let writer = FilesystemMemWriter::new(mem_dir);
3813 let result = MemBackend::read_mem_config(&writer).unwrap();
3814 let bytes = result.expect("config bytes must surface");
3815 let parsed: serde_json::Value = serde_json::from_slice(&bytes).unwrap();
3816 assert_eq!(parsed["schema"], "default@1.0.0");
3817 }
3818
3819 #[test]
3820 fn read_mem_config_via_backend_trait_folder_missing_returns_none() {
3821 let tmp = TempDir::new().unwrap();
3822 let mem_dir = tmp.path().to_path_buf();
3823 let writer = FilesystemMemWriter::new(mem_dir);
3824 let result = MemBackend::read_mem_config(&writer).unwrap();
3825 assert!(result.is_none());
3826 }
3827
3828 #[test]
3829 fn read_mem_config_via_backend_trait_archive_reads_bytes() {
3830 // Build an archive containing .memstead/config.json and verify
3831 // the ArchiveBackend impl returns its bytes.
3832 let tmp = TempDir::new().unwrap();
3833 let archive_path = tmp.path().join("seed.mem");
3834 let body = br#"{
3835 "format": 1,
3836 "schema": "default@1.0.0",
3837 "writeGuidance": { "tone": "archive" }
3838 }"#;
3839 {
3840 let file = std::fs::File::create(&archive_path).unwrap();
3841 let mut writer = zip::ZipWriter::new(file);
3842 writer
3843 .start_file(
3844 ".memstead/config.json",
3845 zip::write::SimpleFileOptions::default(),
3846 )
3847 .unwrap();
3848 use std::io::Write;
3849 writer.write_all(body).unwrap();
3850 writer.finish().unwrap();
3851 }
3852
3853 let backend = ArchiveBackend::new(archive_path);
3854 let result = MemBackend::read_mem_config(&backend).unwrap();
3855 let bytes = result.expect("config bytes must surface");
3856 let parsed: serde_json::Value = serde_json::from_slice(&bytes).unwrap();
3857 assert_eq!(parsed["writeGuidance"]["tone"], "archive");
3858 }
3859
3860 #[test]
3861 fn mem_config_for_returns_none_when_no_config_file_present() {
3862 // Folder backend without a `.memstead/config.json` file. The
3863 // accessor must lenient — return None, not error.
3864 let tmp = TempDir::new().unwrap();
3865 let mem_dir = tmp.path().to_path_buf();
3866 let writer = FilesystemMemWriter::new(mem_dir.clone());
3867 let engine = Engine::from_mounts(vec![(
3868 folder_mount("specs", mem_dir),
3869 Box::new(writer) as Box<dyn MemBackend>,
3870 )])
3871 .unwrap();
3872 assert!(engine.mem_config_for("specs").is_none());
3873 }
3874
3875 #[test]
3876 fn mem_config_for_returns_some_when_config_file_present() {
3877 // Drop a valid `.memstead/config.json` into the mem dir,
3878 // build the engine, and assert the accessor surfaces a
3879 // MemConfig with the right shape (write_guidance entries
3880 // round-trip).
3881 let tmp = TempDir::new().unwrap();
3882 let mem_dir = tmp.path().to_path_buf();
3883 std::fs::create_dir_all(mem_dir.join(".memstead")).unwrap();
3884 let config_body = r#"{
3885 "format": 1,
3886 "schema": "default@1.0.0",
3887 "writeGuidance": {
3888 "tone": "neutral",
3889 "voice": "active"
3890 }
3891 }"#;
3892 std::fs::write(mem_dir.join(".memstead").join("config.json"), config_body).unwrap();
3893
3894 let writer = FilesystemMemWriter::new(mem_dir.clone());
3895 let engine = Engine::from_mounts(vec![(
3896 folder_mount("specs", mem_dir),
3897 Box::new(writer) as Box<dyn MemBackend>,
3898 )])
3899 .unwrap();
3900
3901 let cfg = engine
3902 .mem_config_for("specs")
3903 .expect("mem_config should load");
3904 assert_eq!(cfg.write_guidance.len(), 2);
3905 assert_eq!(
3906 cfg.write_guidance.get("tone").and_then(|v| v.as_str()),
3907 Some("neutral"),
3908 );
3909 assert_eq!(
3910 cfg.write_guidance.get("voice").and_then(|v| v.as_str()),
3911 Some("active"),
3912 );
3913 }
3914
3915 #[test]
3916 fn mem_config_for_unknown_mem_returns_none() {
3917 // Lenient accessor — unknown names get None, not Err.
