core/CHANGELOG.md

Changelog

All notable changes to Chat Switchboard.

[0.22.6] — 2026-03-03

Fixed

• Split FE/BE deployment: page route proxying. Frontend container now supports BACKEND_URL env var. When set, the entrypoint generates nginx proxy_pass blocks for page routes (/, /login, /chat/:id, /admin/*, /editor/*, /notes/*, /settings/*) so Go template rendering works in k8s split-image deployments. When unset, behavior is unchanged (SPA fallback). Both BASE_PATH="" and BASE_PATH="/prefix" variants supported.
• CI test timeout. All go test commands now include -timeout 8m to prevent indefinite hangs from runner resource contention or race-detector compilation stalls. Previously used Go's 10-minute default with no goroutine dump on timeout.

Removed

• router.js (322 lines). Client-side hash router fully replaced by server-side page routes. All Router.* call sites in chat.js replaced with history.replaceState() for URL updates. All typeof Router guards in app.js removed.
• surfaces.js (368 lines). Client-side surface registry replaced by Go template surface architecture. Extension API stubs (surfaces.register, surfaces.activate, etc.) now log deprecation warnings. surfaces.getCurrent() returns window.__SURFACE__ from server-rendered page data.
• index.html SPA shell (1,296 lines → 16 lines). Original monolithic SPA entry point replaced with a minimal redirect page. All real routes now served by Go templates; index.html only serves as nginx try_files fallback for unknown paths.
• editor-mode.js old Surfaces paths (~260 lines). Removed _register(), openDirect(), _unregister(), _activate(), _deactivate(), _embedChat(), _returnChat(), and full check()/checkStartup() implementations. Only mountServerRendered() remains as the entry point. check() and checkStartup() kept as no-ops for backward compat with projects-ui.js callers.
• SPA-only entrypoint code path. docker-entrypoint-fe.sh now requires BACKEND_URL (fails fast if unset). Removed dead %%BASE_HREF%%, %%ENVIRONMENT%%, %%BRANDING_JSON%% injection into index.html. Removed branding config file loading (branding now handled by Go templates).

Changed

• docker-entrypoint-fe.sh: Simplified from 227 → 184 lines. Requires BACKEND_URL. Removed SPA-only fallback path.
• nginx.conf (unified container): Consolidated repeated proxy_set_header blocks. Uses $backend variable. Page routes proxy to Go backend for template rendering.
• base.html: Added __ENV__ and __BRANDING__ globals for JS compatibility (previously injected by index.html sed).
• server/pages/pages.go: Added Environment field to PageData, populated from config.
• extensions.js: ui.replace(), ui.restore(), and surfaces.* API methods now log deprecation warnings instead of calling removed Surfaces system.
• app.js: Removed Surfaces.init(), EditorMode.checkStartup(), and Router.init() calls. Simplified to direct sessionStorage chat restore.
• chat.js: Replaced Router.update() calls with history.replaceState() for URL-bar sync on chat selection/creation.
• repl.js: Removed surfaces.js import reference.

[0.22.5] — 2026-03-03

Added

• Go template engine. Server-side HTML rendering via html/template with //go:embed for compiled-in templates. Custom FuncMap with helpers (roleFilterType, toJSON, hasPrefix, dict). CSP nonce generation per request. Replaces monolithic index.html + client-side DOM construction with composable, server-rendered surfaces.
• Surface architecture. Each page route renders a dedicated surface template that owns its full layout below the classification banner. Surfaces compose from reusable components (model-select, team-select, file-upload) without sharing CSS layout rules. Banner height handled via CSS custom properties (--banner-top-h, --banner-bot-h, --surface-h).
• Chat surface. Server-rendered shell at / and /chat/:chatID. Go template renders banner + layout + script tags; existing JS builds DOM inside the container. Bridge pattern preserves all current chat functionality.
• Admin surface. Full admin panel at /admin/:section with 5 category tabs (People, AI, Routing, System, Monitoring) and section sidebar. Server-rendered pages for providers, models, teams, users, and settings. Hybrid fallback for JS-loaded sections (groups, presets, knowledge, memory, health, capabilities, extensions, storage, usage, audit, stats).
• Editor surface. Server-rendered layout shell at /editor/:wsId with proper CSS height ownership. mountServerRendered() method on EditorMode bypasses the Surfaces registry for Go template mode. Dedicated CSS owns full viewport below banners — no layout collision with other surfaces. (Fixes bugs #4, #5: editor layout collapse)
• Notes surface. Server-rendered layout at /notes/:noteId with notes-main + notes-assist split. Boot script calls openNotes() in standalone mode. Responsive: assist pane hidden on mobile.
• Settings surface. Full-page settings at /settings/:section replacing modal-based settings. Left nav with all sections (general, appearance, providers, models, presets, roles, knowledge, memory, notifications, usage). General and Appearance sections server-rendered with form controls; dynamic sections populated by existing JS.
• Login page. Standalone Go template at /login with cookie-based auth. Sets redirect_after_login cookie for post-login navigation.
• AuthOrRedirect middleware. Cookie-based JWT validation for page routes. Reads sb_token cookie, redirects to /login on missing/invalid token (instead of 401 JSON). RequireAdminPage() helper aborts with 403 for non-admin users. Skips auth entirely in unmanaged mode.
• Cookie-based auth sync. api.js saveTokens() now sets sb_token cookie alongside localStorage on every token save/refresh. clearTokens() clears the cookie. Bridges API auth (Bearer header) with page auth (cookie).
• Reusable model-select component. Go template partial renders <select> with models filtered by type. Server passes filtered list; client-side cascade handler re-filters from window.__PAGE_DATA__.models on provider change. (Fixes bug #1: admin model roles missing embedding models)
• Reusable team-select component. Go template partial renders team dropdown with pre-populated options from server data loader. (Fixes bug #2: routing policy team scope as text field)
• Reusable file-upload component. Go template partial for drop zone + file picker. Used by editor surface.
• Editor file upload. Upload button in editor toolbar + drag-and-drop on file tree with visual feedback. API.uploadWorkspaceFile() sends File objects as raw body to existing WriteFile handler (binary-safe, respects workspace quota). Auto-opens single uploaded text files. (Fixes bug #3: unable to upload files to a project)
• Page route data loaders. Each surface registers a loader that pre-fetches exactly what its template needs: chatLoader, adminLoader (with section-specific data), editorLoader, notesLoader, settingsLoader. No over-fetching, no client-side fetch waterfall.
• Admin provider CRUD. Server-rendered provider table with add/edit form, type dropdown, sync button. Client-side handlers for create, update, delete, sync operations.
• Admin model table. Server-rendered catalog table with client-side search + type/provider filters.
• Admin user management. Server-rendered user table with search, role edit, enable/disable toggles.
• Admin settings page. Registration policy, permissions, banner config, and system prompt — all server-rendered with save handlers.
• pages.js client handlers. Role CRUD, routing policy CRUD, provider/model/team/user management, settings save — all working against existing API endpoints with proper token resolution.

Changed

• nginx.conf: Added page route proxy blocks (/, /login, /chat, /admin, /editor, /notes, /settings) before SPA fallback. Static assets still served directly by nginx. Removed /legacy proxy block.
• server/main.go: Template engine init, pages.SetVersion(), route wiring for all surfaces with auth middleware groups. Removed /legacy fallback route.
• src/js/api.js: saveTokens() / clearTokens() sync sb_token cookie. New uploadWorkspaceFile() method for binary file upload to workspaces.
• src/js/editor-mode.js: Upload button + hidden file input in header. Drag-and-drop on file tree container with .drag-over visual state. _uploadFiles() method with text-file auto-open. mountServerRendered() for Go template boot path.
• src/css/editor-mode.css: .drag-over style for file tree drop target.

Fixed

• Bug #1: Admin model roles now show embedding models. Server-side roleFilterType() sets FilterType="embedding" for the embedding role; model-select component filters by model_type.
• Bug #2: Routing policy team scope uses proper dropdown instead of free-text field. adminLoader pre-fetches team list; team-select component renders options.
• Bug #3: Editor file upload works. Upload button + drag-and-drop + uploadWorkspaceFile() API method.
• Bug #4: Editor chat assist pane is functional. Server-rendered layout with dedicated mount points.
• Bug #5: Editor and chat pane no longer squashed. Each surface owns its full viewport height via dedicated CSS — no shared flex container conflicts.

Removed

• /legacy route and nginx proxy block. All page routes are now server-rendered.

[0.22.4] — 2026-03-02

Added

• Rate limit tracking. Provider health windows now track HTTP 429 / rate limit events separately from general errors. New rate_limit_count column on provider_health table. RecordRateLimit() method on health accumulator. Completion handler and stream loop classify HTTP 429 / rate limit / Too Many Requests errors automatically.
• Auto-disable policy. Providers that report "down" status for N consecutive hourly windows are automatically deactivated. Configurable via PROVIDER_AUTO_DISABLE_THRESHOLD env var (default: 3, set to 0 to disable). AutoDisabler interface on health store, checkAutoDisable() runs on each flush cycle.
• Tool health tracking. Built-in tools (web_search, url_fetch, etc.) now record success/error/latency to tool_health table via the health accumulator. RecordToolSuccess() / RecordToolError() methods. Both tool execution sites in stream loop instrumented. ToolHealthWindow and ToolHealthSummary models.
• Search provider health tracking. web_search and url_fetch tool calls are now included in the tool health pipeline with per-invocation latency and error recording.
• Project-specific file uploads. POST /api/v1/projects/:id/files (multipart upload) and GET /api/v1/projects/:id/files (list). project_id column on attachments table. Files are stored under projects/{id}/ prefix. Project access verified via ownership or team membership. Files queued for text extraction if enabled.
• Workspace settings UI. Git configuration section in project settings panel: remote URL, branch, credential selector. Appears when a workspace is bound. Saves via PATCH /api/v1/workspaces/:id. updateWorkspace() and listGitCredentials() API client methods.
• PDF/DOCX export via pandoc. New POST /api/v1/export endpoint converts markdown content to PDF or DOCX using pandoc. Editor export dropdown now includes "Export as PDF" and "Export as DOCX" options. Graceful error handling when pandoc is unavailable.
• Project files UI. File list and upload button in project settings panel. Shows file names and sizes. Multi-file upload support.
• API client methods. projectUploadFile, projectListFiles, exportDocument, updateWorkspace, listGitCredentials.

Changed

• health/accumulator.go: Added rateLimitCount to bucket, RecordRateLimit(), RecordToolSuccess(), RecordToolError(), flushTools(), checkAutoDisable(), SetAutoDisable(), AutoDisabler interface, toolBucket struct. Store interface extended with UpsertToolWindow(), ListAllToolCurrent().
• handlers/completion.go: HealthRecorder interface extended with RecordRateLimit(), RecordToolSuccess(), RecordToolError(). recordHealth() classifies HTTP 429 errors.
• handlers/stream_loop.go: Both tool execution sites now record tool health. recordHealthFn() classifies rate limit errors.
• models/models.go: RateLimitCount field on ProviderHealthWindow. ToolHealthWindow, ToolHealthSummary types. ErrorCount, RateLimitCount, TimeoutCount fields on ProviderHealthSummary. ProjectID field on Attachment.
• store/postgres/health.go: All queries updated for rate_limit_count. UpsertToolWindow(), ListAllToolCurrent(), DeactivateProvider() methods.
• store/sqlite/health.go: Full rewrite with rate_limit_count, tool health, and auto-disable support.
• store/postgres/attachment.go: project_id in cols, scan, create, and new GetByProject().
• store/sqlite/attachment.go: Same project_id additions.
• store/interfaces.go: GetByProject() on AttachmentStore.
• config/config.go: ProviderAutoDisableThreshold field + env var loading.
• main.go: SQLite health store branching, auto-disable wiring, export handler route, project file routes.
• handlers/health_admin.go: Health summary includes error_count, rate_limit_count, timeout_count.
• src/js/ui-admin.js: Health dashboard cards show "Rate Limits" count with warning color.
• src/js/projects-ui.js: Git settings section, project files section, file upload handler.
• src/js/editor-mode.js: PDF and DOCX export menu items and handler.
• src/js/api.js: 5 new API client methods.

Database

• Migration 015 (Postgres) / 014 (SQLite): rate_limit_count column on provider_health, tool_health table, project_id column on attachments.

