feat(P5): IHF Phase 5 complete — agent-assisted distillation

Adds bounded AI support to the IHF governance loop. All AI outputs are
attributed (model_ref), reviewable (AgentReviewRecord), and reversible.
No autonomous decisions; no silent requirement promotion.

- T01: Schema — agent_proposals, agent_review_records,
  confidence_annotations (migration 1743379200)
- T02: AgentProposalsController (index/show/accept/reject, idempotent
  review guard), global nav "Agent" link
- T03: SummarizeClusterAction — Claude API cluster summary on widget show
- T04: DraftRequirementAction — AI requirement draft; acceptance creates
  RequirementCandidate (human-gated)
- T05: DetectDuplicatesAction — duplicate_flag proposal on candidate show
- T06: DetectPolicySensitivityAction — policy_flag with
  ConfidenceAnnotations per concern scope
- T07: ProposeImplementationAction — impl_proposal from decision show
- T08: AgentAuditDashboardAction — autoRefresh; KPI row, unreviewed queue,
  recent proposals, attribution log matrix
- T09: integration tests, SCOPE.md updated, phase5-summary.md, flake.nix
  adds http-conduit/aeson/string-conversions

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Web/Controller/Widgets.hs

@@ -9,7 +9,8 @@ import Generated.Types
 import IHP.Prelude
 import IHP.ControllerPrelude
 import Data.Aeson (toJSON, object, (.=))
-import Application.Helper.Controller (isInRegression, widgetCycleCounts)
+import Application.Helper.Controller (isInRegression, widgetCycleCounts, callClaudeApi)
+import Data.List (intercalate)
 
 instance Controller WidgetsController where
     beforeAction = ensureIsUser
@@ -121,3 +122,65 @@ instance Controller WidgetsController where
                         |> createRecord
                     setSuccessMessage "Widget updated"
                     redirectTo ShowWidgetAction { widgetId = widget.id }
+
+    -- T03: Summarize feedback cluster via Claude API
+    action SummarizeClusterAction { widgetId } = do
+        annotations <- query @Annotation
+            |> filterWhere (#widgetId, widgetId)
+            |> orderByDesc #createdAt
+            |> limit 20
+            |> fetch
+        threads <- query @AnnotationThread
+            |> filterWhere (#widgetId, widgetId)
+            |> orderByDesc #createdAt
+            |> limit 20
+            |> fetch
+        let annLines = map (\a -> "[" <> a.category <> "/" <> a.severity <> "] " <> a.body) annotations
+            threadLines = map (\t -> "[thread] " <> t.title <> ": " <> fromMaybe "" t.description) threads
+            userMsg = intercalate "\n" (annLines <> threadLines)
+        result <- liftIO $ callClaudeApi
+            "You are a distillation assistant for a governed interaction hub. Summarize the following user feedback cluster into a concise, actionable summary (2\x20134 sentences). Be factual and neutral."
+            userMsg
+            300
+        case result of
+            Left err -> do
+                setErrorMessage ("AI summarization failed: " <> err)
+                redirectTo ShowWidgetAction { widgetId }
+            Right content -> do
+                newRecord @AgentProposal
+                    |> set #proposalType "summary"
+                    |> set #sourceWidgetId (Just widgetId)
+                    |> set #content content
+                    |> set #modelRef "claude-sonnet-4-6"
+                    |> set #status "pending"
+                    |> createRecord
+                setSuccessMessage "Summary proposal created"
+                redirectTo ShowWidgetAction { widgetId }
+
+    -- T04: Draft a requirement candidate via Claude API
+    action DraftRequirementAction { widgetId } = do
+        annotations <- query @Annotation
+            |> filterWhere (#widgetId, widgetId)
+            |> orderByDesc #createdAt
+            |> limit 20
+            |> fetch
+        let annLines = map (\a -> "[" <> a.category <> "/" <> a.severity <> "] " <> a.body) annotations
+            userMsg = intercalate "\n" annLines
+        result <- liftIO $ callClaudeApi
+            "You are a requirements analyst. Given these friction annotations, draft a single structured requirement candidate. Respond with JSON: {\"title\": \"...\", \"description\": \"...\"}."
+            userMsg
+            400
+        case result of
+            Left err -> do
+                setErrorMessage ("AI draft failed: " <> err)
+                redirectTo ShowWidgetAction { widgetId }
+            Right content -> do
+                newRecord @AgentProposal
+                    |> set #proposalType "requirement_draft"
+                    |> set #sourceWidgetId (Just widgetId)
+                    |> set #content content
+                    |> set #modelRef "claude-sonnet-4-6"
+                    |> set #status "pending"
+                    |> createRecord
+                setSuccessMessage "Requirement draft proposal created"
+                redirectTo ShowWidgetAction { widgetId }