3918 let tmp = TempDir::new().unwrap();
3919 let mem_dir = tmp.path().to_path_buf();
3920 let writer = FilesystemMemWriter::new(mem_dir.clone());
3921 let engine = Engine::from_mounts(vec![(
3922 folder_mount("specs", mem_dir),
3923 Box::new(writer) as Box<dyn MemBackend>,
3924 )])
3925 .unwrap();
3926 assert!(engine.mem_config_for("missing").is_none());
3927 }
3928
3929 #[test]
3930 fn mem_config_for_archive_mount_returns_none() {
3931 // Archive backends carry mem_config = None in V1 (the
3932 // read-from-storage path is deferred to a follow-up).
3933 let tmp = TempDir::new().unwrap();
3934 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
3935 let engine = Engine::from_mounts(vec![(
3936 archive_mount("ext", archive_path.clone()),
3937 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
3938 )])
3939 .unwrap();
3940 assert!(engine.mem_config_for("ext").is_none());
3941 }
3942
3943 #[test]
3944 fn mem_configs_named_iterates_only_mounts_with_config() {
3945 // Two folder mounts; one has a config file, one doesn't.
3946 // The iterator yields exactly the configured one — verifies
3947 // the filter_map shape and that the name comes from the
3948 // mount record (authoritative), not the config body.
3949 let tmp = TempDir::new().unwrap();
3950 let with_config = tmp.path().join("specs");
3951 let without_config = tmp.path().join("memos");
3952 std::fs::create_dir_all(with_config.join(".memstead")).unwrap();
3953 std::fs::create_dir_all(&without_config).unwrap();
3954 let config_body = r#"{
3955 "format": 1,
3956 "schema": "default@1.0.0",
3957 "writeGuidance": { "tone": "neutral" }
3958 }"#;
3959 std::fs::write(
3960 with_config.join(".memstead").join("config.json"),
3961 config_body,
3962 )
3963 .unwrap();
3964
3965 let engine = Engine::from_mounts(vec![
3966 (
3967 folder_mount("specs", with_config.clone()),
3968 Box::new(FilesystemMemWriter::new(with_config)) as Box<dyn MemBackend>,
3969 ),
3970 (
3971 folder_mount("memos", without_config.clone()),
3972 Box::new(FilesystemMemWriter::new(without_config)) as Box<dyn MemBackend>,
3973 ),
3974 ])
3975 .unwrap();
3976
3977 let yielded: Vec<(&str, usize)> = engine
3978 .mem_configs_named()
3979 .map(|(name, cfg)| (name, cfg.write_guidance.len()))
3980 .collect();
3981 assert_eq!(yielded, vec![("specs", 1)]);
3982 }
3983
3984 #[test]
3985 fn schema_for_returns_some_for_known_mem_and_none_for_unknown() {
3986 // Every mount registers a schema (resolved from its pin at
3987 // boot). Lookup by mem name surfaces the same Arc that
3988 // mutations resolve internally; unknown names return None.
3989 let tmp = TempDir::new().unwrap();
3990 let mem_dir = tmp.path().to_path_buf();
3991 let writer = FilesystemMemWriter::new(mem_dir.clone());
3992 let engine = Engine::from_mounts(vec![(
3993 folder_mount("specs", mem_dir),
3994 Box::new(writer) as Box<dyn MemBackend>,
3995 )])
3996 .unwrap();
3997 assert!(engine.schema_for("specs").is_some());
3998 assert!(engine.schema_for("missing").is_none());
3999 }
4000
4001 #[test]
4002 fn gitdir_for_unknown_mem_returns_unknown_mem() {
4003 let tmp = TempDir::new().unwrap();
4004 let mem_dir = tmp.path().to_path_buf();
4005 let writer = FilesystemMemWriter::new(mem_dir.clone());
4006 let engine = Engine::from_mounts(vec![(
4007 folder_mount("specs", mem_dir),
4008 Box::new(writer) as Box<dyn MemBackend>,
4009 )])
4010 .unwrap();
4011 let err = engine.gitdir_for("missing").unwrap_err();
4012 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4013 }
4014
4015 #[test]
4016 fn gitdir_for_folder_mount_returns_no_gitdir_error() {
4017 // Folder mounts do not have a gitdir — full's contract surfaces
4018 // a mem-level error, not UnknownMem. Mirror that here.