[0.22.3] — 2026-03-02

Added

• Provider admin UI: "Routing" category. New admin category with three sections: Health, Routing, and Capabilities. Accessible from admin panel sidebar navigation.
• Health dashboard. Admin section showing per-provider status badges, request counts, error rates, average/max latency, timeout counts, and last error messages. Refresh button for live updates. CSS grid layout with responsive cards.
• Routing policy builder. Full CRUD UI for routing policies: create/edit form with name, priority, type (provider_prefer/team_route/cost_limit/model_alias), scope (global/team), team ID, JSON config editor, active toggle. List view with edit/enable/disable/delete actions.
• Routing dry-run test panel. Input model + user ID, evaluates against all active policies with live health status, displays ranked candidates with status badges and selection reasoning.
• Capability override viewer. Table listing all model capability overrides with model ID, provider config, field, value, and delete button. Explains auto-detection fallback when no overrides exist.
• provider_status on models/enabled response. GET /api/v1/models/enabled now includes provider_status field ("healthy"/"degraded"/"down"/"unknown") per model from live health data.
• ModelHandler with health enrichment. Extracted model listing into dedicated handlers/capabilities.go with SetHealthStore() for health status injection. Builds health map from current hourly windows.
• GET /api/v1/admin/capability-overrides endpoint. Returns all capability overrides across all models (admin view).
• API client methods. adminGetAllProviderHealth, adminListRoutingPolicies, adminGetRoutingPolicy, adminCreateRoutingPolicy, adminUpdateRoutingPolicy, adminDeleteRoutingPolicy, adminTestRouting, adminListCapabilityOverrides, adminDeleteModelCapability.

Changed

• src/js/ui-admin.js: New "routing" admin category with health/routing/capabilities sections. ADMIN_SECTIONS, ADMIN_LABELS, ADMIN_LOADERS updated. Added loadAdminHealth(), loadAdminRouting(), loadAdminCapabilities(), _showRoutingForm(), _toggleRoutingPolicy(), _deleteRoutingPolicy(), _runRoutingTest(), _deleteCapOverride().
• src/js/api.js: 9 new API client methods for health, routing, and capability admin endpoints.
• src/index.html: New admin category button for "Routing". Three new admin-section-content divs: adminHealthTab, adminRoutingTab, adminCapabilitiesTab.
• src/css/styles.css: .admin-health-grid and .admin-health-card styles.
• models/models.go: ProviderStatus field on UserModel struct.
• handlers/capabilities.go: Extracted ModelHandler from inline handlers. ListEnabledModels enriches response with provider health status. ResolveModelCaps canonical capability resolver.
• main.go: ModelHandler created with SetHealthStore() wiring. Capability override admin routes registered.

[0.22.2] — 2026-03-02

Added

• Routing policy engine. Rules-based request routing evaluated between "what model was requested" and "which provider serves it." routing/ package with evaluator, fallback runner, and type-safe policy model.
  • provider_prefer: Ordered list of preferred provider configs — fallback to others if preferred are unavailable.
  • team_route: Restrict a team to only use specific provider configs (team-scoped access control).
  • cost_limit: Filter out providers above a heuristic cost threshold per request.
  • model_alias: Map alias names ("fast", "smart") to specific provider+model pairs.
• Health-aware routing. Candidates sorted by health status (healthy > degraded > unknown > down). skip_down flag removes down providers from consideration when healthy alternatives exist. All-down graceful degradation: keeps candidates rather than returning empty.
• Fallback runner. routing.RunWithFallback() tries candidates in order with configurable retry depth. Logs each failure and tries the next candidate. Returns the winning candidate and updated decision with fallback depth.
• routing_policies table. New migration (Postgres 014, SQLite 013) with scope (global/team), priority ordering, JSONB config, and active flag. Index on (is_active, priority) for efficient policy loading.
• Routing policy admin CRUD. Full REST API for policy management:
  • GET /api/v1/admin/routing/policies — list all policies
  • GET /api/v1/admin/routing/policies/:id — get single policy
  • POST /api/v1/admin/routing/policies — create policy (validates type, scope, team_id)
  • PUT /api/v1/admin/routing/policies/:id — update policy
  • DELETE /api/v1/admin/routing/policies/:id — delete policy
• Dry-run routing test. POST /api/v1/admin/routing/test accepts model + user_id + team_ids, evaluates active policies against all global provider configs with live health status, returns ranked candidates and the routing decision.
• X-Switchboard-Provider response header. Every completion response includes providerID/configID for routing observability.
• routing_decision column on usage_log. JSONB field for recording which policy matched, fallback depth, and total candidates evaluated.
• RoutingPolicyStore interface with Postgres and SQLite implementations. Supports Create, Update, Delete, GetByID, ListActive, ListAll, ListForTeam.
• 12 evaluator tests. Coverage: no policies, provider_prefer ordering, team_route filtering (including wrong-team no-op), model_alias rewrite, health skip-down, health sorting, all-down graceful degradation, priority ordering, empty candidates, inactive policy skip.

Changed

• models/models.go: Added RoutingPolicy struct and RoutingDecision field on UsageEntry.
• store/interfaces.go: Added RoutingPolicyStore interface and RoutingPolicies field on Stores.
• store/postgres/stores.go, store/sqlite/stores.go: Wired RoutingPolicyStore into factory.
• handlers/completion.go: Added router, healthStore fields. HealthStatusQuerier interface. evaluateRouting() method builds candidates from accessible configs, queries health status, runs evaluator, reloads credentials on config switch. X-Switchboard-Provider, X-Switchboard-Route, X-Switchboard-Fallback response headers.
• main.go: Creates routing.Evaluator, wires into completion handler via SetRoutingEvaluator() and SetHealthStore(). Registers 6 admin routing routes.

Database

• Postgres: server/database/migrations/014_v0222_routing.sql
• SQLite: server/database/migrations/sqlite/013_v0222_routing.sql

[0.22.1] — 2026-03-02

Added

• Provider profile schemas. Each provider type declares its configurable settings with types, defaults, validation constraints, and dependency rules. Built-in schemas for openai, anthropic, venice, openrouter. Unknown types fall back to the openai schema. providers/profile.go.
• Provider hooks (pre-request / post-stream). Declarative request/response transforms driven by provider_configs.settings JSONB. Replaces any need for hardcoded provider switch statements in the completion path. providers/hooks.go.
  • OpenAI: system_prompt_prefix, frequency_penalty, presence_penalty → injected into ExtraBody.
  • Anthropic: extended_thinking + thinking_budget → injects thinking config into wire JSON, sets anthropic-beta header, clears temperature.
  • Venice: enable_thinking, enable_web_search, include_venice_system_prompt → builds venice_parameters in ExtraBody.
  • OpenRouter: route, require_parameters → serialized into X-Provider-Preferences header.
• ExtraBody on CompletionRequest. Provider hooks write arbitrary key-value pairs that get merged into the wire-format JSON via mergeExtraBody() in both OpenAI and Anthropic doRequest methods. Keeps the canonical request type clean while supporting provider-specific extensions.
• Anthropic extended thinking stream support. content_block_delta with type=thinking_delta now routes to the Reasoning field on StreamEvent, matching the existing reasoning_content path used by OpenAI-compatible providers.
• Provider type registry. providers.RegisterType() combines provider implementation + metadata (name, description, default endpoint, profile schema). providers.ListTypes() returns all registered types.
• GET /api/v1/admin/provider-types endpoint. Returns metadata and profile schemas for all registered provider types. Used by admin UI to render provider creation forms and show available settings per type.
• Preset setting overrides. MergePresetSettings() merges persona-level overrides onto provider-level settings, respecting ProviderOnly fields that cannot be changed at the preset level.
• Comprehensive tests. providers/hooks_test.go: 17 tests covering all four hook implementations, profile schemas, GetHooks, MergePresetSettings, mergeExtraBody, nil safety.

Changed

• providers/registry.go: Init() now uses RegisterType() with full metadata instead of bare Register(). ProviderTypeMeta struct with ID, name, description, default endpoint, profile schema.
• providers/provider.go: CompletionRequest.ExtraBody field (json:"-"). mergeExtraBody() helper function. Model struct unchanged.
• providers/openai.go: doRequest() merges ExtraBody into wire JSON before HTTP dispatch.
• providers/anthropic.go: doRequest() merges ExtraBody + sets anthropic-beta header when thinking is enabled. Stream processing handles thinking_delta content block deltas. Wire type gains Thinking field.
• handlers/completion.go: PreRequest hook invoked at all three dispatch paths (multi-model, streaming, sync).
• handlers/stream_loop.go: PostStreamEvent hook invoked after each streaming event in both streamWithToolLoop and streamModelResponse.
• handlers/health_admin.go: Added GetProviderTypes handler.
• main.go: GET /api/v1/admin/provider-types route registered.

[0.22.0] — 2026-03-02

Added

• Provider health tracking. In-memory accumulator records success/error/timeout per provider config, flushes hourly buckets to provider_health table every 60s. Status derivation: healthy (<5% errors), degraded (5–25%), down (>25%). Background prune removes data older than 7 days.
• Health admin endpoints. GET /api/v1/admin/providers/:id/health returns per-provider summary + hourly windows. GET /api/v1/admin/providers/health returns current-hour status for all providers.
• Capability admin overrides. capability_overrides table supports per-provider or global overrides for any model capability field. Three-tier resolution chain: catalog → heuristic → admin override (highest priority).
• Capability override endpoints. GET /api/v1/admin/models/:id/capabilities with source annotations (catalog/heuristic/override). PUT to set, DELETE to remove, GET /api/v1/admin/capability-overrides to list all.
• Workspace pane layout. Frontend architecture replaces .chat-area + .side-panel with .workspace flex container. Primary and secondary panes are independent with drag-resize handle between them. Surfaces declare layout preferences via primary, secondary, secondaryOpts fields.

Fixed

• scanJSON driver buffer aliasing (v0.21.7 bugfix). safe_json.go now copies the []byte from the database driver instead of aliasing the slice header. Fixes intermittent invalid character errors on multi-row JSON responses caused by driver buffer reuse between rows.Next() calls.
• CreateChannel RETURNING path now uses scanJSON(&ch.Settings) instead of bare &ch.Settings scan.
• UpdateChannel settings validation — json.Valid() check before JSONB merge, returns 400 on invalid JSON instead of storing corrupt data.

Changed

• capabilities/intrinsic.go: ResolveIntrinsic() takes []models.CapabilityOverride as third parameter. applyOverrides() flips bool fields and sets int fields.
• handlers/completion.go: HealthRecorder interface, recordHealth() called after every ChatCompletion call.
• handlers/stream_loop.go: streamWithToolLoop and streamModelResponse accept configID + HealthRecorder params. Health recorded on error, timeout, and success paths.
• store/interfaces.go: CapOverrides CapabilityOverrideStore added to Stores struct.
• models/models.go: ProviderStatus, ProviderHealthWindow, ProviderHealthSummary, CapabilityOverride types.
• main.go: Health accumulator lifecycle (start/stop/prune), admin route registration, startup log includes health status.
• panels.js: Simplified to single _active panel in workspace secondary pane. Removed _dualMode, _splitRatio, _secondary.
• surfaces.js: Layout declarations (primary, secondary, secondaryOpts) in surface registration.
• editor-mode.js: Layout declarations for editor surface.
• index.html: Workspace container structure, removed dual-view split button.
• styles.css: Workspace pane CSS grid, handle, responsive rules.
• app.js: _initWorkspaceResize() replaces _initSidePanelResize() + _initDualDivider().
• ui-settings.js: Zoom selector targets updated class names.

Database

• Migration 013 (Postgres) / 012 (SQLite): provider_health table (hourly bucketed metrics), capability_overrides table (admin corrections).

[0.21.6] — 2026-03-01

Added

• Editor surface: document features merged in. Article mode (outline, export, word count, focus mode) folded into the single Editor surface rather than shipping as a separate surface. One surface, two concerns: code files get tabs + tree, text files get word count + export. Chat panel on the right with show/hide toggle.
• New file button — Visible in both the file tree header (+ icon) and the editor toolbar ("+ New"). Creates files with smart defaults (markdown files get # Title scaffold).
• Export dropdown — Download file, Export as HTML (via marked.js + DOMPurify), Copy to clipboard. Available from editor toolbar for any open file.
• Chat panel toggle — Show/hide the right-side AI chat pane. Editor left pane expands to fill when chat is hidden. Toggle button in toolbar with active state indicator.
• Word count + reading time in status bar for text files (markdown, txt, rst, adoc). ~230 wpm calculation.
• Save indicator in status bar — shows "● Modified" with warning color for unsaved files.
• Git status indicators in editor file tree (closing v0.21.5 deferred) — Tree rows show M/A/U badges with color coding (modified=yellow, added=green, untracked=italic gray, deleted=red strikethrough).
• Auto-save on tab/mode switch in editor surface (closing v0.21.5 deferred) — Modified files auto-save when switching tabs or deactivating the editor surface.
• Ctrl+Shift+F workspace search (closing v0.21.5 deferred) — Maps to Quick Open file finder.
• SEED_PROVIDERS env var — Auto-creates global providers on startup from CSV format provider:api_key[:name]. Supports openai, anthropic, openrouter, venice, mistral, groq, together, fireworks, deepseek, perplexity with auto-filled endpoints. Idempotent. Blocked in production.
• Mode selector labels — Surface buttons in sidebar show icon + label text ("Chat", "Editor"). Labels hide when sidebar collapsed.
• Hash Router (router.js) — URL-driven direct-to-surface navigation:
  • #/chat → default chat view
  • #/chat/ch_abc123 → open specific chat (bookmarkable)
  • #/editor → editor surface (auto-picks workspace from project, or shows picker)
  • #/editor/ws_abc123 → editor with specific workspace Browser back/forward works. Replaces sessionStorage chat restore. Workspace picker overlay when navigating without a workspace. Auto-selects if only one exists.
• openDirect(wsId) method on EditorMode — bypasses check() flow, registers surface directly with a given workspace ID. Used by Router for hash-based navigation.

