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.
91pub fn identity_edges(ws: &Workspace) -> Vec<IdentityEdge> {
92    let mut edges: Vec<IdentityEdge> = Vec::new();
93    for project in &ws.projects {
94        let store = Store::new(project);
95        let Ok(files) = store.list() else { continue };
96        for (r, _) in files {
97            let Ok(value) = store.read(&r) else { continue };
98            match r.kind {
99                ResourceKind::DataSource => datasource_edges(&r.name, &value, &mut edges),
100                ResourceKind::KnowledgeBase
101                    if value
102                        .get("models")
103                        .is_some_and(|m| !m.as_array().is_none_or(|a| a.is_empty()))
104                        || value.get("answerInstructions").is_some() =>
105                {
106                    edges.push(IdentityEdge::rbac(
107                        Principal::SearchService,
108                        None,
109                        "Foundry account (model access)",
110                        roles::COGNITIVE_SERVICES_USER,
111                        format!(
112                            "knowledge base '{}' uses a model for retrieval/synthesis",
113                            r.name
114                        ),
115                    ));
116                }
117                ResourceKind::Skillset if value.to_string().contains("AzureOpenAI") => {
118                    edges.push(IdentityEdge::rbac(
119                        Principal::SearchService,
120                        None,
121                        "Foundry account (model access)",
122                        roles::COGNITIVE_SERVICES_OPENAI_USER,
123                        format!("skillset '{}' calls Azure OpenAI", r.name),
124                    ));
125                }
126                ResourceKind::Index => {
127                    let has_vectorizer = value
128                        .get("vectorSearch")
129                        .and_then(|v| v.get("vectorizers"))
130                        .and_then(Value::as_array)
131                        .is_some_and(|a| !a.is_empty());
132                    if has_vectorizer {
133                        edges.push(IdentityEdge::rbac(
134                            Principal::SearchService,
135                            None,
136                            "Foundry account (model access)",
137                            roles::COGNITIVE_SERVICES_OPENAI_USER,
138                            format!("index '{}' has vectorizers", r.name),
139                        ));
140                    }
141                }
142                ResourceKind::Agent => {
143                    for (kind, name) in registry::extract_references(r.kind, &value) {
144                        if kind == ResourceKind::KnowledgeBase {
145                            edges.push(IdentityEdge::rbac(
146                                Principal::FoundryProject,
147                                None,
148                                "Search service (knowledge base retrieval)",
149                                roles::SEARCH_INDEX_DATA_READER,
150                                format!("agent '{}' grounds on knowledge base '{name}'", r.name),
151                            ));
152                        }
153                    }
154                }
155                _ => {}
156            }
157            // encryption keys → Key Vault
158            if let Some(uri) = value
159                .get("encryptionKey")
160                .and_then(|k| k.get("keyVaultUri"))
161                .and_then(Value::as_str)
162            {
163                edges.push(IdentityEdge::rbac(
164                    Principal::SearchService,
165                    None,
166                    format!("Key Vault {uri}"),
167                    roles::KEY_VAULT_SECRETS_USER,
168                    format!(
169                        "{} '{}' uses customer-managed encryption",
170                        r.kind.display_name(),
171                        r.name
172                    ),
173                ));
174            }
175        }
176    }
177    dedup(edges)
178}
179
180fn datasource_edges(name: &str, value: &Value, edges: &mut Vec<IdentityEdge>) {
181    let ds_type = value
182        .get("type")
183        .and_then(Value::as_str)