4019 let tmp = TempDir::new().unwrap();
4020 let mem_dir = tmp.path().to_path_buf();
4021 let writer = FilesystemMemWriter::new(mem_dir.clone());
4022 let engine = Engine::from_mounts(vec![(
4023 folder_mount("specs", mem_dir),
4024 Box::new(writer) as Box<dyn MemBackend>,
4025 )])
4026 .unwrap();
4027 let err = engine.gitdir_for("specs").unwrap_err();
4028 match err {
4029 EngineError::Mem(msg) => assert!(msg.contains("no resolved gitdir")),
4030 other => panic!("expected EngineError::Mem, got {other:?}"),
4031 }
4032 }
4033
4034 #[test]
4035 fn worktree_for_folder_mount_returns_storage_path() {
4036 let tmp = TempDir::new().unwrap();
4037 let mem_dir = tmp.path().to_path_buf();
4038 let writer = FilesystemMemWriter::new(mem_dir.clone());
4039 let engine = Engine::from_mounts(vec![(
4040 folder_mount("specs", mem_dir.clone()),
4041 Box::new(writer) as Box<dyn MemBackend>,
4042 )])
4043 .unwrap();
4044 let worktree = engine.worktree_for("specs").unwrap();
4045 assert_eq!(worktree, mem_dir);
4046 }
4047
4048 #[test]
4049 fn worktree_for_unknown_mem_returns_unknown_mem() {
4050 let tmp = TempDir::new().unwrap();
4051 let mem_dir = tmp.path().to_path_buf();
4052 let writer = FilesystemMemWriter::new(mem_dir.clone());
4053 let engine = Engine::from_mounts(vec![(
4054 folder_mount("specs", mem_dir),
4055 Box::new(writer) as Box<dyn MemBackend>,
4056 )])
4057 .unwrap();
4058 let err = engine.worktree_for("missing").unwrap_err();
4059 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4060 }
4061
4062 #[test]
4063 fn worktree_for_archive_mount_returns_archive_backed_error() {
4064 let tmp = TempDir::new().unwrap();
4065 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4066 let engine = Engine::from_mounts(vec![(
4067 archive_mount("ext", archive_path.clone()),
4068 Box::new(ArchiveBackend::new(archive_path)) as Box<dyn MemBackend>,
4069 )])
4070 .unwrap();
4071 let err = engine.worktree_for("ext").unwrap_err();
4072 match err {
4073 EngineError::Mem(msg) => assert!(msg.contains("archive-backed")),
4074 other => panic!("expected EngineError::Mem, got {other:?}"),
4075 }
4076 }
4077
4078 #[test]
4079 fn mem_head_sha_for_folder_mount_is_none() {
4080 // Folder backend doesn't track a head; current_head() returns
4081 // Ok(None) at construction; mem_head_sha returns Ok(None).
4082 let tmp = TempDir::new().unwrap();
4083 let mem_dir = tmp.path().to_path_buf();
4084 let writer = FilesystemMemWriter::new(mem_dir.clone());
4085 let engine = Engine::from_mounts(vec![(
4086 folder_mount("specs", mem_dir),
4087 Box::new(writer) as Box<dyn MemBackend>,
4088 )])
4089 .unwrap();
4090 let head = engine.mem_head_sha("specs").unwrap();
4091 assert_eq!(head, None);
4092 }
4093
4094 #[test]
4095 fn mem_head_sha_unknown_mem_returns_unknown_mem() {
4096 let tmp = TempDir::new().unwrap();
4097 let mem_dir = tmp.path().to_path_buf();
4098 let writer = FilesystemMemWriter::new(mem_dir.clone());
4099 let engine = Engine::from_mounts(vec![(
4100 folder_mount("specs", mem_dir),
4101 Box::new(writer) as Box<dyn MemBackend>,
4102 )])
4103 .unwrap();
4104 let err = engine.mem_head_sha("missing").unwrap_err();
4105 assert!(matches!(err, EngineError::UnknownMem(v) if v == "missing"));
4106 }
4107
4108 #[test]
4109 fn capability_surfaces_per_mount() {
4110 let tmp = TempDir::new().unwrap();
4111 let mem_dir = tmp.path().to_path_buf();
4112 let writer = FilesystemMemWriter::new(mem_dir.clone());
4113 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4114
4115 let engine = Engine::from_mounts(vec![
4116 (
4117 folder_mount("writable", mem_dir),
4118 Box::new(writer) as Box<dyn MemBackend>,
4119 ),
4120 (
4121 archive_mount("read-only", archive_path.clone()),
4122 Box::new(ArchiveBackend::new(archive_path)),
4123 ),
4124 ])
4125 .unwrap();
4126
4127 assert_eq!(
4128 engine.capability("writable").unwrap(),
4129 MountCapability::Write
4130 );
4131 assert_eq!(
4132 engine.capability("read-only").unwrap(),
4133 MountCapability::ReadOnly
4134 );
4135 assert!(matches!(
4136 engine.capability("missing"),
4137 Err(EngineError::UnknownMem(_))
4138 ));
4139 }
4140
4141 #[test]
4142 fn read_provenance_routes_through_backend() {
4143 let tmp = TempDir::new().unwrap();
4144 let mem_dir = tmp.path().to_path_buf();
4145 let writer = FilesystemMemWriter::new(mem_dir.clone());
4146
4147 // Append a provenance record via the backend trait directly,
4148 // then read it back through the engine.