Fixed

• chat.switched / chat.created events never emitted — selectChat(), newChat(), and sendMessage() now emit these events. This was blocking EditorMode from detecting workspace bindings on channel switch.
• workspace_id missing from channel responses — Added to all SELECT/RETURNING/Scan paths for both SQLite and Postgres. Frontend chat objects now include workspace_id.
• EditorMode.check() optimization — Checks local App.chats data before API calls.
• defaultEndpoints redeclared — Renamed to seedDefaultEndpoints in seed_providers.go to avoid collision with live_provider_test.go.

Removed

• Article surface (article-mode.js, article-mode.css) — Merged into Editor. Document features (export, word count, focus mode) are now part of the unified Editor surface. Article-specific AI tools (suggest_outline, expand_section) deferred to v0.22+ as editor extensions.

Changed

• editor-mode.js: Rebuilt header with New/Export/ChatToggle buttons. File tree has header with + button. Chat panel show/hide. Word count + save indicator in status bar. _createNewFile(), _exportActiveFile(), _toggleChat() methods. openDirect() for Router.
• editor-mode.css: Header labels, export dropdown, chat toggle, file tree header, save indicator, word count styles.
• chat.js: Event emissions, workspace_id in chat objects, hash sync via Router.update().
• app.js: Router initialization replaces sessionStorage restore.
• projects-ui.js: Project list mapping includes workspace_id.
• channels.go: workspace_id in all query/scan paths.
• surfaces.js: Updated comment (removed article reference).
• styles.css: Mode selector layout, router picker overlay.
• index.html: Removed article-mode.css/article-mode.js, added router.js.
• config.go: SeedProviders field.
• main.go: SeedProviders() call.
• seed_providers.go: seedDefaultEndpoints (renamed from defaultEndpoints).
• ROADMAP.md: Article→Editor merge documented. Go template pages added under v0.25.0.

[0.21.5] — 2026-03-01

Added

• Editor Surface (Development Mode). IDE-like experience built as a v0.21.3 surface consuming workspace primitives. Activates automatically when a channel or project has a bound workspace. Three-pane layout: file tree (sidebar), code editor (CM6), and AI chat panel with resizable split.
• editor-mode.js — Full editor surface module. File tree backed by workspace_ls API with nested directory expansion, file icons by extension, and context menus. Tab bar with modified indicators, close buttons, and multi-file editing. CM6 integration with language auto-detection (20+ languages) and textarea fallback when CM6 bundle unavailable.
• Split pane layout with draggable resize handle. Editor (left) and AI chat (right) share the main region. Chat DOM borrowed from Surfaces saved fragments — real conversation history and input preserved across mode switches.
• Quick Open (Ctrl/Cmd+P): Fuzzy file finder overlay using workspace file index. Ctrl/Cmd+S save, Ctrl/Cmd+W close tab.
• Live editor updates: When AI tool calls modify workspace files (workspace_write, workspace_patch), the server emits workspace.file.changed via WebSocket. If the file is open and unmodified in the editor, content refreshes automatically.
• Status bar: Current file path, language mode, git branch display.
• editor-mode.css: Complete styling for all editor components with dark theme support, mobile responsive (stacked layout at ≤768px).
• Workspace API methods on frontend API module: getWorkspace, listWorkspaceFiles, readWorkspaceFile, writeWorkspaceFile, deleteWorkspaceFile, mkdirWorkspace, getWorkspaceGitStatus, getWorkspaceGitBranches.
• Surfaces.getSavedFragment() / putSavedFragment() — Cross-surface DOM sharing. Allows the editor surface to borrow chat DOM into its split pane without cloning.
• workspace.file. event routing added to server event route table (DirToClient).
• Workspace management UI: Create, list, and bind workspaces from the frontend. Project settings panel gains a Workspace section with dropdown picker and "New" button. Chat context menu gains "Set workspace…" option for per-channel override. Both paths support creating new workspaces inline.
• GET /workspaces — List all workspaces accessible to the current user (user-owned + team-owned).
• workspace_id in channel API: updateChannelRequest accepts workspace_id for binding/unbinding. channelResponse and both list/get queries now include workspace_id.

Changed

• stream_loop.go: Emits workspace.file.changed event after successful workspace_write/workspace_patch tool calls.
• index.html: Added editor-mode.css stylesheet and editor-mode.js script.
• channels.go: channelResponse includes workspace_id, list/get queries select it, update accepts it.
• workspaces.go: Added List handler for GET /workspaces.
• main.go: Added GET /workspaces route.
• projects-ui.js: Workspace section in project panel, workspace picker in chat context menu.
• styles.css: Workspace row layout in project panel.

[0.21.4] — 2026-03-01

Added

• Git Integration. Workspaces can now track a git remote. Operations use os/exec calling the git binary (no CGO, no Go git libraries). Credentials are vault-encrypted (AES-256-GCM) using the same pattern as BYOK provider configs.
• git_credentials table. Postgres + SQLite migrations. Stores encrypted PAT tokens, HTTP basic auth, and SSH private keys. Columns: id, user_id, name, auth_type, encrypted_data, nonce, created_at.
• Workspace git columns. git_remote_url, git_branch, git_credential_id, git_last_sync added to workspaces table via migration.
• workspace.GitOps — exec-based git operations: Clone, Pull, Push, Status, Diff, Commit, Log, BranchList, Checkout. Credential injection via temporary GIT_ASKPASS scripts (PAT), .git-credentials files (basic auth), or GIT_SSH_COMMAND (SSH keys). Temp files securely wiped after each operation.
• Post-operation hooks. Clone, pull, and checkout automatically trigger workspace reconcile (filesystem → DB sync) and re-index changed files through the v0.21.2 indexing pipeline.
• 5 git tools for LLM use: git_status, git_diff, git_commit, git_log, git_branch. Filtered out when no workspace is bound. Each validates git remote configuration before execution.
• 9 git API endpoints: POST clone, POST pull, POST push, GET status, GET diff, POST commit, GET log, GET branches, POST checkout — all under /api/v1/workspaces/:id/git/.
• 3 credential endpoints: POST /api/v1/git-credentials (create), GET /api/v1/git-credentials (list user-scoped summaries), DELETE /api/v1/git-credentials/:id (delete). Encrypted data never exposed in API responses.
• GitCredentialStore interface + Postgres/SQLite implementations. Create, GetByID, ListByUser, Delete with user-scoped authorization.
• Security. file:// and local path URLs rejected in clone. Error output sanitized to strip credential-containing lines. GIT_TERMINAL_PROMPT=0 set on all operations to prevent interactive prompts.

Changed

• WorkspaceStore interface: added SetGitLastSync() method + both implementations.
• models.Workspace: added git tracking fields. models.WorkspacePatch: added git patch fields.
• Completion handler: git tools disabled alongside workspace tools when no workspace bound.
• store.Stores struct: added GitCredentials field.
• Postgres/SQLite NewStores(): wires GitCredentialStore.

[0.21.3] — 2026-03-01

Added

• Surface Registry (surfaces.js). Mode-switching system that allows extensions to register "surfaces" (UI modes) that take over named regions of the interface. DOM nodes are detached and preserved in DocumentFragments during mode switches — not destroyed — ensuring CM6 editor instances and other stateful components survive transitions.
• Surface region attributes. Four data-surface-region attributes on index.html containers: surface-header (model bar), surface-main (chat messages), surface-footer (input area), sidebar-content (search + chat history). Extensions swap these regions via ctx.ui.replace() / ctx.ui.restore().
• Mode selector. Appears in the sidebar (#modeSelectorWrap) when ≥1 extension surface is registered beyond the default chat. Shows icon buttons for each mode with active state highlighting.
• Extension context API. ctx.surfaces.register(), ctx.surfaces.unregister(), ctx.surfaces.activate(), ctx.surfaces.getCurrent() on the scoped extension context. ctx.ui.replace() and ctx.ui.restore() for region DOM management.
• Surface events. surface.activated, surface.deactivated, surface.registered, surface.unregistered on the EventBus (all localOnly).
• REPL console (repl.js). Fourth tab in the debug modal (Ctrl+Shift+L). AsyncFunction wrapper for top-level await. Injected globals: API, Events, Extensions, Surfaces, DebugLog, Panels, UI, plus $(), $$(), sleep() helpers. Features: collapsible JSON pretty-printing, red errors with expandable stack traces, command history via ↑/↓ (persisted to sessionStorage), tab-completion on live object graphs and EventBus labels, multi-line input (Shift+Enter), admin-gated (admin role OR ?debug=1 URL param).
• CSS. Mode selector styles (.mode-selector, .mode-btn, active/hover states). REPL styles (output formatting, value-type color coding, collapsible JSON, input prompt).

Changed

• extensions.js context builder: added ctx.surfaces namespace and ctx.ui.replace() / ctx.ui.restore() methods.
• app.js startup: initializes Surfaces.init() before Extensions.loadAll(), and REPL.init() after auth confirmation.
• index.html: added data-surface-region attributes, mode selector container, REPL tab in debug modal, script tags for surfaces.js and repl.js.
• EXTENSIONS.md §6: updated with implementation details, actual API signatures, and event catalog.

[0.21.2] — 2026-03-01

Added

• Workspace Indexing Pipeline. Background text chunking and embedding for workspace files, reusing the existing knowledge.SplitText chunker and knowledge.Embedder for vector generation. Indexable file types include 40+ source code extensions (Go, Rust, Python, TypeScript, Perl, etc.) plus all text/* MIME types. Code-aware chunking uses 1500-character chunks with function/class/type boundary separators (\n\nfunc , \n\nfn , \n\nclass , \n\ndef , \n\ntype , \n\nsub , \n\npub , \n\nimpl ), falling back to paragraph and line breaks.
• Content-addressed skip. SHA256 computed during WriteFile is compared against the prior hash before triggering re-indexing. Unchanged files skip the chunk + embed pipeline entirely, making archive extracts and bulk writes efficient.
• workspace_chunks table. New table (Postgres + SQLite migrations) storing text chunks with vector embeddings per workspace file. Postgres uses sequential scan with pgvector <=> cosine operator (no IVFFlat/HNSW index — 3072-dim embeddings exceed the 2000-dim index limit; sequential scan is adequate at workspace scale). SQLite uses app-level cosine in Go. Columns: workspace_id, file_id, chunk_index, content, token_count, embedding, metadata (JSONB with line_start, language).
• workspace_files additions. index_status column (pending, indexing, ready, error, skipped) and chunk_count for tracking indexing progress per file. workspaces table gains indexing_enabled boolean (default true).
• WorkspaceStore chunk methods. Four new store interface methods implemented for both Postgres and SQLite: InsertChunks (batch insert with pgvector cast or JSON text), DeleteChunksByFile, SimilaritySearch (pgvector <=> operator or app-level cosine), UpdateFileIndexStatus.
• workspace.Indexer. Background indexing orchestrator with configurable concurrency semaphore. Single-file indexing (IndexFile) triggered on WriteFile when content changes. Batch indexing (IndexBatch) triggered after ExtractArchive. Async goroutines with 5-minute timeout per file. Embedding usage tracked via Embedder.LogUsage.
• workspace_search tool. Semantic search across workspace files via natural language query. Embeds the query, runs cosine similarity against workspace chunks, returns file paths, content snippets, relevance scores, and approximate line numbers. Optional file_pattern glob filtering (e.g. *.go, src/*.ts) applied post-search. Top-K capped at 20.
• Indexer integration. workspace.FS.SetIndexer() hooks the indexer into the write path. WriteFile triggers single-file indexing on content change. ExtractArchive triggers batch indexing for all extracted files. Both use the workspace owner's identity for embedding provider resolution.
• Configuration. WORKSPACE_INDEXING_ENABLED env var (global kill switch, default true). WORKSPACE_INDEX_CONCURRENCY env var (default 2). Per-workspace indexing_enabled toggle via WorkspacePatch. Indexing gracefully degrades when no embedding role is configured.
• Index status endpoint. GET /api/v1/workspaces/:id/index-status returns aggregated indexing progress: per-status file counts (pending, indexing, ready, error, skipped), total chunk count, and workspace indexing_enabled flag.

