Changeset 0.15.0 patches (#72)

.gitea/workflows/ci.yaml

@@ -161,7 +161,7 @@ jobs:
           PGPORT: ${{ env.POSTGRES_PORT }}
           PGUSER: ${{ secrets.POSTGRES_USER }}
           PGPASSWORD: ${{ secrets.POSTGRES_PASSWORD }}
-        run: go test -v -race -count=1 ./...
+        run: go test -v -race -count=1 -p 1 ./...
 
       - name: Drop CI test database
         if: always()

VERSION

@@ -1 +1 @@
-0.14.0
+0.15.0

docs/ROADMAP.md

@@ -749,15 +749,30 @@ See [DESIGN-0.14.0.md](DESIGN-0.14.0.md) for full spec.
 
 ---
 
-## v0.15.0 — Compaction
+## v0.15.0 — Compaction ✅
 
 Depends on: utility model role (v0.10.0).
 
-- [ ] Auto-compaction service: background job that calls utility model to summarize
+- [x] Compaction service: extracts messages, calls utility model to summarize, replaces with summary
-- [ ] Channel-scoped: triggers when context exceeds threshold
+- [x] `Summarize & Continue` button: user-triggered compaction via existing summarize endpoint
-- [ ] Summaries stored as system messages
+- [x] Channel-scoped context threshold triggers (configurable min messages, min chars, activity gap)
-- [ ] Configurable: per-channel opt-in/out, summary model override
+- [x] Summaries stored as system messages with `is_summary = true` metadata
-- [ ] Admin controls for resource limits
+- [x] Background scanner: periodic candidate detection with cooldown and in-flight dedup
+- [x] Configurable: per-channel opt-in/out (`compaction_opt_out`), global threshold overrides
+- [x] Admin controls: global enable/disable, threshold/cooldown settings via `global_settings`
+- [x] Context budget guard rail: estimated tokens must exceed 80% of ceiling before compaction
+- [x] Token estimation: `(len + 3) / 4` heuristic with path-from-summary support
+- [x] Rate limiting on summarize endpoint
+- [x] Comprehensive test suite: scanner candidates, guard rails, live pipeline integration tests
+
+Bug fixes during CI stabilization:
+- [x] Fixed `channels.deleted_at` → `channels.is_archived` (channels use soft-archive, not soft-delete)
+- [x] Fixed NULL model column scan in scanner (`COALESCE(c.model, '')`)
+- [x] Fixed test channel backdating (BEFORE UPDATE trigger override via INSERT)
+- [x] Fixed vault nil-check fallback in role resolver for plaintext API keys
+- [x] Fixed parallel test DB races (`-p 1` serialization)
+- [x] Fixed proactive token refresh cascade: failed refresh no longer nukes active session
+- [x] Fixed model visibility reset: preferences fetch failure preserves existing hidden state
 
 ---
 
@@ -1160,6 +1175,7 @@ based on need.
 
 **UX / Multi-Seat**
 - ~~Per-chat model/preset persistence (server-side)~~ ✅ _(v0.12.0 — `channels.settings.last_selector_id` JSONB merge, localStorage write-through cache)_
+- Per-provider model preferences: `user_model_settings` unique key is `(user_id, model_id)` — same model from different providers shares one visibility toggle. Needs `provider_config_id` dimension in DB constraint, store, API, and frontend `hiddenModels` keying. Frontend composite key (`configId:modelId`) already exists in `App.models[].id`.
 
 **Knowledge Bases — Future**
 - Hybrid search: combine vector similarity with full-text `tsvector`, re-rank

server/compaction/compaction_test.go

@@ -43,14 +43,25 @@ func setGlobalSetting(t *testing.T, key string, value interface{}) {
 	}
 }
 
-// touchChannelTime backdates a channel's updated_at.
+// seedChannelBackdated creates a channel with updated_at set to (now - age).
-func touchChannelTime(t *testing.T, channelID string, age time.Duration) {
+// Uses INSERT with an explicit timestamp so the BEFORE UPDATE trigger on
+// channels (which overwrites updated_at = NOW()) never fires.
+func seedChannelBackdated(t *testing.T, userID, title string, age time.Duration, msgCount, contentSize int) (channelID string, msgIDs []string) {
 	t.Helper()
 	target := time.Now().Add(-age)
-	_, err := database.DB.Exec(`UPDATE channels SET updated_at = $1 WHERE id = $2`, target, channelID)
+	err := database.DB.QueryRow(`
+		INSERT INTO channels (user_id, title, type, updated_at)
+		VALUES ($1, $2, 'direct', $3)
+		RETURNING id
+	`, userID, title, target).Scan(&channelID)
 	if err != nil {
-		t.Fatalf("touchChannelTime: %v", err)
+		t.Fatalf("seedChannelBackdated: %v", err)
 	}
+	msgIDs = database.SeedTestMessages(t, channelID, msgCount, contentSize)
+	if len(msgIDs) > 0 {
+		database.SeedTestCursor(t, channelID, userID, msgIDs[len(msgIDs)-1])
+	}
+	return
 }
 