4149 let backend_handle: &dyn MemBackend = &writer;
4150 backend_handle
4151 .append_provenance(&Provenance::new(
4152 std::time::UNIX_EPOCH + std::time::Duration::from_secs(1_700_000_000),
4153 crate::ProvenanceKind::Create,
4154 Some("v:e".into()),
4155 crate::vcs::Actor::Cli,
4156 None,
4157 Some("first".into()),
4158 ))
4159 .unwrap();
4160
4161 let engine = Engine::from_mounts(vec![(
4162 folder_mount("specs", mem_dir),
4163 Box::new(writer) as Box<dyn MemBackend>,
4164 )])
4165 .unwrap();
4166
4167 let records = engine.read_provenance("specs", None).unwrap();
4168 assert_eq!(records.len(), 1);
4169 assert_eq!(records[0].kind, crate::ProvenanceKind::Create);
4170 assert_eq!(records[0].entity.as_deref(), Some("v:e"));
4171 assert_eq!(records[0].note.as_deref(), Some("first"));
4172 }
4173
4174 #[test]
4175 fn archive_mount_returns_sealed_indirectly_through_backend_layer() {
4176 // The engine doesn't yet expose mutation methods, but an
4177 // archive backend held on a Mount with ReadOnly capability is
4178 // still a `&dyn MemBackend` whose write methods return
4179 // Sealed. This test locks the trait routing — when the engine
4180 // gains write methods in a later session, capability gating +
4181 // backend Sealed errors must agree.
4182 let tmp = TempDir::new().unwrap();
4183 let archive_path = build_archive(tmp.path(), "ext", &[("a.md", b"a")]);
4184 let backend = ArchiveBackend::new(archive_path);
4185 match MemBackend::write_entity(&backend, Path::new("x.md"), b"x") {
4186 Err(BackendError::Sealed) => {}
4187 other => panic!("expected Sealed, got {other:?}"),
4188 }
4189 }
4190
4191 // ---- Read-side delegates ----------------------------------------
4192 //
4193 // These tests pin the surface that the MCP migration consumes
4194 // (stats, health, context, communities, search, list, orphans,
4195 // stubs, most_connected, missing_required_outgoing). They run
4196 // against a folder-mount engine with a small fixture of created
4197 // entities and one relate edge — enough to exercise both the
4198 // graph-query path and the cache-invalidation hooks.
4199
4200 /// Generation-keyed memos (flywheel W8/01). Four claims: repeated
4201 /// reads serve the memo; a REFUSED engine batch leaves the memo
4202 /// untouched (the refusal path calls no hook — pinned so it stays
4203 /// cheap); a memo computed from an INTERIM (mid-batch) state never
4204 /// survives a rollback as fresh — the store-carried generation is
4205 /// what lets the invalidation hook adjudicate that correctly; and
4206 /// a real mutation invalidates, with the recomputation identical
4207 /// to a from-scratch detection (the criterion-3 identity oracle
4208 /// for the mechanism).
4209 #[test]
4210 fn generation_keyed_memos_survive_rollback_and_track_mutations() {
4211 use indexmap::IndexMap;
4212
4213 let tmp = TempDir::new().unwrap();
4214 let mut engine = build_demo_engine(&tmp);
4215
4216 // 1. Repeated reads: cell filled once, generation stable.
4217 let _ = engine.communities();
4218 let memo_gen_before = engine.community_memo.get().expect("memo filled").0;
4219 let _ = engine.communities();
4220 assert_eq!(
4221 engine.community_memo.get().expect("still filled").0,
4222 memo_gen_before,
4223 "repeated reads must serve the memo"
4224 );
4225
4226 // 2. A REFUSED engine batch rolls the store back and leaves
4227 // the memo untouched — refusal stays recompute-free.