Changed

• workspace_files scanner updated in both Postgres and SQLite stores to include index_status and chunk_count columns.
• workspaces scanner updated to include indexing_enabled column.
• WorkspacePatch model extended with IndexingEnabled field.
• WorkspaceToolNames() now includes workspace_search in the disabled set when no workspace is bound.
• RegisterWorkspaceSearchTool() added for late registration of the search tool (requires embedder).

[0.21.1] — 2026-03-01

Added

• Workspace Tools. Six AI-callable tools in the workspace category: workspace_ls (list files with content type/size), workspace_read (text content with 50KB soft limit, binary detection), workspace_write (create/overwrite with auto-mkdir), workspace_rm (delete with recursive guard), workspace_mv (move/rename with auto-mkdir), workspace_patch (find/replace operations, each match must be unique, 1MB file limit). Late registration via RegisterWorkspaceTools() after stores + FS init. WorkspaceToolNames() returns the full list for filtering.
• Channel Workspace Binding. channels.workspace_id nullable FK column (Postgres + SQLite migration 010/009). Channels with a bound workspace get workspace tools auto-injected into the completion tool set.
• Project Workspace Binding. projects.workspace_id nullable FK column. Projects can bind a workspace shared by all channels in the project.
• Workspace Resolution Chain. ChannelStore.ResolveWorkspaceID() resolves the effective workspace for a channel: channel workspace_id takes priority, falling back to the project's workspace_id via project_channels join. Both Postgres and SQLite implementations. Returns empty string when no workspace is bound.
• Tool Injection in Completion Handler. buildToolDefs() accepts workspaceID parameter; when empty, all workspace tool names are added to the disabled set. ExecutionContext.WorkspaceID field carries the resolved workspace ID through tool execution and streaming.
• Workspace resolution wired into streaming. Both CompletionHandler and streamWithToolLoop resolve the workspace ID from the channel and pass it through to tool execution contexts.

Deferred

• Frontend workspace UI (channel settings toggle, project Files tab, chat bar workspace indicator) deferred to a future release.
• Archive-to-workspace upload flow (detect archive MIME + workspace binding → extract to workspace) deferred.

[0.21.0] — 2026-03-01

Added

• Workspace Storage Primitive. Platform-level file storage bound to users, projects, channels, or teams via polymorphic owner model. Dual-layer architecture: PVC filesystem (source of truth) with DB metadata index (queryable cache). Workspaces support configurable quotas, status lifecycle (active/archived/deleting), and owner-based authorization inheritance.
• Workspace data model. Two new tables: workspaces (polymorphic owner_type/owner_id, root_path, max_bytes quota, status) and workspace_files (path, MIME type, size, sha256, is_directory). Unique index on (workspace_id, path) enables upsert-on-conflict for file metadata sync. Migrations for both Postgres and SQLite.
• workspace.FS package. Filesystem operations layer with security-first design: atomic writes (temp file + rename with SHA256 computed via tee reader), path traversal guards (cleanPath normalization + absPath containment validation), symlink rejection on write targets. Operations: ReadFile, WriteFile, DeleteFile (with recursive guard), Mkdir, Stat, ListDir, Tree, Reconcile (filesystem→DB drift sync).
• Archive operations. Extract zip and tar.gz archives into workspaces with bomb protection: 10K file limit, 100MB single file cap, workspace quota enforcement during extraction. Common-prefix stripping handles GitHub-style project-name/ wrapper directories. CreateArchive packages workspaces into downloadable zip or tar.gz.
• Content type detection. Extension-based MIME detection covering 40+ source code types (Go, Rust, Python, TypeScript, etc.) with http.DetectContentType sniffing fallback for unknown extensions.
• WorkspaceStore interface. Full CRUD for workspaces, file index operations (upsert, delete, delete-by-prefix, get, list with recursive/non-recursive modes), ownership lookup (GetByOwner, ListByOwner), and aggregate stats. Postgres and SQLite implementations.
• Workspace API. 15 new endpoints under /api/v1/workspaces: workspace CRUD (create, get, update, delete), file operations (list, read, write, delete, mkdir), archive management (upload with extraction, download), reconcile (FS→DB sync), stats. Owner-based authorization: user workspaces require self, channel workspaces require channel owner, project workspaces require project member, team workspaces require team member.
• Unit tests. Path cleaning (13 cases including traversal, dotfiles, whitespace), absPath traversal detection, MIME detection (14 extensions), unsafe path filtering (7 cases), common prefix detection (5 cases), write/read round-trip with mock store, delete verification, mkdir with index sync.

[0.20.0] — 2026-03-01

Added

• Notifications Core (Phase 1): Persistent, user-targeted notification infrastructure with real-time WebSocket delivery. notifications table (Postgres + SQLite), NotificationStore with paginated queries, Service.Notify() / NotifyMany() for centralized creation and dispatch. Five API endpoints: list (paginated, filterable), unread count, mark read, mark all read, delete. Bell icon in header bar with unread count badge (capped at 9+). Notification dropdown (latest 10, grouped, click-to-navigate via resource_type/resource_id). Full notification panel registered with PanelRegistry. WebSocket push via notification.new event with toast for high-priority types (kb.error, role.fallback). Background cleanup goroutine (configurable retention, default 90 days). Initial sources: role.fallback (via EventBus subscription), kb.ready/kb.error (knowledge base processing), grant.changed (group membership).
• @mention Parsing + Multi-model Routing (Phase 2): Channels support multiple AI models with @mention-based routing. mentions.Parse() extracts @mentions from message content, resolves against channel model roster (case-insensitive, longest-match-first, trailing punctuation tolerant). Completion handler fans out sequentially to mentioned model(s), producing one assistant response per target. Without @mention, default channel model responds (fully backward compatible). Channel model CRUD: 4 new endpoints for add/remove/update/list. Frontend: model pills in chat header, @mention autocomplete (CM6 mentionCompletion extension with roster-backed suggestions), model attribution labels on multi-model responses. Message model_display field for human-readable attribution. SSE streaming tagged with model info per response.
• Email Transport + Notification Preferences (Phase 3): SMTP email delivery via EmailTransport supporting implicit TLS (port 465) and STARTTLS (port 587). Multipart MIME messages (HTML + plaintext) with branded templates using Go html/template. notification_preferences table with three-tier resolution: specific type → user wildcard * → system default (in_app=true, email=false). Three preference API endpoints (list, set with partial patch, delete). Admin SMTP configuration in settings (host, port, user, password, from address, TLS mode) with test email endpoint. User notification preferences UI in Settings → Notifications tab with per-type in-app/email checkboxes. Async email delivery (goroutine with 30s timeout, failures logged non-blocking).
• Gin Release Mode: Backend now automatically sets gin.SetMode(gin.ReleaseMode) when ENVIRONMENT=production, suppressing per-request access logs for health checks and reducing log noise. GIN_MODE env var also added to K8s backend deployment as explicit override.

Fixed

• Model preferences 500 on every page load: GET /api/v1/models/preferences returned HTTP 500 due to NULL values in user_model_settings.hidden and sort_order columns failing Go sql.Scan() into non-pointer types. Root cause: the Set upsert passed NULL for unset patch fields, bypassing DEFAULT false/DEFAULT 0 on INSERT. Fixed with COALESCE in both SELECT (read path) and INSERT VALUES (write path) for Postgres and SQLite. Includes data-fix SQL for existing NULL rows.

[0.19.2] — 2026-02-28

Added

• Project Persona Default: Bind a persona to a project via the detail panel. All chats in the project inherit the persona's model, parameters, and system prompt as a fallback when no explicit preset is selected. Resolution chain: explicit request → project persona → none.
• Project Archive Toggle: Archive/unarchive projects from the detail panel. Archived projects hidden from sidebar by default with "Show archived (N)" toggle. Dimmed visual treatment when shown.
• Channel Reorder: Right-click a chat within a project for "Move up" / "Move down" options. Server-persisted positions via project_channels.position, loaded on startup, maintained across moves.

[0.19.1] — 2026-02-28

Added

• Active Project: Pin a project as active; new chats auto-assign to it. Persists across reloads via localStorage. Visual indicator (📌 + accent border) in sidebar.
• Project System Prompt: Per-project instructions stored in projects.settings JSONB. Injected between persona/channel prompt and KB hint during completion. Merge semantics on update (preserves other settings keys).
• Project Detail Panel: Side panel (via PanelRegistry) for managing project system prompt, KB bindings, and notes. Accessible from project ⋯ menu → "Project settings".
• Enriched ListKBs and ListNotes responses with JOIN-sourced name/title fields for display in the project panel.

Changed

• ProjectPatch model now accepts settings field for partial settings merge.
• Project context menu expanded: "Pin as active", "Project settings", plus existing rename/color/delete.

[0.19.0] — 2026-02-28

Added

• Projects / Workspaces. Organizational containers that group related conversations, knowledge bases, and notes into a single workspace. Projects provide a scope-aware organizational layer above individual channels, with support for personal, team, and global visibility.
• Project data model. Four new tables: projects (with scope, owner, team, color, icon, settings JSONB), project_channels (ordered membership with UNIQUE constraint), project_knowledge_bases (with auto_search flag), and project_notes. Channels gain a denormalized project_id FK for efficient filtering.
• Project API. 17 new endpoints under /api/v1/projects: full CRUD, channel add/remove/list/reorder, KB add/remove/list, note add/remove/list. Access checks enforce owner-or-team-member visibility with owner-only delete.
• Project KB resolution. Knowledge bases bound to a project are automatically available to all channels within that project. Resolution chain extended: Persona KBs → Project KBs → Channel KBs → Personal KBs. Both BuildKBHint (system prompt injection) and kbsearch (tool-time retrieval) updated.
• Note auto-association. Notes created from a channel that belongs to a project are automatically added to that project's note collection.
• Sidebar project groups. Projects appear as collapsible groups in the sidebar above the existing time-based "Recent" section. Each group shows a color dot, chat count, and an options menu (⋯) for rename, color picker, and delete.
• Drag-and-drop. Drag chat items between project groups or back to Recent. Visual feedback with outline and background highlight on valid drop targets.
• Right-click context menu. Right-click any chat to move it to a project, remove it from its current project, or create a new project and assign in one step.
• New Project button. Added to the New Chat split-button dropdown for quick project creation.
• Channel project_id filter. GET /api/v1/channels accepts ?project_id=<uuid> to filter by project, or ?project_id=none for unassigned channels. Response includes project_id field.

Changed

• renderChatList() rewritten to support project grouping while preserving the original time-based layout when no projects exist. Chat items now include draggable, oncontextmenu, and drag event handlers.
• Channel response struct, SELECT queries, and scan calls updated across ListChannels, GetChannel, and CreateChannel (both Postgres and SQLite paths) to include project_id.
• resource_grants CHECK constraint extended to include 'project' as a valid resource type.
• db-validate.sh updated: former "Dropped tables" assertions for projects and project_channels replaced with positive checks for all four project tables plus channels.project_id column check.
• Test helper truncation list updated with project junction tables.

Fixed

• JSON corruption defense (hotfix carry-forward from 0.18.2): SafeJSON wrapper with scanJSON and scanTags hardened helpers that replace bare json.RawMessage / pq.Array scanning. Prevents encoding/json: invalid character panics from NULL or empty columns.
• Service worker (hotfix carry-forward): chrome-extension:// URL filtering to avoid opaque response cache errors.

