rigg_core/
identity.rs

1//! Identity-graph extraction: which service-to-service RBAC edges does this
2//! workspace's configuration require? (spec §8.2)
3//!
4//! Edges are derived from the actual resource files — data source connection
5//! strings, knowledge-base model wiring, agent→knowledge-base grounding, and
6//! encryption key references. `rigg auth doctor` verifies and repairs them.
7
8use serde_json::Value;
9
10use crate::registry;
11use crate::resources::traits::ResourceKind;
12use crate::store::Store;
13use crate::workspace::Workspace;
14
15/// Built-in Azure role definition IDs.
16pub mod roles {
17    pub const STORAGE_BLOB_DATA_READER: (&str, &str) = (
18        "ba92f5b4-2d11-453d-a403-e96b0029c9fe",
19        "Storage Blob Data Reader",
20    );
21    pub const COGNITIVE_SERVICES_OPENAI_USER: (&str, &str) = (
22        "5e0bd9bd-7b93-4f28-af87-19fc36ad61bd",
23        "Cognitive Services OpenAI User",
24    );
25    pub const COGNITIVE_SERVICES_USER: (&str, &str) = (
26        "a97b65f3-24c7-4388-baec-2e87135dc908",
27        "Cognitive Services User",
28    );
29    pub const SEARCH_INDEX_DATA_READER: (&str, &str) = (
30        "1407120a-92aa-4202-b7e9-c0e197c71c8f",
31        "Search Index Data Reader",
32    );
33    pub const KEY_VAULT_SECRETS_USER: (&str, &str) = (
34        "4633458b-17de-408a-b874-0445c86b69e6",
35        "Key Vault Secrets User",
36    );
37}
38
39/// Whose identity must hold the role.
40#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize)]
41#[serde(rename_all = "kebab-case")]
42pub enum Principal {
43    SearchService,
44    FoundryProject,
45}
46
47/// How the edge can be verified.
48#[derive(Debug, Clone, Copy, PartialEq, Eq, serde::Serialize)]
49#[serde(rename_all = "kebab-case")]
50pub enum EdgeKind {
51    /// ARM RBAC role assignment — verifiable and fixable.
52    Rbac,
53    /// Data-plane permission model outside ARM RBAC (Cosmos SQL roles,
54    /// Azure SQL contained users) — reported with guidance only.
55    Informational,
56}
57
58#[derive(Debug, Clone, serde::Serialize)]
59pub struct IdentityEdge {
60    pub principal: Principal,
61    /// ARM resource id of the target scope, when derivable from config.
62    pub scope: Option<String>,
63    pub target: String,
64    pub role_id: String,
65    pub role_name: String,
66    pub kind: EdgeKind,
67    pub reason: String,
68}
69
70impl IdentityEdge {
71    fn rbac(
72        principal: Principal,
73        scope: Option<String>,
74        target: impl Into<String>,
75        role: (&str, &str),
76        reason: impl Into<String>,
77    ) -> Self {
78        IdentityEdge {
79            principal,
80            scope,
81            target: target.into(),
82            role_id: role.0.to_string(),
83            role_name: role.1.to_string(),
84            kind: EdgeKind::Rbac,
85            reason: reason.into(),
86        }
87    }
88}
89
90/// Extract every identity edge the workspace's resources require, for one
91/// environment's tree.