184        .unwrap_or_default();
185    let conn = value
186        .get("credentials")
187        .and_then(|c| c.get("connectionString"))
188        .and_then(Value::as_str)
189        .unwrap_or_default();
190    let scope = parse_resource_id(conn);
191    match ds_type {
192        "azureblob" | "adlsgen2" | "azurefile" | "azurefiles" | "azuretable" => {
193            edges.push(IdentityEdge::rbac(
194                Principal::SearchService,
195                scope.clone(),
196                scope
197                    .clone()
198                    .unwrap_or_else(|| "storage account (set ResourceId= in the connection string)".into()),
199                roles::STORAGE_BLOB_DATA_READER,
200                format!("data source '{name}' ({ds_type}) reads from storage"),
201            ));
202        }
203        "cosmosdb" => edges.push(IdentityEdge {
204            principal: Principal::SearchService,
205            scope: scope.clone(),
206            target: scope.unwrap_or_else(|| "Cosmos DB account".into()),
207            role_id: String::new(),
208            role_name: "Cosmos DB Built-in Data Reader (SQL role)".into(),
209            kind: EdgeKind::Informational,
210            reason: format!(
211                "data source '{name}' reads Cosmos DB — grant via `az cosmosdb sql role assignment create` \
212                 (Cosmos data-plane roles are not ARM RBAC)"
213            ),
214        }),
215        "azuresql" => edges.push(IdentityEdge {
216            principal: Principal::SearchService,
217            scope: scope.clone(),
218            target: scope.unwrap_or_else(|| "Azure SQL database".into()),
219            role_id: String::new(),
220            role_name: "db_datareader (contained AAD user)".into(),
221            kind: EdgeKind::Informational,
222            reason: format!(
223                "data source '{name}' reads Azure SQL — CREATE USER [search-service-name] FROM EXTERNAL PROVIDER; \
224                 ALTER ROLE db_datareader ADD MEMBER [...] (SQL AAD users are not ARM RBAC)"
225            ),
226        }),
227        _ => {}
228    }
229}
230
231/// Parse `ResourceId=/subscriptions/...;<rest>` connection strings.
232pub fn parse_resource_id(conn: &str) -> Option<String> {
233    let start = conn.find("ResourceId=")? + "ResourceId=".len();
234    let rest = &conn[start..];
235    let end = rest.find(';').unwrap_or(rest.len());
236    let id = rest[..end].trim();
237    (id.starts_with("/subscriptions/") && !id.contains('<')).then(|| id.to_string())
238}
239
240fn dedup(edges: Vec<IdentityEdge>) -> Vec<IdentityEdge> {
241    let mut seen = std::collections::BTreeSet::new();
242    edges
243        .into_iter()
244        .filter(|e| {
245            seen.insert((
246                format!("{:?}", e.principal),
247                e.scope.clone().unwrap_or_else(|| e.target.clone()),
248                e.role_id.clone(),
249                e.role_name.clone(),
250            ))
251        })
252        .collect()
253}
254
255#[cfg(test)]
256mod tests {
257    use super::*;
258    use crate::resources::traits::ResourceRef;
259    use crate::workspace::{PROJECT_FILE, PROJECTS_DIR, WORKSPACE_FILE};
260    use serde_json::json;
261
262    fn ws_with(resources: &[(ResourceKind, &str, Value)]) -> (tempfile::TempDir, Workspace) {
263        let tmp = tempfile::tempdir().unwrap();
264        std::fs::write(
265            tmp.path().join(WORKSPACE_FILE),
266            "environments:\n  dev:\n    default: true\n    search: { service: s }\n    foundry: { account: f, project: p }\n",
267        )
268        .unwrap();
269        let pdir = tmp.path().join(PROJECTS_DIR).join("demo");
270        std::fs::create_dir_all(&pdir).unwrap();
271        std::fs::write(pdir.join(PROJECT_FILE), "{}\n").unwrap();
272        let ws = Workspace::load(tmp.path()).unwrap();
273        {
274            let store = Store::new(ws.project("demo").unwrap());
275            for (kind, name, value) in resources {
276                store.write(&ResourceRef::new(*kind, *name), value).unwrap();
277            }
278        }
279        let ws = Workspace::load(tmp.path()).unwrap();
280        (tmp, ws)
281    }
282
283    #[test]
284    fn blob_datasource_yields_storage_edge_with_scope() {
285        let (_tmp, ws) = ws_with(&[(