[0.18.1] — 2026-02-28

Added

• Side panel architecture. Complete rewrite of the side panel system from shared-tab layout to independent single-slot panels. Any action (preview, notes, diagram pop-out) fills the slot, replacing whatever was there — no tabs, no association between panel types.
• Panel registry (PanelRegistry): named panels with independent open/close state, scroll/state preservation across switches, keyboard shortcuts (Ctrl+\ cycle, Ctrl+Shift+\ toggle dual-view).
• Dual-view mode: two panels side-by-side via CSS grid with a drag-adjustable split ratio (0.2–0.8 range, 6px divider handle). Toggle button in header actions area alongside fullscreen and close.
• Live HTML preview: streaming responses update the preview iframe in real-time (500ms debounce). Extracts the last fenced HTML block from partial content and renders it as the response streams in.
• Extension pop-out: ⧉ button on rendered extension blocks (mermaid, KaTeX, etc.) clones the content into the preview iframe and opens the side panel.
• Extension UI primitives (ctx.ui): seven methods exposed to browser extensions through the scoped extension context:
  • toast(msg, type) — toast notifications via UI.toast()
  • openPreview(html) — load HTML into side panel preview iframe
  • isDark() — theme detection without DOM sniffing
  • isMobile() — viewport width check (≤768px)
  • isPanelOpen() — side panel container visibility
  • confirm(msg, opts) — modal confirm dialog (Promise<boolean>)
  • createMenu(anchor, opts) — popup menu (unchanged from stub)
• Mermaid context-aware expand: single ⛶ button replaces the previous two-button (pop-out + fullscreen) approach. When the side panel is open, expand pops the diagram into it; when closed, expand goes fullscreen.
• Mermaid fullscreen close button: 40px circular close button overlaid top-right in fullscreen mode, 48px on mobile. Fixes the previous Escape-key-only exit which was unusable on touch devices.
• Mobile side panel: swipe navigation between panels (80px threshold, 1.5× horizontal-to-vertical ratio), tap-to-close overlay, responsive auto-collapse of dual mode on narrow viewports, enlarged touch targets for all panel controls.
• Side panel header label: simple text label showing the active panel name, replacing the previous tab bar UI.

Changed

• Side panel model changed from tabbed (Preview + Notes tabs visible simultaneously) to single-slot (one panel fills the space, actions replace it). No user-selectable tabs — content is entirely action-driven.
• Dual-view toggle button moved from tab bar (dynamically injected) to static header actions area next to fullscreen and close.
• Mermaid extension refactored to use ctx.ui primitives exclusively. Zero direct references to UI.*, PanelRegistry.*, or DOM class sniffing for theme detection. Extension remains fully self-contained in extensions/builtin/mermaid-renderer/.
• Mermaid source copy uses ctx.ui.toast() instead of inline button text swap. Theme detection uses ctx.ui.isDark() instead of manual document.body.classList.contains('dark-theme') check.
• Side panel outer resize minimum bumped to 480px in dual mode (vs 280px single).

Removed

• Tab bar UI (.side-panel-tabs, .side-panel-tab, tab rendering logic). Panels no longer have user-selectable tabs.
• Separate pop-out and fullscreen buttons in mermaid toolbar (collapsed into single context-aware expand button).

[0.18.0] — 2026-02-28

Added

• Memory system. Long-term memory across conversations with scope-aware isolation. Three memory scopes: user (personal facts/preferences), persona (shared across all users of a Persona), and persona_user (per-user within a Persona context, e.g. tutoring progress per student). Memories persist across channels and are injected into the system prompt at completion time with scope priority (persona_user > persona > user).
• Memory tools. Two new LLM-callable tools:
  • memory_save — LLM explicitly stores a fact with key/value/confidence. Scope-aware: saves to the active scope for the current Persona context.
  • memory_recall — LLM queries stored facts relevant to current context. Merges results from applicable scopes with semantic search via embeddings and keyword fallback.
• Automatic memory extraction. Background scanner finds conversations with sufficient new activity, sends them to the utility model role for fact extraction, and stores results as pending_review memories. Configurable extraction prompt per Persona (e.g. "extract FAQ-worthy Q&A pairs"). Scanner runs on a configurable interval with concurrency control. Global kill switch via admin settings.
• Memory review pipeline. Extracted memories start in pending_review status. Admin panel shows pending review queue with bulk approve. Users can approve/reject/edit their own pending memories from the Settings → Memory tab.
• User memory management. Settings → Memory tab shows active/pending memory counts, filterable/searchable memory list with inline edit and delete. Status filter (active, pending_review, archived). Approve All button for batch operations on pending memories.
• Admin memory controls. Admin panel → Memory section shows system-wide pending review queue. Memory extraction toggle in admin Settings panel (memory_extraction_enabled). Bulk approve endpoint for batch review.
• Hybrid semantic recall. Memory recall supports both keyword search and vector similarity (cosine distance). Postgres uses <=> operator with pgvector; SQLite computes cosine similarity in Go with app-level vector loading. Results are merged and deduplicated.
• Memory injection at completion time. BuildMemoryHint() loads relevant memories and formats them as a system prompt section, injected after the knowledge base hint. Context-budget aware with configurable character limit. Supports embedding-based relevance filtering when the user's latest message is available.
• Persona memory configuration. Personas gain memory_enabled (bool) and memory_extraction_prompt (text) fields. When memory is enabled on a Persona, conversations with that Persona contribute to persona-scoped memory extraction. Custom extraction prompts allow specialization (e.g. helpdesk FAQ extraction vs. tutoring progress tracking).
• Database migrations. Four new migration files (Postgres + SQLite):
  • 004_v0180_memories.sql / sqlite/003_v0180_memories.sql — memories table with composite unique index, full-text search index (GIN/keyword), memory_extraction_log tracking table.
  • 005_v0180_memory_phase2.sql / sqlite/004_v0180_memory_phase2.sql — Persona memory columns, extraction log unique constraint.
• API endpoints. Eight new authenticated endpoints:
  • GET /memories — list user's memories (filterable by status/query)
  • GET /memories/count — active + pending counts
  • PUT /memories/:id — edit a memory's key/value/confidence
  • DELETE /memories/:id — delete a memory
  • POST /memories/:id/approve — approve a pending memory
  • POST /memories/:id/reject — reject (archive) a pending memory
  • GET /admin/memories/pending — admin pending review queue
  • POST /admin/memories/bulk-approve — admin bulk approve

Changed

• CompletionHandler now accepts an *knowledge.Embedder parameter for memory injection with semantic relevance filtering.
• Persona create/update forms include memory configuration fields (enabled toggle, extraction prompt textarea).
• Admin settings save handler includes memory_extraction and memory_extraction_enabled keys.
• store.Stores struct includes Memories MemoryStore field.
• Admin panel sections include Memory with pending review loader.

Technical Notes

• Memory embeddings use vector(3072) matching the existing KB/notes schema. HNSW index is not used (pgvector limits HNSW to 2000 dims); filtered sequential scans are performant for per-user memory tables. IVFFlat or dimension reduction available as future optimizations.
• SQLite hybrid recall loads embeddings into Go and computes cosine similarity at the application level, reusing the existing cosineSimilarity() function from the knowledge base store.
• Memory tools use late registration (like KB search and note tools) because they require stores and embedder dependencies initialized in main.go.

[0.17.3] — 2026-02-28

Added

• Wikilink bi-directional linking. Notes support [[Title]] and [[Title|display text]] syntax. Links are extracted on save via regex, resolved to target note IDs by case-insensitive title match, and stored in a note_links junction table. Dangling links (references to notes that don't exist yet) are preserved and automatically resolved when a matching note is created later.
• Transclusion embeds. ![[Title]] syntax renders embedded note content inline in read mode. Content is fetched asynchronously with a recursion guard (max depth 1 — nested transclusions render as plain text references). Transclusion links are visually distinct in both the editor (border-left, italic) and the graph (dashed edges).
• Backlinks panel. Read mode shows a collapsible "Linked mentions" panel below the note content listing all notes that link to the current note. Each backlink is clickable to navigate directly. Count badge updates on every note open.
• Knowledge graph visualization. Canvas-based force-directed graph with no external dependencies (~480 lines). Physics: O(n²) Coulomb repulsion, Hooke spring attraction on edges, center gravity. Interaction: pan, zoom (0.15–4.0x toward cursor), drag nodes, hover highlights node + neighbors. Click opens note editor. Nodes sized by √(link_count), colored by folder path (10-color palette). Ghost nodes for unresolved [[links]] with toggle button. Energy-based pause (stops rAF when total kinetic energy falls below threshold). ResizeObserver for responsive canvas sizing.
• CM6 note editor. New CM.noteEditor() factory function with live markdown preview (heading sizes for h1–h3, blockquote styling, fenced code block decorations), [[wikilink]] autocomplete triggered by [[ with async title search, and clickable wikilink chip rendering. Falls back to plain <textarea> if CM6 bundle is unavailable.
• Wikilink autocomplete. Typing [[ in the note editor triggers an async completion popup backed by GET /notes/search-titles?q=. Results are fetched via ILIKE title match, limited to 8 suggestions. Selecting a result inserts the full [[Title]] syntax.
• Daily notes. "Today" button in the notes toolbar creates or opens a daily note titled Daily — YYYY-MM-DD in the /daily/ folder with a starter template (Tasks + Notes sections). Idempotent — repeated clicks navigate to the existing daily note.
• Save-to-note from chat. "Note" button in message action bar captures the full message content (or the current text selection within that message) into a new note with pre-filled title extracted from the first line. Provenance is tracked via source_message_id column on the notes table, enabling future jump-to-source navigation.
• API endpoints. Three new authenticated endpoints:
  • GET /notes/search-titles?q=&limit= — lightweight title search for autocomplete (returns [{id, title}]).
  • GET /notes/:id/backlinks — notes linking to the specified note, with display text and metadata.
  • GET /notes/graph — full graph topology: nodes (with inbound/outbound link counts), resolved edges, and unresolved (dangling) link titles.
• note_links table. New junction table with source_note_id, target_note_id (nullable for dangling links), target_title, display_text, is_transclusion, and created_at. Primary key on (source_note_id, target_title). Index on target_note_id for backlink queries. Migrations for both Postgres and SQLite.
• source_message_id column. New nullable UUID column on notes table for chat-to-note provenance tracking.
• Wikilink extraction package. Standalone notelinks/ package with regex-based parser handling [[Title]], [[Title|Display]], ![[Title]], and ![[Title|Display]]. Deduplicates by lowercase title, preserves first occurrence. 10 unit tests covering edge cases.

Changed

• Notes editor replaces <textarea> with CM6 noteEditor instance. The container #noteEditorContentContainer is lazily initialized on first edit; destroyed on panel close to avoid stale state.
• Note create and update handlers now extract wikilinks from content and call ReplaceLinks() (transactional DELETE + batch INSERT). Create also calls ResolveByTitle() to fix dangling links from other notes.
• showNotesList() destroys the CM6 editor instance and hides graph view.
• saveNote() and deleteNote() call invalidateNoteGraph() to clear the cached graph data.
• Read mode renders [[wikilinks]] as styled clickable chips via post- processing of the formatted HTML. Clicking navigates to the linked note or offers to create it if not found.
• ARCHITECTURE.md updated with notelinks/ package, note-graph.js, CM.noteEditor() factory, and expanded notes description covering the linking model.
• ROADMAP.md updated with full v0.17.3 section including all checklist items.
• CM6 bundle entrypoint (index.mjs) exports noteEditor alongside existing chatInput and codeEditor.
• theme.mjs adds noteEditorTheme with full-height layout (min 200px, max 60vh), heading size decorations, and blockquote styling.

[0.17.2] — 2026-02-28

Added

• CodeMirror 6 integration. Rich editor infrastructure compiled at Docker build time via esbuild (IIFE bundle, ~295KB min / ~90KB gzip). Two factory functions exposed on window.CM:
  • CM.chatInput() — Markdown-mode editor for the chat input with auto-growing height, Enter=send / Shift+Enter=newline, spell check, and WYSIWYG fenced code block decorations (visual container with monospace font and accent border, matching claude.ai UX).
  • CM.codeEditor() — Full-featured code editor for admin extension panel with line numbers, bracket matching, search/replace, fold gutter, and 10 bundled language modes (Markdown, JavaScript, JSON, SQL, HTML, CSS, YAML, Go, Python, Rust).
• Graceful degradation. All CM6 integration points check window.CM availability. If the bundle fails to load, the app falls back to native <textarea> with zero breakage.
• Dark/Light/System theme toggle. New appearance setting with three modes. Light theme overrides all CSS variables via [data-theme="light"] selector. System mode tracks prefers-color-scheme media query in real time. Theme changes emit theme.changed on the EventBus; CM6 editors toggle oneDark syntax theme via compartment reconfiguration.
• Vim/Emacs keybinding preference. Editor keybinding mode (Standard / Vim / Emacs) configurable in appearance settings. Applies to code editors and extension editors only — chat input always uses standard keybindings. Live-switchable on already-open editors via keymap.changed event. Bundled statically (~40KB for both modes).
• Inline code shortcut. Ctrl/Cmd+E wraps selection in backticks or inserts an empty inline code pair with cursor between them.
• Code block shortcut. Typing ``` at the start of a line expands to a fenced code block with cursor positioned inside. The decoration plugin renders code blocks with a styled visual container in the chat input.
• CI path-based change detection. New detect-changes job classifies changed files into frontend/backend/infra/docs buckets. Test jobs skip when irrelevant (FE-only changes skip Go tests, docs-only changes skip all tests and deploy). Tags always run the full pipeline.

