knowledge: stamp model + backend on facts for audit

Adds two nullable TEXT columns to entity_facts —
`created_by_model` (LLM identifier) and `created_by_backend`
("local" / "hybrid" / "manual" / NULL) — so the curator can audit
which configurations produce good fact-keeping and which produce
noise.

photo_insights already carries model_version + backend, and
entity_facts.source_insight_id links to it, but:
  - source_insight_id is set post-loop, so chat-continuation and
    regenerated-insight facts lose the link.
  - JOINing per read is more friction than embedding provenance on
    the row itself.
  - Manual facts (POST /knowledge/facts) have no insight at all and
    need their own "manual" provenance marker.

Threading: execute_tool grows `model` + `backend` params, passed
from the three call sites (agentic insight loop, chat single-turn,
chat stream) using the loop-time `chat_backend.primary_model()` +
`effective_backend` already in scope. tool_store_fact stamps the
new fact accordingly; manual create_fact stamps backend="manual".
Legacy rows leave both NULL — pre-tracking data can't be back-
filled reliably from training_messages without burning compute.

Indexes are partial (WHERE NOT NULL) so legacy rows don't bloat
them, and "show me all facts from model X" stays fast.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

migrations/2026-05-10-000300_entity_facts_provenance/down.sql

@@ -0,0 +1,4 @@
+DROP INDEX IF EXISTS idx_entity_facts_created_by_backend;
+DROP INDEX IF EXISTS idx_entity_facts_created_by_model;
+ALTER TABLE entity_facts DROP COLUMN created_by_backend;
+ALTER TABLE entity_facts DROP COLUMN created_by_model;

migrations/2026-05-10-000300_entity_facts_provenance/up.sql

@@ -0,0 +1,30 @@
+-- Track which model + backend generated each fact so the curator
+-- can audit which configurations produce trustworthy knowledge.
+--
+-- photo_insights already carries `model_version` + `backend`, and
+-- entity_facts.source_insight_id links to it — but:
+--   1. source_insight_id is only set after an insight is stored
+--      (post-loop), so chat-continuation facts and facts whose insight
+--      was regenerated lose the link.
+--   2. JOINing for every read is more friction than just embedding the
+--      provenance on the fact row itself.
+--   3. Manual facts (POST /knowledge/facts) have no insight at all and
+--      need to record "manual" as their provenance.
+--
+-- Two nullable TEXT columns are enough for the audit use case: model
+-- (e.g. "qwen2.5:7b", "anthropic/claude-sonnet-4") and backend
+-- ("local", "hybrid", "manual"). Pre-existing rows leave both NULL —
+-- legacy facts predate this tracking and can't be back-filled
+-- reliably from training_messages without burning compute.
+
+ALTER TABLE entity_facts ADD COLUMN created_by_model TEXT;
+ALTER TABLE entity_facts ADD COLUMN created_by_backend TEXT;
+
+-- Indexes are cheap and useful for "show me all facts from model X"
+-- audit queries — partial so the legacy NULL rows don't bloat them.
+CREATE INDEX idx_entity_facts_created_by_model
+    ON entity_facts(created_by_model)
+    WHERE created_by_model IS NOT NULL;
+CREATE INDEX idx_entity_facts_created_by_backend
+    ON entity_facts(created_by_backend)
+    WHERE created_by_backend IS NOT NULL;

src/ai/insight_chat.rs

@@ -497,6 +497,8 @@ impl InsightChatService {
                             &normalized,
                             req.user_id,
                             &active_persona,
+                            &model_used,
+                            &effective_backend,
                             &loop_cx,
                         )
                         .await;
@@ -870,6 +872,8 @@ impl InsightChatService {
                 &normalized,
                 req.user_id,
                 &active_persona,
+                &model_used,
+                &effective_backend,
                 max_iterations,
                 &tx,
             )
@@ -1059,6 +1063,8 @@ impl InsightChatService {
                 &normalized,
                 req.user_id,
                 &active_persona,
+                &model_used,
+                &effective_backend,
                 max_iterations,
                 &tx,
             )
@@ -1210,6 +1216,10 @@ impl InsightChatService {
         normalized: &str,
         user_id: i32,
         active_persona: &str,
+        // Provenance — stamped onto any store_fact tool call made
+        // during this loop. Mirrors the non-streaming chat path.
+        model_used: &str,
+        effective_backend: &str,
         max_iterations: usize,
         tx: &tokio::sync::mpsc::Sender<ChatStreamEvent>,
     ) -> Result<AgenticLoopOutcome> {
@@ -1290,6 +1300,8 @@ impl InsightChatService {
                             normalized,
                             user_id,
                             active_persona,
+                            model_used,
+                            effective_backend,
                             &cx,
                         )
                         .await;