92pub fn identity_edges(ws: &Workspace, env: &str) -> Vec<IdentityEdge> {
93    let mut edges: Vec<IdentityEdge> = Vec::new();
94    for project in &ws.projects {
95        let store = Store::new(project, env);
96        let Ok(files) = store.list() else { continue };
97        for (r, _) in files {
98            let Ok(value) = store.read(&r) else { continue };
99            match r.kind {
100                ResourceKind::DataSource => datasource_edges(&r.name, &value, &mut edges),
101                ResourceKind::KnowledgeBase
102                    if value
103                        .get("models")
104                        .is_some_and(|m| !m.as_array().is_none_or(|a| a.is_empty()))
105                        || value.get("answerInstructions").is_some() =>
106                {
107                    edges.push(IdentityEdge::rbac(
108                        Principal::SearchService,
109                        None,
110                        "Foundry account (model access)",
111                        roles::COGNITIVE_SERVICES_USER,
112                        format!(
113                            "knowledge base '{}' uses a model for retrieval/synthesis",
114                            r.name
115                        ),
116                    ));
117                }
118                ResourceKind::Skillset => skillset_edges(&r.name, &value, &mut edges),
119                ResourceKind::Index => {
120                    let has_vectorizer = value
121                        .get("vectorSearch")
122                        .and_then(|v| v.get("vectorizers"))
123                        .and_then(Value::as_array)
124                        .is_some_and(|a| !a.is_empty());
125                    if has_vectorizer {
126                        edges.push(IdentityEdge::rbac(
127                            Principal::SearchService,
128                            None,
129                            "Foundry account (model access)",
130                            roles::COGNITIVE_SERVICES_OPENAI_USER,
131                            format!("index '{}' has vectorizers", r.name),
132                        ));
133                    }
134                }
135                ResourceKind::Agent => {
136                    for (kind, name) in registry::extract_references(r.kind, &value) {
137                        if kind == ResourceKind::KnowledgeBase {
138                            edges.push(IdentityEdge::rbac(
139                                Principal::FoundryProject,
140                                None,
141                                "Search service (knowledge base retrieval)",
142                                roles::SEARCH_INDEX_DATA_READER,
143                                format!("agent '{}' grounds on knowledge base '{name}'", r.name),
144                            ));
145                        }
146                    }
147                }
148                _ => {}
149            }
150            // encryption keys → Key Vault
151            if let Some(uri) = value
152                .get("encryptionKey")
153                .and_then(|k| k.get("keyVaultUri"))
154                .and_then(Value::as_str)
155            {
156                edges.push(IdentityEdge::rbac(
157                    Principal::SearchService,
158                    None,
159                    format!("Key Vault {uri}"),
160                    roles::KEY_VAULT_SECRETS_USER,
161                    format!(
162                        "{} '{}' uses customer-managed encryption",
163                        r.kind.display_name(),
164                        r.name
165                    ),
166                ));
167            }
168        }
169    }
170    dedup(edges)
171}
172
173/// Skillset edges: Azure OpenAI skills need model access on the Foundry
174/// account; an identity-based `cognitiveServices` connection
175/// (AIServicesByIdentity) needs 'Cognitive Services User' on that specific
176/// AI services account (named by its subdomain URL).
177fn skillset_edges(name: &str, value: &Value, edges: &mut Vec<IdentityEdge>) {
178    if value.to_string().contains("AzureOpenAI") {
179        edges.push(IdentityEdge::rbac(
180            Principal::SearchService,
181            None,
182            "Foundry account (model access)",
183            roles::COGNITIVE_SERVICES_OPENAI_USER,
184            format!("skillset '{name}' calls Azure OpenAI"),
185        ));
186    }
187    // Custom Web API skills: AAD auth via `authResourceId` means the search
188    // identity must be authorized on the target app — app-level, not ARM
189    // RBAC, so informational.