286            ResourceKind::DataSource,
287            "ds",
288            json!({
289                "name": "ds",
290                "type": "azureblob",
291                "credentials": {"connectionString": "ResourceId=/subscriptions/s1/resourceGroups/rg/providers/Microsoft.Storage/storageAccounts/acct;"},
292                "container": {"name": "c"}
293            }),
294        )]);
295        let edges = identity_edges(&ws);
296        assert_eq!(edges.len(), 1);
297        let e = &edges[0];
298        assert_eq!(e.principal, Principal::SearchService);
299        assert_eq!(e.kind, EdgeKind::Rbac);
300        assert_eq!(e.role_name, "Storage Blob Data Reader");
301        assert_eq!(
302            e.scope.as_deref(),
303            Some(
304                "/subscriptions/s1/resourceGroups/rg/providers/Microsoft.Storage/storageAccounts/acct"
305            )
306        );
307    }
308
309    #[test]
310    fn cosmos_and_sql_are_informational() {
311        let (_tmp, ws) = ws_with(&[
312            (
313                ResourceKind::DataSource,
314                "cds",
315                json!({"name": "cds", "type": "cosmosdb", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.DocumentDB/databaseAccounts/c;Database=d"}, "container": {"name": "x"}}),
316            ),
317            (
318                ResourceKind::DataSource,
319                "sds",
320                json!({"name": "sds", "type": "azuresql", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.Sql/servers/sv;Database=d"}, "container": {"name": "t"}}),
321            ),
322        ]);
323        let edges = identity_edges(&ws);
324        assert_eq!(edges.len(), 2);
325        assert!(edges.iter().all(|e| e.kind == EdgeKind::Informational));
326    }
327
328    #[test]
329    fn kb_models_agent_grounding_and_vectorizers() {
330        let (_tmp, ws) = ws_with(&[
331            (
332                ResourceKind::KnowledgeBase,
333                "kb",
334                json!({"name": "kb", "models": [{"kind": "azureOpenAI"}], "knowledgeSources": [{"name": "ks"}]}),
335            ),
336            (
337                ResourceKind::Agent,
338                "agent",
339                json!({"name": "agent", "model": "m", "tools": [{"type": "mcp", "x-rigg-ref": "knowledge-bases/kb"}]}),
340            ),
341            (
342                ResourceKind::Index,
343                "idx",
344                json!({"name": "idx", "fields": [], "vectorSearch": {"vectorizers": [{"name": "v"}]}}),
345            ),
346        ]);
347        let edges = identity_edges(&ws);
348        let roles: Vec<&str> = edges.iter().map(|e| e.role_name.as_str()).collect();
349        assert!(roles.contains(&"Cognitive Services User"), "{roles:?}");
350        assert!(roles.contains(&"Search Index Data Reader"), "{roles:?}");
351        assert!(
352            roles.contains(&"Cognitive Services OpenAI User"),
353            "{roles:?}"
354        );
355        let grounding = edges
356            .iter()
357            .find(|e| e.role_name == "Search Index Data Reader")
358            .unwrap();
359        assert_eq!(grounding.principal, Principal::FoundryProject);
360    }
361
362    #[test]
363    fn placeholder_resource_ids_are_not_scopes() {
364        assert_eq!(
365            parse_resource_id("ResourceId=/subscriptions/<subscription-id>/...;"),
366            None
367        );
368        assert_eq!(parse_resource_id("AccountKey=zzz"), None);
369        assert_eq!(
370            parse_resource_id("ResourceId=/subscriptions/a/resourceGroups/b;Database=d").as_deref(),
371            Some("/subscriptions/a/resourceGroups/b")
372        );
373    }
374
375    #[test]
376    fn duplicate_edges_dedup() {
377        let ds = json!({"name": "ds", "type": "azureblob", "credentials": {"connectionString": "ResourceId=/subscriptions/s/resourceGroups/r/providers/Microsoft.Storage/storageAccounts/a;"}, "container": {"name": "c"}});
378        let mut ds2 = ds.clone();
379        ds2["name"] = json!("ds2");
380        let (_tmp, ws) = ws_with(&[
381            (ResourceKind::DataSource, "ds", ds),
382            (ResourceKind::DataSource, "ds2", ds2),
383        ]);
384        assert_eq!(identity_edges(&ws).len(), 1, "same scope+role dedups");
385    }
386}
rigg_core/identity.rs

rigg_core/
identity.rs