Changed

• Docker build pipeline: both Dockerfile.frontend and unified Dockerfile now include a cm6-build stage (Node 20 Alpine → esbuild → IIFE bundle).
• ChatInput abstraction in chat.js replaces direct textarea access across 7 callsites (chat.js, tokens.js, attachments.js).
• Extension editor in admin-handlers.js uses CM.codeEditor() with JSON and JavaScript modes, replacing bare <textarea> with manual Tab handler.
• Service worker excludes /vendor/codemirror/ from cache (version-busted).
• Debug state snapshot includes CM6 version and language list.
• ARCHITECTURE.md updated to v0.17 reflecting CM6 integration, SQLite dual-driver, modular frontend file structure, and theme system.
• CI pipeline header updated to v0.17.2 with path gating documentation.
• build-editor.sh falls back to npm install if package-lock.json is missing (belt-and-suspenders for local dev).

Fixed

• App initialization crash. Events.publish (nonexistent) → Events.emit (correct API). The unhandled exception during initAppearance() killed initListeners(), leaving the entire app half-initialized — settings modal unclosable, keyboard shortcuts unwired, paste handlers missing.
• Cursor invisible in dark and light mode. CM6 cursor used --accent color (low contrast on both themes). Switched to --text for consistent visibility. Added explicit borderLeftWidth: 2px.
• Placeholder text offset. Double padding between .cm-editor wrapper (12px) and .cm-content (8px) pushed placeholder 20px below expected position. Zeroed .cm-content padding — wrapper is the single source of truth.

[0.17.1] — 2026-02-27

Added

• SQLite backend. Full dual-driver database layer — set DB_DRIVER=sqlite to run with an embedded SQLite database. Pure Go (no CGO), zero external dependencies. 19 store files covering all domain stores: channels, messages, users, teams, personas, knowledge bases, notes, usage, audit, extensions, and more. Feature parity with Postgres including knowledge base vector search via app-level cosine similarity computed in Go.
• Dialect-aware test infrastructure. database.SetupTestDB() detects DB_DRIVER and provisions either a Postgres test database or a SQLite temp file. Exported database.PH(n) returns $N or ? per dialect. database.TruncateAll() uses TRUNCATE CASCADE on Postgres and DELETE FROM with PRAGMA foreign_keys toggling on SQLite. dialectSQL() helper in handler tests converts $N placeholders and strips ::jsonb casts at runtime.
• SQLite CI pipeline. New test-sqlite job runs the full handler integration test suite and store tests against an embedded SQLite database. Parallel with the existing Postgres test job. Build gate verifies CGO_ENABLED=0 compilation.
• Generic provider test config. Live provider integration tests now read PROVIDER, PROVIDER_KEY, and PROVIDER_URL environment variables instead of hardcoded VENICE_API_KEY. Legacy fallback preserved. Model selection prefers non-reasoning models to avoid thinking budget requirements with low max_tokens.

Changed

• kb_chunks table in SQLite schema includes embedding TEXT column for JSON-encoded float64 vectors (previously omitted as feature-gated).
• SimilaritySearch on SQLite loads candidate chunks, decodes JSON embeddings, and computes cosine distance in Go — replacing the previous "not available" error.
• InsertChunks on SQLite now stores embedding vectors as JSON text.
• Live test names genericized: TestLive_Venice* → TestLive_*.
• CI build-and-deploy depends on [test, test-frontend, test-sqlite].

Fixed

• SQLite RETURNING + time.Time scan failure. The modernc/sqlite driver cannot scan datetime('now') TEXT columns into time.Time via RETURNING. All 13 SQLite store Create methods rewritten to set timestamps in Go (time.Now().UTC()) and use ExecContext instead of QueryRowContext(...).Scan(). Format: 2006-01-02 15:04:05 (timeFmt constant in helpers.go).
• SQLite missing id in INSERTs. Unlike Postgres (DEFAULT gen_random_uuid()), SQLite TEXT PRIMARY KEY columns have no auto-generation. Added store.NewID() / uuid.New() to: usage_log, model_pricing (both upsert paths), team_members (AddMember), group_members (AddMember), refresh_tokens (CreateRefreshToken).
• Test seed helpers SQLite-aware. SeedTestUser, SeedTestChannel, SeedTestTeam, SeedTestTeamMember, SeedTestGroup, SeedGroupMember now branch on IsSQLite() to provide application-generated UUIDs instead of relying on RETURNING id.
• Postgres-isms in SQLite stores. extension.go Update used now() → datetime('now'); ListForUser COALESCE used true → 1.
• SQLite store bool/int type mismatches: persona.go auto-fetch flag, usage.go exclude-BYOK filter, user_settings.go visibility map values — all corrected from 0/1 to false/true.

[0.17.0] — 2026-02-27

Added

• Persona-KB binding. Personas can now have knowledge bases directly bound to them. When a user selects a persona with bound KBs, the kb_search tool is automatically scoped to those KBs and the persona's system prompt includes a KB listing hint. Admin and team admin preset forms include a KB picker with per-KB auto-search toggles. Migration adds persona_knowledge_bases join table.
• Enterprise KB mode. New discoverable flag on knowledge bases controls whether users can see and attach KBs directly. When kb_direct_access platform policy is set to true, users cannot attach KBs to channels directly — they access KBs exclusively through persona bindings curated by admins. New ListDiscoverableKBs and SetDiscoverable endpoints.
• Role fallback alerts (issue #69). When a role's primary provider fails and the fallback activates, the resolver emits a role.fallback event on the EventBus with audit log entry. Admin users see a persistent dismissable banner. 5-minute per-role cooldown prevents flooding.
• Chat rename. Double-click any chat title in the sidebar to edit inline. Enter saves, Escape cancels.
• Utility model auto-naming. After the first assistant response, a background request to the utility role generates a concise title. Falls back to truncation when no utility model is configured. New endpoint: POST /channels/:id/generate-title.
• Chat token count. Conversation token estimate shown in the model bar, color-coded against context budget.
• State restore on refresh. Active chat ID persisted to sessionStorage, auto-restored on page reload.

Fixed

• ResolvePreset group access bypass. ResolvePreset() used raw SQL that skipped resource_grants checks from v0.16.0. Now uses PersonaStore.UserCanAccess().
• KB create scope authorization. Now enforces admin role for global KBs and team admin role for team KBs.
• Completion handler persona ID scoping. personaID variable scoped inside an if block but referenced downstream. Fixed by threading through function signatures.
• Nil slice JSON marshaling. ListDiscoverableKBs returns [] instead of null when no KBs match.

Changed

• UpdateDocumentStorageKey moved from raw ExecContext to store method.
• EventBus route table: role.fallback → DirToClient.
• Resolver gains .WithBus(bus) builder (nil-safe, backwards compatible).
• Service worker: /extensions/ excluded from fetch handler.
• Mermaid renderer v2.0 (pan/zoom, export) promoted to builtin.
• Removed DIAG diagnostics from TestGroupBasedPersonaAccess.
• Paste-to-file threshold synced from backend PublicSettings (storage.paste_to_file_chars, admin-configurable, default 2000). Resolves hardcoded constant in attachments.js.
• Embedding dropdown already had tolerant type filter, manual model ID fallback, and auto-switch on empty — confirmed complete, checkbox updated.

[0.16.0] — 2026-02-27

Added

• User groups. Global and team-scoped groups decoupled from team membership. groups and group_members tables. Admin and team admin CRUD for group management. Groups serve as ACL targets for resources.
• Resource grants. Three-way grant model (team_only, global, groups) for Personas and Knowledge Bases. resource_grants table with grant_scope and granted_groups UUID[] columns. Grant picker UI on Persona and KB forms with group multi-select.
• Schema consolidation. 9 incremental migrations collapsed into single 001_v016_schema.sql. Fresh installs use the consolidated file; upgrade path preserved via migration version tracking.

Changed

• Persona and KB list queries now filter through resource_grants for non-admin users. Team-only remains the default scope.
• Admin panel shows group membership counts and grant summaries.

[0.15.1] — 2026-02-26

Added

• attachment_recall tool. Two operations: list returns filenames and metadata for the current channel's attachments; read extracts and returns content by attachment ID. Channel-scoped access control.
• conversation_search tool. Full-text search across the current channel's message history using PostgreSQL plainto_tsquery. Returns matching messages with timestamps and role context.
• Token estimator attachment awareness. Tokens.estimateAttachments() accounts for staged file sizes in context budget calculations. Warning thresholds include attachment estimates.

[0.15.0] — 2026-02-26

Added

• Background compaction scanner. Periodic scan identifies channels exceeding configurable context thresholds. Automatic summarization via utility role compresses old messages into summary nodes. Per-channel opt-in/out via auto_compact channel setting.
• Compaction service. compaction.Service orchestrates summary generation: estimates token usage, selects messages for compression, calls utility role, inserts summary as tree boundary node with metadata.
• Context budget guard rail. 80% ceiling prevents compaction from triggering mid-generation. Cooldown timer prevents repeated compaction of the same channel.
• Summarize & Continue button. User-triggered compaction from the context warning bar. Reuses compaction service with immediate execution.

Changed

• Scanner configurable via global_settings: threshold percentage, cooldown duration, enabled/disabled toggle. Channel-level overrides.

[0.14.0] — 2026-02-26

Added

• Knowledge bases. RAG pipeline: upload documents → chunk (recursive text splitter) → embed via pgvector → kb_search tool for semantic retrieval. Team and personal KB scopes. Channel KB toggle enables per-conversation knowledge access.
• Document ingestion. knowledge.Ingest() pipeline: file upload → text extraction (reuses v0.12.0 pipeline) → chunking with configurable overlap → embedding via the embedding role → storage as kb_chunks with vector index.
• KB admin panel. Knowledge Bases section under AI category. Create, delete, upload documents, view chunk counts and storage usage. Team admin scoped to team KBs.
• Notes semantic search. Note search upgraded from exact text match to pgvector cosine similarity when embeddings are available.
• kb_search tool. Registered when embedding role is configured. Accepts query text, returns top-K chunks with source document attribution. Channel KB bindings control which KBs are searched.

Changed

• pgvector extension enabled in schema (CREATE EXTENSION IF NOT EXISTS vector). kb_chunks table includes embedding vector(1536) column with IVFFlat index.

[0.13.1] — 2026-02-26

Added

• web_search tool. Search provider abstraction with two backends: DuckDuckGo (HTML scraping, zero config) and SearXNG (self-hosted, JSON API). Returns title, URL, and snippet for each result. Configured via SEARCH_PROVIDER and SEARXNG_URL env vars.
• url_fetch tool. Fetches and extracts text content from URLs. Respects robots.txt. Content-type detection with HTML-to-text conversion. Configurable timeout and size limits.
• Tool categories. Tools now have a category field (builtin, search, knowledge, browser). Tools toggle UI in chat bar groups by category with per-tool enable/disable.
• Tools toggle UI. Popup menu on chat input toolbar showing all available tools. Per-tool checkbox state sent as disabled_tools[] in completion requests. Browser extension tools included.

[0.13.0] — 2026-02-25

Changed

• Admin panel refactor. Replaced 12-tab modal with fullscreen admin panel. Four categories (People, AI, System, Monitoring) with section sidebar navigation. URL-based routing (#admin/people/users). Responsive layout, classification banner-aware positioning.
• CSS design token cleanup. Consolidated duplicate color/spacing variables. Admin panel uses shared token system with main UI.