190    if let Some(skills) = value.get("skills").and_then(Value::as_array) {
191        for skill in skills {
192            let is_webapi = skill
193                .get("@odata.type")
194                .and_then(Value::as_str)
195                .is_some_and(|t| t.ends_with("WebApiSkill"));
196            if !is_webapi {
197                continue;
198            }
199            let uri = skill.get("uri").and_then(Value::as_str).unwrap_or("?");
200            if let Some(auth) = skill.get("authResourceId").and_then(Value::as_str) {
201                if !auth.trim().is_empty() {
202                    edges.push(IdentityEdge {
203                        principal: Principal::SearchService,
204                        scope: None,
205                        target: format!("custom Web API {uri}"),
206                        role_id: String::new(),
207                        role_name: "app authorization (AAD)".into(),
208                        kind: EdgeKind::Informational,
209                        reason: format!(
210                            "skillset '{name}' calls a custom Web API with AAD auth \
211                             (authResourceId: {auth}) — the search identity must be \
212                             assigned/permitted on that app registration"
213                        ),
214                    });
215                }
216            } else {
217                edges.push(IdentityEdge {
218                    principal: Principal::SearchService,
219                    scope: None,
220                    target: format!("custom Web API {uri}"),
221                    role_id: String::new(),
222                    role_name: "endpoint authorization".into(),
223                    kind: EdgeKind::Informational,
224                    reason: format!(
225                        "skillset '{name}' calls a custom Web API without AAD auth \
226                         (no authResourceId) — the endpoint must accept the call \
227                         (function key in httpHeaders, or make it identity-based \
228                         with authResourceId)"
229                    ),
230                });
231            }
232        }
233    }
234    let cs = value.get("cognitiveServices");
235    let identity_based = cs
236        .and_then(|c| c.get("@odata.type"))
237        .and_then(Value::as_str)
238        .is_some_and(|t| t.ends_with("AIServicesByIdentity"));
239    if identity_based {
240        if let Some(account) = cs
241            .and_then(|c| c.get("subdomainUrl"))
242            .and_then(Value::as_str)
243            .and_then(ai_services_account_from_subdomain)
244        {
245            edges.push(IdentityEdge::rbac(
246                Principal::SearchService,
247                None,
248                format!("AI services account '{account}'"),
249                roles::COGNITIVE_SERVICES_USER,
250                format!("skillset '{name}' uses identity-based AI services enrichment"),
251            ));
252        }
253    }
254}
255
256/// `https://<name>.cognitiveservices.azure.com/` → `<name>`.
257fn ai_services_account_from_subdomain(url: &str) -> Option<String> {
258    url.strip_prefix("https://")
259        .and_then(|rest| rest.split('.').next())
260        .filter(|s| !s.is_empty())
261        .map(str::to_string)
262}
263
264fn datasource_edges(name: &str, value: &Value, edges: &mut Vec<IdentityEdge>) {
265    let ds_type = value
266        .get("type")
267        .and_then(Value::as_str)
268        .unwrap_or_default();
269    let conn = value
270        .get("credentials")
271        .and_then(|c| c.get("connectionString"))
272        .and_then(Value::as_str)
273        .unwrap_or_default();
274    let scope = parse_resource_id(conn);
275    match ds_type {
276        "azureblob" | "adlsgen2" | "azurefile" | "azurefiles" | "azuretable" => {
277            edges.push(IdentityEdge::rbac(
278                Principal::SearchService,
279                scope.clone(),
280                scope
281                    .clone()
282                    .unwrap_or_else(|| "storage account (set ResourceId= in the connection string)".into()),
283                roles::STORAGE_BLOB_DATA_READER,
284                format!("data source '{name}' ({ds_type}) reads from storage"),
285            ));
286        }
287        "cosmosdb" => edges.push(IdentityEdge {
288            principal: Principal::SearchService,
289            scope: scope.clone(),
290            target: scope.unwrap_or_else(|| "Cosmos DB account".into()),
291            role_id: String::new(),
292            role_name: "Cosmos DB Built-in Data Reader (SQL role)".into(),
293            kind: EdgeKind::Informational,
294            reason: format!(
295                "data source '{name}' reads Cosmos DB — grant via `az cosmosdb sql role assignment create` \
296                 (Cosmos data-plane roles are not ARM RBAC)"
297            ),
298        }),
299        "azuresql" => edges.push(IdentityEdge {
300            principal: Principal::SearchService,
301            scope: scope.clone(),
302            target: scope.unwrap_or_else(|| "Azure SQL database".into()),
303            role_id: String::new(),
304            role_name: "db_datareader (contained AAD user)".into(),
305            kind: EdgeKind::Informational,
306            reason: format!(
307                "data source '{name}' reads Azure SQL — CREATE USER [search-service-name] FROM EXTERNAL PROVIDER; \
308                 ALTER ROLE db_datareader ADD MEMBER [...] (SQL AAD users are not ARM RBAC)"
309            ),
310        }),
311        _ => {}
312    }
313}
314
315/// Parse `ResourceId=/subscriptions/...;<rest>` connection strings.