 // setChannelSettings sets the channel.settings JSONB.
@@ -89,10 +100,8 @@ func TestScanner_FindCandidates_ReturnsQualifying(t *testing.T) {
 
 	// Create a channel with enough messages to qualify
 	// 20 messages × 2000 chars = 40K chars (> candidateMinChars=20K)
-	channelID, _ := seedChannel(t, userID, "Big Chat", 20, 2000)
+	// Backdated 5min so it passes activity gap (>2min) and recency (<7 days)
-
+	channelID, _ := seedChannelBackdated(t, userID, "Big Chat", 5*time.Minute, 20, 2000)
-	// Backdate so it passes the activity gap (>2min old) and recency (<7 days)
-	touchChannelTime(t, channelID, 5*time.Minute)
 
 	ctx := context.Background()
 	candidates := sc.findCandidates(ctx)
@@ -122,8 +131,7 @@ func TestScanner_FindCandidates_ExcludesTooFewMessages(t *testing.T) {
 	userID := database.SeedTestUser(t, "scanuser2", "scan2@test.com")
 
 	// Only 4 messages — below candidateMinMessages=10
-	channelID, _ := seedChannel(t, userID, "Small Chat", 4, 2000)
+	channelID, _ := seedChannelBackdated(t, userID, "Small Chat", 5*time.Minute, 4, 2000)
-	touchChannelTime(t, channelID, 5*time.Minute)
 
 	ctx := context.Background()
 	candidates := sc.findCandidates(ctx)
@@ -164,8 +172,7 @@ func TestScanner_FindCandidates_ExcludesArchived(t *testing.T) {
 	sc := NewScanner(svc, stores, ScannerConfig{})
 
 	userID := database.SeedTestUser(t, "scanuser4", "scan4@test.com")
-	channelID, _ := seedChannel(t, userID, "Archived Chat", 20, 2000)
+	channelID, _ := seedChannelBackdated(t, userID, "Archived Chat", 5*time.Minute, 20, 2000)
-	touchChannelTime(t, channelID, 5*time.Minute)
 
 	// Archive the channel
 	database.DB.Exec(`UPDATE channels SET is_archived = true WHERE id = $1`, channelID)
@@ -377,8 +384,9 @@ func TestScanner_GetCooldownDuration_Override(t *testing.T) {
 
 func TestEstimateTokens_GuardRailMath(t *testing.T) {
 	// Simulate a large conversation that exceeds a 32K utility model
-	// 100K chars ≈ 25K tokens of conversation + system prompt overhead
+	// 120K chars ≈ 30K tokens of conversation + system prompt overhead
-	largeContent := make([]byte, 100000)
+	// inputCeiling = 32K * 0.80 = 25.6K → 30K exceeds it
+	largeContent := make([]byte, 120000)
 	contentTokens := EstimateTokens(string(largeContent))
 	systemTokens := EstimateTokens("You are a conversation summarizer...") + 8 // +overhead
 
@@ -387,7 +395,7 @@ func TestEstimateTokens_GuardRailMath(t *testing.T) {
 	inputCeiling := int(float64(utilityBudget) * 0.80)
 
 	if totalPrompt <= inputCeiling {
-		t.Errorf("100K chars (%d tokens) should exceed 32K model ceiling (%d tokens)",
+		t.Errorf("120K chars (%d tokens) should exceed 32K model ceiling (%d tokens)",
 			totalPrompt, inputCeiling)
 	}
 

server/compaction/scanner.go

@@ -237,13 +237,12 @@ func (sc *Scanner) findCandidates(ctx context.Context) []models.Channel {
 	maxAge := time.Now().Add(-candidateMaxAge)
 
 	rows, err := database.DB.QueryContext(ctx, `
-		SELECT c.id, c.user_id, c.model, c.settings::text,
+		SELECT c.id, c.user_id, COALESCE(c.model, ''), COALESCE(c.settings::text, '{}'),
 		       COUNT(m.id) AS msg_count,
 		       COALESCE(SUM(LENGTH(m.content)), 0) AS total_chars
 		FROM channels c
 		JOIN messages m ON m.channel_id = c.id AND m.deleted_at IS NULL
-		WHERE c.deleted_at IS NULL
+		WHERE c.type = 'direct'
-		  AND c.type = 'direct'
 		  AND c.is_archived = false
 		  AND c.updated_at < $1
 		  AND c.updated_at > $2

server/handlers/completion.go

@@ -829,18 +829,6 @@ func (h *CompletionHandler) loadConversation(channelID, userID, presetSystemProm
 	return messages, nil
 }
 