[0.12.0] — 2026-02-25

Added

• File handling and vision support. Upload images, PDFs, and documents into chat via 📎 button, drag-and-drop, or paste. Multimodal message assembly injects base64 images for vision-capable models and extracted text for documents. Staged attachment strip shows upload progress and extraction status. Auth-aware blob rendering with Bearer tokens. Image lightbox viewer. Vision capability hints on image attachments.
• Storage backend abstraction. ObjectStore interface with two implementations: PVC (local filesystem) for single-node/dev and S3 (minio-go) for multi-node production. S3 backend works with any S3-compatible API: MinIO, Ceph RGW, AWS S3. Auto-detection when STORAGE_BACKEND is not set (tries PVC, disables if not writable). Both backends share identical semantics — all handlers, attachment CRUD, multimodal assembly, and orphan cleanup work regardless of backend.
• S3 configuration. S3_ENDPOINT, S3_BUCKET, S3_ACCESS_KEY, S3_SECRET_KEY, S3_REGION, S3_PREFIX, S3_FORCE_PATH_STYLE env vars. Endpoint auto-detects SSL from scheme. Path-style URLs default on (required for MinIO/Ceph). Optional key prefix for shared buckets. CI pipeline syncs S3 secrets to K8s when STORAGE_BACKEND=s3.
• Text extraction pipeline. PDF, DOCX, XLSX, PPTX, ODT, RTF text extraction via filesystem-based queue. Inline and sidecar modes. Crash recovery for items stuck in processing state.
• Admin storage panel. Backend health, file count, total size, orphan detection and cleanup. Shows PVC path or S3 endpoint/bucket depending on active backend. Extraction queue status.
• Vault CLI commands. switchboard vault rekey re-encrypts all provider API keys when rotating the encryption key. switchboard vault status shows encryption health. Admin UI encryption status indicator in Settings tab.
• Per-chat model persistence. Server-side channels.settings JSONB field stores last_selector_id. Roams across devices with localStorage write-through cache as fallback.

[0.11.0] — 2026-02-25

Added

• Browser extension system. Full lifecycle: manifest registration, script injection, scoped context (ctx.renderers, ctx.tools, ctx.events, ctx.storage, ctx.ui), permission-aware API proxying. Extensions self- register via Extensions.register() and receive isolated contexts during initAll(). Admin CRUD endpoints, asset serving (public, no auth needed for <script> tag loading), auto-seeder for builtin extensions.
• Custom renderer pipeline. Block renderers intercept code fences by language tag, post renderers process the DOM after insertion. Case- insensitive language matching handles LLM capitalization (Mermaid, Diff, CSV). Nested ```markdown ``` fence unwrapping prevents the common LLM pattern of wrapping responses in markdown fences from breaking inner code blocks.
• Browser tool bridge. Extensions register tool handlers via ctx.tools.handle(). Server collects browser tool schemas alongside server tools, includes them in LLM completions. Tool calls route through WebSocket EventBus (tool.call.* → browser → tool.result.* → server). 30-second timeout with error recovery.
• Server tools: calculator and datetime. Auto-register via init(), zero wiring changes. Calculator: recursive-descent evaluator with arithmetic, 17 math functions, constants. Datetime: current date/time/ timezone/unix/ISO-week in any IANA timezone.
• 6 built-in browser extensions (self-contained, own CSS via _injectStyles()/destroy() lifecycle):
  • Mermaid: block + post renderer, SVG diagrams, dark mode detection, local vendor with CDN fallback, max-height constraint with scroll
  • KaTeX: block renderer (```latex/math/tex ```) + post renderer (inline$...and...$` in text nodes)
  • CSV Table: block renderer, RFC 4180 parser with quoted fields, sortable columns (click headers), numeric-aware sorting
  • Diff Viewer: block renderer, red/green syntax highlighting for unified diffs, stats badge (+N/-N), hunk/file headers
  • JS Sandbox: browser tool (js_eval), sandboxed iframe (allow-scripts only), console capture, 10s timeout
  • Regex Tester: browser tool (regex_test), multi-input matching, full match details with named groups and indices
• Admin extension editor. Edit button on each extension in Admin → Extensions. Inline editor shows name, description, manifest JSON, and script source with tab-key support. System extensions show overwrite warning.
• Enhanced diagnostics. Test 5: browser extensions (loaded, active, renderers, tool handlers). Test 6: Service Worker cache (registration, scope, state, cache names, entry counts). Purge Cache button in Debug Log modal footer.
• Extension-owned styles. All 4 rendering extensions (mermaid, katex, csv, diff) inject their own <style> tags during init() and remove them during destroy(). Global stylesheet only keeps .ext-rendered wrapper. User-created extensions can bring their own CSS without touching the global stylesheet.
• Notes extension rendering. runExtensionPostRender() called in both note read mode and preview mode. Mermaid diagrams and KaTeX math now render correctly in notes.
• Database migration 006_extensions.sql. extensions and extension_user_settings tables.

Fixed

• WebSocket token field mismatch. API saved { accessToken: '...' } but EventBus read tokens.access. Tool bridge was dead on arrival.
• Extension asset 401. Asset route was behind auth middleware, but <script> tags don't send Authorization headers. Moved to public group.
• runExtensionPostRender is not defined. Stale Service Worker cache served old ui-format.js without the function. Added typeof guard.
• Extension init order. Extensions loaded after loadChats() — block renderers not registered when messages first rendered. Moved extension loading before chat loading.
• K8s Ingress WebSocket routing. Traefik pathType: Exact doesn't reliably win over Prefix rules in the same Ingress resource. Changed /ws and /health to Prefix for longest-prefix-wins routing.
• Mermaid SVG oversized. Added max-height: 600px on diagram container with overflow scroll, removed mermaid.js hardcoded height attribute from SVGs so CSS constraints apply.

Changed

• Extension CSS removed from global styles.css. Each extension owns its styles via _injectStyles() in init() with destroy() cleanup. Idempotent injection guards prevent duplicates.
• Markdown fence language extraction lowercased at the point of extraction in ui-format.js, making all block renderer pattern matches case-insensitive without per-extension workarounds.
• _unwrapMarkdownFence() pre-processor strips outer ```markdown ``` wrappers when they contain nested fences. Handles think-block placeholders and trailing explanation text. Only triggers when nested fences are present — plain markdown code blocks still render normally.

[0.10.5] — 2026-02-24

Added

• ui-primitives.js — shared rendering primitives and registries. Single source of truth for provider types, role definitions, and reusable UI components. Extension-ready via registry pattern (Providers.add(), Roles.add()). Primitives follow the renderPresetForm() pattern: (container, options) → control object with getValues/setValues/clear.
  • Providers registry — types, labels, default endpoints (was 5 places → 1)
  • Roles registry — names, type filters, hints (was hardcoded in 2 places → 1)
  • renderCapBadges(caps, opts) — consolidates 3 badge builders (compact + detailed)
  • renderProviderForm(container, opts) — replaces 3 form implementations
  • renderProviderList(container, opts) — replaces 3 list renderers
  • renderRoleConfig(container, opts) — replaces 2 role UIs + 4 handlers
  • renderUsageDashboard(container, opts) — replaces 3 usage renderers

Fixed

• Admin provider form now auto-fills endpoint on type change. Was missing from admin (worked in user BYOK and team forms). Now all three scopes use the same primitive with identical behavior.
• Team provider form consolidated. Was two separate HTML forms (create + edit) with separate listeners. Now a single dual-mode form matching the pattern used by admin and user BYOK scopes.

Changed

• Provider type definitions removed from index.html (2 static <select>s) and settings-handlers.js (1 dynamic build + 2 endpoint maps). All now sourced from Providers registry in ui-primitives.js.
• Provider list rendering uses event delegation instead of inline onclick handlers. Each list returns { refresh, getCache } control handles.
• Role configuration uses data-role-* attributes for event delegation instead of id-based selectors and global onchange handlers.
• Usage dashboards accept options for compact/full mode, custom API calls, and extension-provided extra columns.
• All 16 confirm() calls replaced with showConfirm() — styled modal dialog matching the app design instead of browser-native dialog. Supports danger styling, Escape/Enter keys, click-outside dismiss.
• Sidebar collapse icon now always visible (dimmed) next to the logo, brightens on hover. Previously the icon replaced the logo on hover only. When sidebar is collapsed, only the collapse icon shows (logo hidden).
• New .popup-menu + .popup-menu-item shared CSS base for all dropdown and flyout menus. createPopupMenu(anchor, opts) primitive available for future menu creation with consistent behavior.
• Removed ~360 lines of duplicated code across admin-handlers.js, settings-handlers.js, ui-admin.js, and ui-settings.js.
• Removed ~40 lines of static HTML form markup from index.html.

[0.10.4] — 2026-02-24

Added

• model_type field across the full pipeline. Models now carry a type classification (chat, embedding, image) sourced from provider APIs at sync time — no hardcoded lists. Venice's /v1/models returns type per model; OpenAI-compatible APIs pass through the field when present.
  • New DB column: model_catalog.model_type VARCHAR(20) DEFAULT 'chat'
  • Migration: 005_model_type.sql
  • Propagation: providers.Model.Type → CatalogSyncEntry.ModelType → CatalogEntry.ModelType → UserModel.ModelType → frontend model_type

Fixed

• Admin role save didn't refresh UI. adminSaveRole() showed "✓ Saved" but never called UI.loadAdminRoles(), so dropdowns appeared stale after save. Now reloads the roles panel after a successful save.
• Role model dropdowns showed all models regardless of type. Embedding role showed chat models, utility role showed embedding models. Both admin and user role UIs now filter the model dropdown by model_type:
  • "embedding" role → only model_type === 'embedding' models
  • "utility" role → only model_type === 'chat' models

Changed

• Venice provider now reads the type field from each model in the API response and normalizes it (text → chat, embedding → embedding, image → image).
• OpenAI provider wire type extended with optional type field for OpenAI-compatible APIs that include it.

[0.10.3] — 2026-02-24

Changed

• Frontend refactor: 2 monolith files → 13 domain-scoped files. Split ui.js (2,582 lines) and app.js (2,940 lines) into 13 focused files averaging ~544 lines each. No features added, no functions renamed, no architectural changes. Vanilla JS, no modules, no build step.

  New file structure:

  File	Lines	Domain
  ui-format.js	353	Markdown rendering, esc(), code blocks, side panel
  ui-core.js	974	UI object: sidebar, chat list, messages, streaming, model selector
  ui-settings.js	640	Settings tabs, teams, providers, user preferences
  ui-admin.js	645	Admin tabs, users, roles, usage, teams
  tokens.js	123	Context tracking, token estimation
  notes.js	364	Notes panel, editor, multi-select
  chat.js	584	Chat ops, send, regen, edit, branch, summarize
  settings-handlers.js	692	Settings save, provider CRUD, command palette
  admin-handlers.js	652	Admin actions, presets, team management
  app.js	567	State, init, boot, auth, listener dispatch

  Unchanged: api.js (575), debug.js (580), events.js (327).

• initListeners() decomposed into domain-specific init functions: _initChatListeners(), _initSettingsListeners(), _initAdminListeners(), _initNotesListeners(), _initGlobalKeyboard(). The orchestrator in app.js dispatches to each.

• Side panel resize changed from self-invoking IIFE to _initSidePanelResize() called during listener init, avoiding DOM timing issues.

• Service worker updated with new file list for pre-caching.

• Policy-gating tests updated to read from the correct source files after the split. All 159 tests pass.

[0.10.2] — 2026-02-24

Added

• Summarize & Continue — User-triggered conversation compaction using the utility model role. Button appears in context warning bar at ≥75% context usage. POST /channels/:id/summarize calls the utility role to generate a summary, inserts it as a tree node with metadata.type = "summary", and updates the cursor. Subsequent completions use the summary as a context boundary — messages before it are replaced by the summary as a system message. Multiple summaries stack. Fork-aware (summaries are tree nodes with their own branch position). "Show full history" toggle reveals collapsed earlier messages.
• BYOK Role Overrides — Users with personal providers can override the org's utility and embedding model roles. Resolution chain: personal → team → global. New "Model Roles" tab in Settings (visible when BYOK is enabled). Stored in user.settings JSONB under model_roles key, same shape as team overrides.
• Utility Rate Limiting — utility_rate_limit global setting (default: 20 calls/hour/user, 0 = unlimited). Org-funded utility calls check against usage_log before executing. BYOK calls exempt (user pays their own way). Returns 429 with clear message when exceeded.
• Message metadata in path — PathMessage now includes metadata JSONB field, enabling summary detection and future message-type extensibility.
• Per-chat model/preset restore — Switching between chats now restores the last-used model or preset in the selector. Stored in localStorage keyed by channel ID. Falls back to the channel's base model, then the global default, if the original selection is no longer available. Cleaned up on chat deletion.