316pub fn parse_resource_id(conn: &str) -> Option<String> {
317    let start = conn.find("ResourceId=")? + "ResourceId=".len();
318    let rest = &conn[start..];
319    let end = rest.find(';').unwrap_or(rest.len());
320    let id = rest[..end].trim();
321    (id.starts_with("/subscriptions/") && !id.contains('<')).then(|| id.to_string())
322}
323
324fn dedup(edges: Vec<IdentityEdge>) -> Vec<IdentityEdge> {
325    let mut seen = std::collections::BTreeSet::new();
326    edges
327        .into_iter()
328        .filter(|e| {
329            seen.insert((
330                format!("{:?}", e.principal),
331                e.scope.clone().unwrap_or_else(|| e.target.clone()),
332                e.role_id.clone(),
333                e.role_name.clone(),
334            ))
335        })
336        .collect()
337}
338
339#[cfg(test)]
340mod tests {
341    use super::*;
342    use crate::resources::traits::ResourceRef;
343    use crate::workspace::{PROJECT_FILE, PROJECTS_DIR, WORKSPACE_FILE};
344    use serde_json::json;
345
346    fn ws_with(resources: &[(ResourceKind, &str, Value)]) -> (tempfile::TempDir, Workspace) {
347        let tmp = tempfile::tempdir().unwrap();
348        std::fs::write(
349            tmp.path().join(WORKSPACE_FILE),
350            "environments:\n  dev:\n    default: true\n    search: { service: s }\n    foundry: { account: f, project: p }\n",
351        )
352        .unwrap();
353        let pdir = tmp.path().join(PROJECTS_DIR).join("demo");
354        std::fs::create_dir_all(&pdir).unwrap();
355        std::fs::write(pdir.join(PROJECT_FILE), "{}\n").unwrap();
356        let ws = Workspace::load(tmp.path()).unwrap();
357        {
358            let store = Store::new(ws.project("demo").unwrap(), "dev");
359            for (kind, name, value) in resources {
360                store.write(&ResourceRef::new(*kind, *name), value).unwrap();
361            }
362        }
363        let ws = Workspace::load(tmp.path()).unwrap();
364        (tmp, ws)
365    }
366
367    #[test]
368    fn blob_datasource_yields_storage_edge_with_scope() {
369        let (_tmp, ws) = ws_with(&[(
370            ResourceKind::DataSource,
371            "ds",
372            json!({
373                "name": "ds",
374                "type": "azureblob",
375                "credentials": {"connectionString": "ResourceId=/subscriptions/s1/resourceGroups/rg/providers/Microsoft.Storage/storageAccounts/acct;"},
376                "container": {"name": "c"}
377            }),
378        )]);
379        let edges = identity_edges(&ws, "dev");
380        assert_eq!(edges.len(), 1);
381        let e = &edges[0];
382        assert_eq!(e.principal, Principal::SearchService);
383        assert_eq!(e.kind, EdgeKind::Rbac);
384        assert_eq!(e.role_name, "Storage Blob Data Reader");
385        assert_eq!(
386            e.scope.as_deref(),
387            Some(
388                "/subscriptions/s1/resourceGroups/rg/providers/Microsoft.Storage/storageAccounts/acct"
389            )
390        );
391    }
392
393    #[test]
394    fn cosmos_and_sql_are_informational() {
395        let (_tmp, ws) = ws_with(&[
396            (
397                ResourceKind::DataSource,
398                "cds",
399                json!({"name": "cds", "type": "cosmosdb", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.DocumentDB/databaseAccounts/c;Database=d"}, "container": {"name": "x"}}),
400            ),
401            (
402                ResourceKind::DataSource,
403                "sds",
404                json!({"name": "sds", "type": "azuresql", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.Sql/servers/sv;Database=d"}, "container": {"name": "t"}}),
405            ),
406        ]);
407        let edges = identity_edges(&ws, "dev");
408        assert_eq!(edges.len(), 2);
409        assert!(edges.iter().all(|e| e.kind == EdgeKind::Informational));
410    }
411
412    #[test]
413    fn kb_models_agent_grounding_and_vectorizers() {
414        let (_tmp, ws) = ws_with(&[
415            (
416                ResourceKind::KnowledgeBase,
417                "kb",
418                json!({"name": "kb", "models": [{"kind": "azureOpenAI"}], "knowledgeSources": [{"name": "ks"}]}),
419            ),
420            (