4228 let bare = |id: crate::EntityId| crate::engine::UpdateEntityArgs {
4229 anchors: Vec::new(),
4230 anchors_unset: Vec::new(),
4231 id,
4232 expected_hash: None,
4233 sections: IndexMap::new(),
4234 append_sections: IndexMap::new(),
4235 patch_sections: IndexMap::new(),
4236 metadata: IndexMap::new(),
4237 metadata_unset: Vec::new(),
4238 declare_relations: Vec::new(),
4239 dry_run: false,
4240 relations_unset: Vec::new(),
4241 };
4242 let mut real = bare(crate::EntityId::new("specs", "source-one"));
4243 real.append_sections
4244 .insert("identity".to_string(), "appended line".to_string());
4245 let missing = bare(crate::EntityId::new("specs", "does-not-exist"));
4246 let (actor, client) = cli_actor();
4247 let result = engine
4248 .batch_update(
4249 vec![(real, None), (missing, None)],
4250 actor,
4251 Some(&client),
4252 false,
4253 )
4254 .expect("refused batch returns a report-all envelope");
4255 assert!(
4256 !result.applied,
4257 "the missing target refuses the whole batch"
4258 );
4259 assert_eq!(
4260 engine.community_memo.get().map(|m| m.0),
4261 Some(memo_gen_before),
4262 "a refused batch must leave the pre-batch memo standing"
4263 );
4264 assert_eq!(
4265 engine.store().generation(),
4266 memo_gen_before.store_generation,
4267 "rollback restored the store to the memo's generation"
4268 );
4269
4270 // 3. The dangerous direction: a memo computed from an INTERIM
4271 // state (simulated batch staging) must not survive the
4272 // rollback as fresh. The store-carried generation is what the
4273 // invalidation hook adjudicates with.
4274 let snapshot = engine.store.clone();
4275 let interim_id = crate::EntityId::new("specs", "interim-only");
4276 let mut interim = engine
4277 .store
4278 .get(&crate::EntityId::new("specs", "source-one"))
4279 .expect("demo entity present")
4280 .clone();
4281 interim.id = interim_id.clone();
4282 interim.title = "Interim Only".to_string();
4283 engine.store.upsert(interim_id, interim);
4284 engine.invalidate_communities();
4285 let _ = engine.communities();
4286 let interim_gen = engine.community_memo.get().expect("interim memo").0;
4287 assert_ne!(interim_gen, memo_gen_before);
4288 engine.store = snapshot; // rollback, generation restored with it
4289 engine.invalidate_communities();
4290 engine.invalidate_search_indexes();
4291 if let Some((g, _)) = engine.community_memo.get() {
4292 assert_ne!(
4293 *g, interim_gen,
4294 "a rolled-back interim state must never be served as fresh"
4295 );
4296 }
4297 assert!(
4298 !engine
4299 .communities()
4300 .entity_cluster_map
4301 .keys()
4302 .any(|id| id.contains("interim-only")),
4303 "the partition served after rollback reflects the restored store, not the interim one"
4304 );
4305
4306 // 4. A real mutation invalidates; the recomputation equals a
4307 // from-scratch detection and sees the new entity.
4308 let (actor, client) = cli_actor();
4309 engine
4310 .create_entity(
4311 empty_create_args("specs", "Fourth Entity"),
4312 actor,
4313 Some(&client),
4314 None,
4315 )
4316 .unwrap();
4317 assert!(
4318 engine
4319 .communities()
4320 .entity_cluster_map
4321 .keys()
4322 .any(|id| id.contains("fourth-entity")),
4323 "recomputed partition sees the new entity"
4324 );
4325 let fresh = {
4326 let schema = engine
4327 .schemas
4328 .iter()
4329 .min_by(|a, b| a.0.cmp(b.0))
4330 .map(|(_, s)| s.clone())
4331 .expect("demo engine has a schema");
4332 let schema_for_weights = schema.clone();
4333 crate::graph::community::detect_communities(
4334 engine.store(),
4335 schema.manifest.community.resolution,
4336 schema.manifest.community.seed,
4337 move |rel_type| {
4338 schema_for_weights
4339 .manifest
4340 .relationships
4341 .definitions
4342 .iter()
4343 .find(|d| d.name == rel_type)
4344 .map(|d| d.default_weight as f64)
4345 .unwrap_or(1.0)
4346 },
4347 )
4348 };
4349 assert_eq!(
4350 engine.communities().entity_cluster_map,
4351 fresh.entity_cluster_map,
4352 "memo must equal a from-scratch detection over the current store"
4353 );