-// isSummaryMessage checks if a PathMessage has summary metadata.
-func isSummaryMessage(m *PathMessage) bool {
-	if m.Metadata == nil {
-		return false
-	}
-	var meta map[string]interface{}
-	if err := json.Unmarshal(*m.Metadata, &meta); err != nil {
-		return false
-	}
-	return meta["type"] == "summary"
-}
-
 // ── Message Persistence ─────────────────────
 
 // persistMessage inserts a message into the tree, updates the cursor, and

server/handlers/summarize.go

@@ -33,7 +33,7 @@ func (h *SummarizeHandler) Summarize(c *gin.Context) {
 	// ── Verify channel ownership ──
 	var ownerID string
 	err := database.DB.QueryRow(
-		`SELECT user_id FROM channels WHERE id = $1 AND deleted_at IS NULL`, channelID,
+		`SELECT user_id FROM channels WHERE id = $1`, channelID,
 	).Scan(&ownerID)
 	if err != nil {
 		c.JSON(http.StatusNotFound, gin.H{"error": "channel not found"})

server/main.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/gin-gonic/gin"
 
+	"git.gobha.me/xcaliber/chat-switchboard/compaction"
 	"git.gobha.me/xcaliber/chat-switchboard/config"
 	"git.gobha.me/xcaliber/chat-switchboard/crypto"
 	"git.gobha.me/xcaliber/chat-switchboard/database"
@@ -248,8 +249,9 @@ func main() {
 			protected.POST("/chat/completions", comp.Complete)
 			protected.GET("/tools", comp.ListTools)
 
-			// Summarize & Continue
+			// Summarize & Continue (backed by compaction service)
-			summarize := handlers.NewSummarizeHandler(stores, roleResolver)
+			compactionSvc := compaction.NewService(stores, roleResolver)
+			summarize := handlers.NewSummarizeHandler(compactionSvc)
 			protected.POST("/channels/:id/summarize", summarize.Summarize)
 
 			// Provider Configs (user-facing — replaces /api-configs)
@@ -400,6 +402,7 @@ func main() {
 			admin.PUT("/users/:id/role", adm.UpdateUserRole)
 			admin.PUT("/users/:id/active", adm.ToggleUserActive)
 			admin.POST("/users/:id/reset-password", adm.ResetPassword)
+			admin.POST("/users/:id/vault/reset", adm.ResetVault)
 			admin.DELETE("/users/:id", adm.DeleteUser)
 
 			// Global settings

server/roles/roles.go

@@ -279,6 +279,9 @@ func (r *Resolver) resolveBinding(ctx context.Context, binding *RoleBinding) (pr
 		if err != nil {
 			return nil, providers.ProviderConfig{}, "", fmt.Errorf("decrypt API key: %w", err)
 		}
+	} else if len(cfg.APIKeyEnc) > 0 {
+		// No vault — key stored as raw bytes (unencrypted fallback)
+		apiKey = string(cfg.APIKeyEnc)
 	}
 
 	// Parse headers

src/js/api.js

@@ -131,7 +131,15 @@ const API = {
         this._refreshTimer = setTimeout(async () => {
             if (!this.refreshToken) return;
             console.debug('🔄 Proactive token refresh');
-            await this.refresh();
+            try {
+                const data = await this._post('/api/v1/auth/refresh', { refresh_token: this.refreshToken }, true);
+                this._setAuth(data);
+            } catch (e) {
+                // Proactive refresh failed — keep session alive.
+                // Access token is still valid for ~20% of its lifetime.
+                // When it naturally expires, 401-retry in _authed() handles cleanup.
+                console.warn('⚠ Proactive refresh failed, access token still valid');
+            }
         }, ms);
     },
 

src/js/app.js

@@ -62,7 +62,11 @@ async function fetchModels() {
             App.hiddenModels = new Set(
                 (prefData.preferences || []).filter(p => p.hidden).map(p => p.model_id)
             );
-        } catch (e) { App.hiddenModels = new Set(); }
+        } catch (e) {
+            // Keep existing preferences on failure (auth dead, network issue).
+            // Only init to empty if there's nothing to preserve.
+            if (!App.hiddenModels) App.hiddenModels = new Set();
+        }
 
         App.models = (data.models || []).map(m => {
             const isPreset = !!m.is_preset;

src/js/ui-settings.js

@@ -509,7 +509,9 @@ Object.assign(UI, {
                     App.hiddenModels = new Set(
                         (prefData.preferences || []).filter(p => p.hidden).map(p => p.model_id)
                     );
-                } catch (e) { App.hiddenModels = new Set(); }
+                } catch (e) {
+                    if (!App.hiddenModels) App.hiddenModels = new Set();
+                }
             }
 
             const data = await API.listEnabledModels();