421                ResourceKind::Agent,
422                "agent",
423                json!({"name": "agent", "model": "m", "tools": [{"type": "mcp", "x-rigg-ref": "knowledge-bases/kb"}]}),
424            ),
425            (
426                ResourceKind::Index,
427                "idx",
428                json!({"name": "idx", "fields": [], "vectorSearch": {"vectorizers": [{"name": "v"}]}}),
429            ),
430        ]);
431        let edges = identity_edges(&ws, "dev");
432        let roles: Vec<&str> = edges.iter().map(|e| e.role_name.as_str()).collect();
433        assert!(roles.contains(&"Cognitive Services User"), "{roles:?}");
434        assert!(roles.contains(&"Search Index Data Reader"), "{roles:?}");
435        assert!(
436            roles.contains(&"Cognitive Services OpenAI User"),
437            "{roles:?}"
438        );
439        let grounding = edges
440            .iter()
441            .find(|e| e.role_name == "Search Index Data Reader")
442            .unwrap();
443        assert_eq!(grounding.principal, Principal::FoundryProject);
444    }
445
446    #[test]
447    fn skillset_identity_ai_services_yields_account_edge() {
448        let (_tmp, ws) = ws_with(&[(
449            ResourceKind::Skillset,
450            "ss",
451            json!({"name": "ss", "skills": [], "cognitiveServices": {
452                "@odata.type": "#Microsoft.Azure.Search.AIServicesByIdentity",
453                "subdomainUrl": "https://myaisvc.cognitiveservices.azure.com/"
454            }}),
455        )]);
456        let edges = identity_edges(&ws, "dev");
457        let edge = edges
458            .iter()
459            .find(|e| e.target == "AI services account 'myaisvc'")
460            .expect("ai services edge");
461        assert_eq!(edge.kind, EdgeKind::Rbac);
462        assert_eq!(edge.role_name, "Cognitive Services User");
463        assert_eq!(edge.principal, Principal::SearchService);
464    }
465
466    #[test]
467    fn skillset_webapi_edges_are_informational() {
468        let (_tmp, ws) = ws_with(&[(
469            ResourceKind::Skillset,
470            "ss",
471            json!({"name": "ss", "skills": [
472                {"@odata.type": "#Microsoft.Skills.Custom.WebApiSkill",
473                 "uri": "https://fn.azurewebsites.net/api/enrich",
474                 "authResourceId": "api://12345"},
475                {"@odata.type": "#Microsoft.Skills.Custom.WebApiSkill",
476                 "uri": "https://fn2.azurewebsites.net/api/other"}
477            ]}),
478        )]);
479        let edges = identity_edges(&ws, "dev");
480        let webapi: Vec<_> = edges
481            .iter()
482            .filter(|e| e.target.starts_with("custom Web API"))
483            .collect();
484        assert_eq!(webapi.len(), 2);
485        assert!(webapi.iter().all(|e| e.kind == EdgeKind::Informational));
486        assert!(
487            webapi
488                .iter()
489                .any(|e| e.reason.contains("authResourceId: api://12345"))
490        );
491        assert!(webapi.iter().any(|e| e.reason.contains("without AAD auth")));
492    }
493
494    #[test]
495    fn placeholder_resource_ids_are_not_scopes() {
496        assert_eq!(
497            parse_resource_id("ResourceId=/subscriptions/<subscription-id>/...;"),
498            None
499        );
500        assert_eq!(parse_resource_id("AccountKey=zzz"), None);
501        assert_eq!(
502            parse_resource_id("ResourceId=/subscriptions/a/resourceGroups/b;Database=d").as_deref(),
503            Some("/subscriptions/a/resourceGroups/b")
504        );
505    }
506
507    #[test]
508    fn duplicate_edges_dedup() {
509        let ds = json!({"name": "ds", "type": "azureblob", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.Storage/storageAccounts/a;"}, "container": {"name": "c"}});
510        let mut ds2 = ds.clone();
511        ds2["name"] = json!("ds2");
512        let (_tmp, ws) = ws_with(&[
513            (ResourceKind::DataSource, "ds", ds),
514            (ResourceKind::DataSource, "ds2", ds2),
515        ]);
516        assert_eq!(
517            identity_edges(&ws, "dev").len(),
518            1,
519            "same scope+role dedups"
520        );
521    }
522}
rigg_core/identity.rs

rigg_core/
identity.rs