4354 }
4355
4356 fn build_demo_engine(tmp: &TempDir) -> Engine {
4357 let mem_dir = tmp.path().to_path_buf();
4358 let writer = FilesystemMemWriter::new(mem_dir.clone());
4359 let mut engine = Engine::from_mounts(vec![(
4360 folder_mount("specs", mem_dir),
4361 Box::new(writer) as Box<dyn MemBackend>,
4362 )])
4363 .unwrap();
4364 let (actor, client) = cli_actor();
4365 let source = engine
4366 .create_entity(
4367 empty_create_args("specs", "Source One"),
4368 actor,
4369 Some(&client),
4370 None,
4371 )
4372 .unwrap();
4373 let target = engine
4374 .create_entity(
4375 empty_create_args("specs", "Target Two"),
4376 actor,
4377 Some(&client),
4378 None,
4379 )
4380 .unwrap();
4381 engine
4382 .create_entity(
4383 empty_create_args("specs", "Lonely Three"),
4384 actor,
4385 Some(&client),
4386 None,
4387 )
4388 .unwrap();
4389 engine
4390 .relate_entity(
4391 RelateEntityArgs {
4392 source: source.id.clone(),
4393 expected_hash: Some(source.content_hash.clone()),
4394 rel_type: "USES".to_string(),
4395 target: target.id.clone(),
4396 remove: false,
4397 description: None,
4398 dry_run: false,
4399 },
4400 actor,
4401 Some(&client),
4402 None,
4403 )
4404 .unwrap();
4405 engine
4406 }
4407
4408 #[test]
4409 fn status_reports_per_engine_counts() {
4410 let tmp = TempDir::new().unwrap();
4411 let engine = build_demo_engine(&tmp);
4412 let stats = engine.status();
4413 assert_eq!(stats.entity_count, 3);
4414 assert_eq!(stats.edge_count, 1);
4415 assert_eq!(stats.mem_count, 1);
4416 assert_eq!(stats.types_in_use, vec!["spec".to_string()]);
4417 assert_eq!(stats.edge_types.get("USES"), Some(&1));
4418 }
4419
4420 #[test]
4421 fn orphans_lists_unconnected_real_entities() {
4422 let tmp = TempDir::new().unwrap();
4423 let engine = build_demo_engine(&tmp);
4424 let orphans = engine.orphans();
4425 assert_eq!(orphans.len(), 1);
4426 assert_eq!(orphans[0].as_ref(), "specs--lonely-three");
4427 }
4428
4429 /// #49: the orphan/community headlines can be attributed per pinned
4430 /// schema. Single-mem here, so one bucket — but it proves the
4431 /// attribution keys by `schema_of(mem)` and that the per-schema
4432 /// counts sum to the raw total (which a health surface keeps verbatim).
4433 #[test]
4434 fn schema_breakdowns_attribute_to_mem_pin() {
4435 let tmp = TempDir::new().unwrap();
4436 let engine = build_demo_engine(&tmp);
4437
4438 let orphans = engine.orphans();
4439 let orphans_by_schema = engine.orphans_by_schema(&orphans);
4440 assert_eq!(
4441 orphans_by_schema.values().sum::<usize>(),
4442 orphans.len(),
4443 "per-schema orphan counts must sum to the raw total"
4444 );
4445 assert_eq!(orphans_by_schema.len(), 1, "one mem ⇒ one schema bucket");
4446 let (schema, count) = orphans_by_schema.iter().next().unwrap();
4447 assert!(!schema.is_empty(), "specs mem is pinned: {schema:?}");
4448 assert_eq!(*count, 1);
4449
4450 // communities_by_schema buckets the demo mem's clusters under the
4451 // same pin; with one schema, its values sum to the global count.
4452 let mems: Vec<String> = engine.mounts().iter().map(|m| m.mem.clone()).collect();
4453 let communities_by_schema = engine.communities_by_schema(&mems);
4454 assert_eq!(communities_by_schema.len(), 1);
4455 assert_eq!(
4456 communities_by_schema.values().sum::<usize>(),
4457 engine.communities().count,
4458 );
4459 }
4460
4461 #[test]
4462 fn stubs_lists_unresolved_link_targets() {
4463 let tmp = TempDir::new().unwrap();
4464 let mem_dir = tmp.path().to_path_buf();
4465 let writer = FilesystemMemWriter::new(mem_dir.clone());
4466 let mut engine = Engine::from_mounts(vec![(
4467 folder_mount("specs", mem_dir),
4468 Box::new(writer) as Box<dyn MemBackend>,
4469 )])
4470 .unwrap();
4471 let (actor, client) = cli_actor();
4472 let source = engine
4473 .create_entity(
4474 empty_create_args("specs", "Holder"),
4475 actor,
4476 Some(&client),
4477 None,
4478 )
4479 .unwrap();
4480 // Relate to a non-existent target — relate_entity creates a
4481 // stub for the target so the edge can land.