src/ai/insight_generator.rs

@@ -1554,6 +1554,13 @@ Return ONLY the summary, nothing else."#,
         file_path: &str,
         user_id: i32,
         persona_id: &str,
+        // Provenance — written into entity_facts.created_by_* when
+        // the loop calls store_fact. The caller knows the actual
+        // chat-runtime model and backend (which may differ from
+        // ollama.primary_model in hybrid mode where chat lives on
+        // OpenRouter while Ollama still handles vision).
+        model: &str,
+        backend: &str,
         cx: &opentelemetry::Context,
     ) -> String {
         let result = match tool_name {
@@ -1574,8 +1581,10 @@ Return ONLY the summary, nothing else."#,
             }
             "store_entity" => self.tool_store_entity(arguments, ollama, cx).await,
             "store_fact" => {
-                self.tool_store_fact(arguments, file_path, user_id, persona_id, cx)
+                self.tool_store_fact(
-                    .await
+                    arguments, file_path, user_id, persona_id, model, backend, cx,
+                )
+                .await
             }
             "get_current_datetime" => Self::tool_get_current_datetime(),
             unknown => format!("Unknown tool: {}", unknown),
@@ -2632,6 +2641,8 @@ Return ONLY the summary, nothing else."#,
         file_path: &str,
         user_id: i32,
         persona_id: &str,
+        model: &str,
+        backend: &str,
         cx: &opentelemetry::Context,
     ) -> String {
         use crate::database::models::{InsertEntityFact, InsertEntityPhotoLink};
@@ -2700,6 +2711,8 @@ Return ONLY the summary, nothing else."#,
             valid_from,
             valid_until: None,
             superseded_by: None,
+            created_by_model: Some(model.to_string()),
+            created_by_backend: Some(backend.to_string()),
         };
 
         let mut kdao = self
@@ -3730,6 +3743,8 @@ Return ONLY the summary, nothing else."#,
                             &file_path,
                             user_id,
                             &persona_id,
+                            chat_backend.primary_model(),
+                            &backend_label,
                             &loop_cx,
                         )
                         .await;

src/database/knowledge_dao.rs

@@ -1470,6 +1470,8 @@ mod tests {
                     valid_from: None,
                     valid_until: None,
                     superseded_by: None,
+                    created_by_model: None,
+                    created_by_backend: None,
                 },
             )
             .unwrap();
@@ -1691,6 +1693,8 @@ mod tests {
                 valid_from: None,
                 valid_until: None,
                 superseded_by: None,
+                created_by_model: None,
+                created_by_backend: None,
             },
         );
         assert!(
@@ -1923,6 +1927,8 @@ mod tests {
                     valid_from: None,
                     valid_until: None,
                     superseded_by: None,
+                    created_by_model: None,
+                    created_by_backend: None,
                 },
             )
             .unwrap();

src/database/models.rs

@@ -262,6 +262,12 @@ pub struct InsertEntityFact {
     /// the supersede endpoint; status flips to 'superseded' in the
     /// same transaction. See migration 2026-05-10-000200.
     pub superseded_by: Option<i32>,
+    /// Provenance for model audit — see migration 2026-05-10-000300.
+    /// `created_by_model` is the LLM identifier (e.g. "qwen2.5:7b",
+    /// "anthropic/claude-sonnet-4") or NULL for legacy / manual rows.
+    /// `created_by_backend` is "local" / "hybrid" / "manual" / NULL.
+    pub created_by_model: Option<String>,
+    pub created_by_backend: Option<String>,
 }
 
 #[derive(Serialize, Queryable, Clone, Debug)]
@@ -281,6 +287,8 @@ pub struct EntityFact {
     pub valid_from: Option<i64>,
     pub valid_until: Option<i64>,
     pub superseded_by: Option<i32>,
+    pub created_by_model: Option<String>,
+    pub created_by_backend: Option<String>,
 }
 
 #[derive(Insertable)]

src/database/schema.rs

@@ -62,6 +62,8 @@ diesel::table! {
         valid_from -> Nullable<BigInt>,
         valid_until -> Nullable<BigInt>,
         superseded_by -> Nullable<Integer>,
+        created_by_model -> Nullable<Text>,
+        created_by_backend -> Nullable<Text>,
     }
 }
 

src/knowledge.rs

@@ -119,6 +119,10 @@ pub struct FactDetail {
     /// supersession, migration 2026-05-10-000200). Only set when
     /// status == 'superseded'.
     pub superseded_by: Option<i32>,
+    /// Provenance — see migration 2026-05-10-000300. NULL on legacy
+    /// rows. `created_by_backend` is "local" / "hybrid" / "manual".
+    pub created_by_model: Option<String>,
+    pub created_by_backend: Option<String>,
     /// Set when another active fact has the same subject+predicate,
     /// a different object, AND their valid-time intervals overlap.
     /// Detected at read time by the get_entity handler grouping
@@ -432,6 +436,8 @@ async fn get_entity<D: KnowledgeDao + 'static>(
             valid_from: f.valid_from,
             valid_until: f.valid_until,
             superseded_by: f.superseded_by,
+            created_by_model: f.created_by_model,
+            created_by_backend: f.created_by_backend,
             in_conflict: false,
         });
     }
@@ -768,6 +774,11 @@ async fn create_fact<D: KnowledgeDao + 'static>(
         valid_from: body.valid_from,
         valid_until: body.valid_until,
         superseded_by: None,
+        // Manual creation via curation UI — provenance recorded as
+        // "manual" with no model, distinguishing user-entered facts
+        // from agent-generated ones in the audit view.
+        created_by_model: None,
+        created_by_backend: Some("manual".to_string()),
     };
 
     match dao.upsert_fact(&cx, insert) {