4482 engine
4483 .relate_entity(
4484 RelateEntityArgs {
4485 source: source.id.clone(),
4486 expected_hash: Some(source.content_hash.clone()),
4487 rel_type: "USES".to_string(),
4488 target: EntityId::new("specs", "ghost"),
4489 remove: false,
4490 description: None,
4491 dry_run: false,
4492 },
4493 actor,
4494 Some(&client),
4495 None,
4496 )
4497 .unwrap();
4498 let stubs = engine.stubs();
4499 assert!(
4500 stubs.iter().any(|(id, _)| id.as_ref() == "specs--ghost"),
4501 "expected ghost stub: {stubs:?}"
4502 );
4503 }
4504
4505 #[test]
4506 fn most_connected_orders_by_degree() {
4507 let tmp = TempDir::new().unwrap();
4508 let engine = build_demo_engine(&tmp);
4509 let top = engine.most_connected(5);
4510 assert_eq!(top.len(), 3);
4511 // Source and Target each have one edge; Lonely has zero.
4512 let zero_degree: Vec<_> = top
4513 .iter()
4514 .filter(|c| c.total == 0)
4515 .map(|c| c.id.as_ref().to_string())
4516 .collect();
4517 assert_eq!(zero_degree, vec!["specs--lonely-three".to_string()]);
4518 }
4519
4520 #[test]
4521 fn health_returns_per_engine_summary() {
4522 let tmp = TempDir::new().unwrap();
4523 let engine = build_demo_engine(&tmp);
4524 let health = engine.health();
4525 // `memstead_create` refuses on missing required sections, so
4526 // entities built through `empty_create_args` carry the
4527 // helper-seeded `identity` + `purpose` bodies and no longer
4528 // surface as missing-fields. Health remains the read-side
4529 // tolerance surface for legacy on-disk drift — covered by
4530 // the loader-tolerance tests that hand-craft pre-strict
4531 // markdown files.
4532 assert!(
4533 health
4534 .missing_fields
4535 .iter()
4536 .all(|r| r.id.as_ref() != "specs--source-one"),
4537 "post-strict-create fixture must not surface as missing-fields; got {:?}",
4538 health.missing_fields,
4539 );
4540 }
4541
4542 #[test]
4543 fn context_carries_neighbors_and_community() {
4544 let tmp = TempDir::new().unwrap();
4545 let engine = build_demo_engine(&tmp);
4546 let source_id = EntityId::new("specs", "source-one");
4547 let ctx = engine.context(&source_id).unwrap();
4548 assert_eq!(ctx.entity_id, source_id);
4549 assert_eq!(ctx.neighbors.len(), 1);
4550 assert_eq!(ctx.neighbors[0].relationship, "USES");
4551 assert!(matches!(ctx.neighbors[0].direction, Direction::Outgoing));
4552 }
4553
4554 #[test]
4555 fn communities_caches_louvain_until_invalidated() {
4556 let tmp = TempDir::new().unwrap();
4557 let mut engine = build_demo_engine(&tmp);
4558 // Population reflects the current store at first call.
4559 let entities_before = engine.communities().entity_cluster_map.len();
4560 // Cache hit — repeat call returns same data.
4561 assert_eq!(
4562 engine.communities().entity_cluster_map.len(),
4563 entities_before
4564 );
4565 // Mutation invalidates the cache; next call re-runs against
4566 // the post-mutation store and includes the new entity.
4567 let (actor, client) = cli_actor();
4568 engine
4569 .create_entity(
4570 empty_create_args("specs", "Disturber"),
4571 actor,
4572 Some(&client),
4573 None,
4574 )
4575 .unwrap();
4576 let entities_after = engine.communities().entity_cluster_map.len();
4577 assert_eq!(
4578 entities_after,
4579 entities_before + 1,
4580 "create_entity should have invalidated community cache and added the new entity"
4581 );
4582 }
4583
4584 #[test]
4585 fn list_filters_by_metadata_only() {
4586 let tmp = TempDir::new().unwrap();
4587 let engine = build_demo_engine(&tmp);
4588 let scope = SearchScope {
4589 entity_type: Some("spec".to_string()),
4590 ..Default::default()
4591 };
4592 let result = engine.list(&scope);
4593 // Three real spec entities created; stubs / non-spec types absent.
4594 assert_eq!(result.hits.len(), 3);
4595 }
4596
4597 #[test]
4598 fn list_applies_schema_declared_filter_on_non_default_schema_mem() {
4599 // A mem pinned to `planning` (non-default schema). The
4600 // `decision` type declares `status` with `filterable: equality`.
4601 // Pre-fix, filter dispatch consulted only the built-in default
4602 // schema via `type_by_name`, missed `status`, silently bypassed
4603 // the filter, and emitted the misleading "unknown filter key"
4604 // warning. Post-fix, the filter is honored and no warning fires.
4605 let tmp = TempDir::new().unwrap();
4606 let mem_dir = tmp.path().to_path_buf();
4607 let writer = FilesystemMemWriter::new(mem_dir.clone());
4608 let mount = Mount {
4609 mem: "planning".to_string(),
4610 schema: Some(memstead_schema::SchemaRef::new(
4611 "planning",
4612 semver::Version::new(0, 1, 0),
4613 )),
4614 storage: MountStorage::Folder { path: mem_dir },
4615 capability: MountCapability::Write,
4616 lifecycle: MountLifecycle::Eager,
4617 cross_linkable: true,
4618 migration_target: None,
4619 };
4620 let mut engine =
4621 Engine::from_mounts(vec![(mount, Box::new(writer) as Box<dyn MemBackend>)]).unwrap();
4622 let (actor, client) = cli_actor();
4623
4624 // Two decisions with different status values; required fields
4625 // (decision/context/consequences sections, decided_on, deciders)
4626 // get placeholder defaults — the test only cares about the
4627 // status field's filterability.
4628 for (title, status) in &[("Skip Postgres", "accepted"), ("Use SQLite", "proposed")] {
4629 let mut metadata = indexmap::IndexMap::new();
4630 metadata.insert("status".to_string(), status.to_string());
4631 metadata.insert("deciders".to_string(), "alice".to_string());
4632 metadata.insert("decided_on".to_string(), "2026-05-19".to_string());
4633 let args = crate::engine::CreateEntityArgs {
4634 anchors: Vec::new(),
4635 mem: "planning".to_string(),
4636 title: title.to_string(),
4637 entity_type: "decision".to_string(),
4638 sections: indexmap::IndexMap::from_iter([
4639 ("decision".to_string(), "We chose this.".to_string()),
4640 ("context".to_string(), "Single-user dev.".to_string()),
4641 ("consequences".to_string(), "Lose multi-writer.".to_string()),
4642 ]),
4643 metadata,
4644 relations: Vec::new(),
4645 dry_run: false,
4646 };
4647 engine
4648 .create_entity(args, actor, Some(&client), None)
4649 .unwrap();
4650 }
4651
4652 // Filter on the schema-declared filterable field.
4653 let scope = SearchScope {
4654 entity_type: Some("decision".to_string()),
4655 filters: std::collections::HashMap::from([(
4656 "status".to_string(),
4657 "accepted".to_string(),
4658 )]),
4659 ..Default::default()
4660 };
4661 let result = engine.list(&scope);
4662 assert_eq!(
4663 result.hits.len(),
4664 1,
4665 "filter on schema-declared field must select only matching entities"
4666 );
4667 assert_eq!(result.hits[0].title, "Skip Postgres");
4668 assert!(
4669 result.warnings.is_empty(),
4670 "no warning should fire when the filter is declared by the mem's pinned schema: {:?}",
4671 result.warnings
4672 );
4673 }
4674
4675 #[test]
4676 fn search_returns_results_against_built_index() {
4677 let tmp = TempDir::new().unwrap();
4678 let engine = build_demo_engine(&tmp);
4679 let scope = SearchScope {
4680 query: Some(crate::ops::Query {
4681 any: vec!["source".to_string()],
4682 ..Default::default()
4683 }),
4684 ..Default::default()
4685 };
4686 let result = engine.search(&scope).expect("native search returns Ok");
4687 assert!(result.total >= 1, "expected ≥1 hit for source: {result:?}");
4688 assert!(
4689 result
4690 .hits
4691 .iter()
4692 .any(|h| h.id.as_ref() == "specs--source-one"),
4693 "expected source-one in hits: {result:?}"
4694 );
4695 }
4696
4697 // ---- Engine::from_workspace_root (lean boot path) --------------
4698}