Changed

• Generation role removed — RoleGeneration ("generation") removed from ValidRoles and admin Roles tab. Image/media generation will be extension-managed (v0.11.0) with its own provider config, not a role slot. The current completion/embedding abstraction doesn't fit image gen's fundamentally different API surface.
• Resolver.Complete/Embed signatures — Now accept userID parameter for personal role override resolution. Empty string skips personal lookup.
• Proactive token refresh — Access token is now refreshed at 80% of its lifetime (~12min for 15min tokens). On page reload, a conservative 60s refresh is scheduled since token age is unknown. Eliminates the race condition where profile succeeds on a near-expired token but subsequent calls fail.
• Auth guard in startApp() — If token is invalidated during startup (e.g., 401 on loadChats after profile succeeded), app returns to login instead of rendering a broken UI with cascading 401 errors.
• Per-chat model restore fallback — When the stored model/preset is removed, falls back to admin default → first visible model (was keeping stale selection). Also cleans up the stale localStorage entry.
• BYOK Role Overrides — Fixed response parsing (configs.configs not configs.data) and model field mapping (configId/baseModelId not provider_config_id/model_id).
• GetRole endpoint integration test — GET /admin/roles/:role now exercised by CI via TestIntegration_Roles_GetSingleRole, catching the GetConfig signature mismatch that caused the build failure.
• Auth resilience frontend tests — 10 tests covering startup auth guard, token refresh lifecycle, and the profile-success-then-401 edge case.
• JS syntax lint in CI — node --check runs on all src/js/*.js files before frontend tests, catching parse errors that tests alone can't detect.

[0.10.1] — 2026-02-24

Added

• Admin System Prompt — Global system prompt configured in Admin › Settings. Injected as the first system message in every conversation, before user/preset system prompts. Users cannot override or disable it. Stored in global_settings['system_prompt'].
• Personas tab — User presets moved from the Models tab to their own "Personas" tab in Settings. Tab is hidden when admin disables user presets (allow_user_personas policy).
• Team Management modal — Team admin functionality extracted from Settings into its own tabbed modal (Members, Providers, Presets, Usage, Activity). Accessed via "Team Management" flyout menu item, or "Manage →" on team cards in Settings. Multi-team admins see a picker first; single-team admins go straight to the tabbed view. Back arrow in header returns to picker.
• Preview pane: clear button — Trash icon in the preview header resets the iframe and shows the empty hint. Also auto-clears when deleting a chat that had active preview content.
• Preview/Notes pane: fullscreen — Expand button in the header toggles fullscreen mode (panel fills entire viewport width).
• Preview/Notes pane: resizable — Drag the left edge of the side panel to resize (280px–70vw). Width resets on close.
• Code block: download — "Download" button on every code block. Infers file extension from language tag (e.g. python → .py, go → .go).

Changed

• Personal Usage scoped to BYOK — GET /usage and the user Usage tab now only show consumption against personal (BYOK) providers. Global provider usage is the org's cost, not the user's. New QueryByUserPersonal store method filters on scope = 'personal' AND owner_id = user_id. Admin usage views remain unaffected. Usage tab hidden when admin disables user providers (allow_user_byok policy).
• OpenAI streaming usage race — Deferred the Done event until the usage chunk arrives. Previously, finish_reason fired Done before the usage chunk (which has choices: []), so streaming token counts were always 0 for all OpenAI-compatible providers.

Fixed

• Usage logging zero-token guard — Removed early exit in logUsage that suppressed rows when tokens were 0. Combined with the streaming race above, this meant no streaming usage was ever recorded.
• Live chat completion test — TestLive_VeniceChatCompletion sent wrong field names (messages/config_id instead of content/provider_config_id).

[0.10.0] — 2026-02-24

Added

• Model Roles — Named model slots (utility, embedding, generation) with primary + fallback bindings. Stored in global_settings['model_roles']. Team-level overrides via teams.settings JSONB. New server/roles/ package with Resolver.Complete() and Resolver.Embed() for automatic failover. Admin API: GET/PUT /admin/roles/:role, POST /admin/roles/:role/test. Team API: GET/PUT/DELETE /teams/:id/roles/:role.
• Provider Embed() interface — All providers implement Embed(). OpenAI, Venice, OpenRouter support /v1/embeddings; Anthropic returns ErrNotSupported. New types: EmbeddingRequest, EmbeddingResponse.
• Usage Tracking — Every completion (streaming and non-streaming) logs token counts and cost to usage_log table. Cost calculated at insert time from model_pricing (no retroactive recalculation). Provider scope denormalized for efficient admin filtering. Admin views exclude BYOK. New stores: UsageStore, PricingStore. Admin API: GET /admin/usage, GET /admin/usage/users/:id, GET /admin/usage/teams/:id. User API: GET /usage (personal summary).
• Model Pricing — model_pricing table with catalog sync and manual admin overrides. Sync from provider APIs (Venice, OpenRouter) auto-populates pricing with source='catalog'; manual entries never overwritten. Admin API: GET/PUT/DELETE /admin/pricing.
• Streaming token capture — OpenAI: stream_options.include_usage + parse usage/cache from final chunk. Anthropic: parse message_start for input/cache tokens, message_delta for output tokens. Cache token fields (CacheCreationTokens, CacheReadTokens) on CompletionResponse, StreamEvent, and streamResult.
• Admin Roles tab — Configure primary/fallback provider+model per role, test-fire from UI.
• Admin Usage dashboard — Period selector (7d/30d/90d), group-by (model/user/day/provider), totals cards, breakdown table, pricing table.
• Team Usage dashboard — Team admins can view usage against their team-owned providers via GET /teams/:teamId/usage. Filters to provider_configs WHERE scope='team' AND owner_id=teamId. Integrated into team management panel with period/group-by selectors.
• Admin Reset Password — Button in user list with vault destruction warning dialog (two-step confirmation).
• User Usage tab — Settings modal "Usage" tab shows personal token consumption and estimated costs across all conversations, including BYOK. Period selector (7d/30d/90d) and group-by (model/day).
• Live Venice integration tests — Gated behind VENICE_API_KEY env var. Tests: non-streaming completion, streaming completion, usage logging for both modes, pricing from catalog sync, and direct embeddings via BGE-M3. Uses qwen3-4b (Venice Small) — cheapest at $0.05/$0.15 per 1M tokens.
• Migration 004 — usage_log table, model_pricing table, model_roles seed in global_settings.

Fixed

• UEK re-wrap on password change — ChangePassword now decrypts UEK with old password and re-encrypts with new password + fresh salt. Previously, password changes silently broke all personal BYOK keys.
• UEK destruction on admin password reset — ResetPassword now nullifies vault columns, evicts UEK from cache, deletes personal provider configs, and logs an audit event. Previously, admin resets left orphaned encrypted UEK that silently broke personal keys.
• Streaming completions logged zero tokens — StreamEvent lacked token fields and providers discarded usage data from final chunks. Both OpenAI and Anthropic streaming parsers now capture and propagate token counts.
• OpenAI streaming usage race condition — The OpenAI protocol sends chunks in order: content → finish_reason → usage → [DONE]. The parser sent Done=true on the finish_reason chunk (step 2) before the usage chunk arrived (step 3, with choices:[]). The receiver returned immediately on Done, so usage was captured but never delivered. Fix: defer the Done event as pendingFinish until the usage chunk arrives, then flush with tokens attached. Tool-call finishes flush immediately (tool loop needs the event). Affects all OpenAI-compatible providers (OpenAI, Venice, OpenRouter).
• Token accumulation across tool iterations — streamResult in stream_loop.go now accumulates input/output/cache tokens across multi-tool-call iterations instead of only capturing the final iteration.
• Admin pricing leaked BYOK entries — PricingStore.List() returned all model_pricing rows including those from personal BYOK providers. Admin panel showed pricing entries for user-private providers they shouldn't see, with O(users × models) scale explosion. Now joins on provider_configs and filters scope != 'personal'. Admin UpsertPricing also validates provider scope, rejecting manual pricing on personal providers.
• Team role handlers used wrong param name — ListTeamRoles, UpdateTeamRole, DeleteTeamRole read c.Param("id") but routes use :teamId. Every team role operation silently got an empty team ID.
• Usage logging suppressed for zero-token streams — logUsage had an early exit when inputTokens == 0 && outputTokens == 0. Combined with the OpenAI streaming parser bug (below), this silently dropped all streaming usage rows. Removed the guard — requests are always recorded.
• Live chat completion test used wrong field names — TestLive_VeniceChatCompletion sent messages and config_id but the handler expects content and provider_config_id. Test silently skipped on the binding error.

[0.9.4] — 2026-02-24

Added

• API key encryption (vault) — Two-tier AES-256-GCM encryption for stored API keys. Global/team keys encrypted with env-var-derived key (HKDF-SHA256); personal BYOK keys encrypted with per-user encryption key (UEK) derived from password via Argon2id. Admin cannot recover personal keys without user's passphrase. New server/crypto/ package: vault.go, cache.go, resolver.go, backfill.go with 11 round-trip tests.
• UEK lifecycle — UEK generated on registration, unwrapped on login (Argon2id → AES-GCM), cached in sync.Map for session duration, evicted with memory zeroing on logout. Pre-migration users auto-initialize vault on first login.
• Migration 003_vault.sql — Adds encrypted_uek, uek_salt, uek_nonce, vault_set to users table. Adds api_key_enc (BYTEA), key_nonce, key_scope to provider_configs. Backfills key_scope from existing scope.
• Startup backfill — BackfillEncryptedKeys() encrypts plaintext API keys on first startup with ENCRYPTION_KEY set. EnforceEncryptionKey() refuses startup if encrypted keys exist but env var is missing.
• CI/CD: encryption secret — ENCRYPTION_KEY Gitea secret synced to k8s switchboard-encryption secret. Backend manifest references it as optional secretKeyRef.

Fixed

• HTML code blocks render live in chat (XSS) — When a model's </think> tag directly abutted a code fence (no newline), the fence wasn't recognized by marked.js, causing raw HTML to render as live DOM elements (canvas games, styles, etc.). Three-layer fix: (1) DOMPurify switched from permissive ADD_TAGS (default allows canvas, style, form, etc.) to strict ALLOWED_TAGS allowlist of only markdown-produced elements; (2) think-block placeholders padded with \n\n to ensure adjacent fences start on fresh lines; (3) unclosed code fences auto-closed before marked.parse for streaming protection.

[0.9.3] — 2026-02-23

Changed

• Code blocks: button-driven collapse — Replaced <details> wrapper with inline collapse/expand toggle button in the code toolbar. Auto-collapses at

  15 lines with a fade mask; toggle available on all blocks >5 lines. User can always expand/collapse regardless of threshold. Smooth CSS transition instead of native <details> jump.

• Thinking blocks: always visible — Thinking blocks are always rendered in the DOM regardless of the showThinking setting. The setting now controls whether blocks start expanded (<details open>) or collapsed. User can always click to toggle. Setting label updated to "Auto-expand thinking blocks".

Added

• Admin default model — New default_model policy in Admin → Settings. Dropdown populated from enabled models. When a user has no saved selection (fresh login, cleared browser) or their saved model is no longer available, the admin default is used before falling back to first visible. Resolution chain: localStorage → admin default → first visible.
• Reasoning/thinking support for OpenAI-compatible providers — Models that send reasoning_content in stream deltas (Grok, DeepSeek, etc.) now have thinking blocks streamed live and persisted as <think> tags. Renders as collapsible thinking blocks in both streaming and history views.
• Favicon animation during generation — Browser tab favicon pulses with cascading dot opacity animation while a completion is in progress, restoring to the static favicon when done.
• Tool calls in message history — Tool call metadata (name, arguments, result, error status) is now persisted alongside assistant messages in the database. History view renders tool calls as collapsed <details> blocks showing "🔧 tool_name → done" with expandable input/output JSON.
• Notes tool: "View note" link — When a notes tool (note_create, note_update, etc.) completes, a "📝 View note" button appears in both the live streaming tool indicator and the history tool call block. Clicking opens the Notes panel and navigates directly to the note.
• Notes panel: Copy button — "Copy" button added to note read mode, copies title + content as markdown to clipboard.
• Brand hover: jitter-free crossfade — Sidebar brand logo→collapse icon swap uses opacity crossfade instead of display: none toggle, eliminating layout reflow jitter on hover.

Fixed

• Model selector shows unavailable model — Selector restoration searched all models including user-hidden ones. Saved selection (localStorage) for a hidden model like Claude Opus would re-select it on every page load even when only Grok was visible. Resolution now filters to visible models only.
• Refresh toast shows wrong count — "Loaded 33 models" counted all models including hidden; now shows only visible count ("Loaded 1 model").
• Tool calls vanish after completion — Live streaming tool indicators disappeared when reloadActivePath() rebuilt messages from DB. Fixed: the getActivePath query now includes tool_calls column, and PathMessage carries the data through to the frontend renderer.
• Tool result "undefined results" text — Operator precedence bug in tool result summary parser caused undefined results to display for note_create. Fixed summary logic to properly branch on title vs count.
• Regenerate streams at wrong position — Regen'd response appeared below the full conversation (appended at bottom) then snapped to correct position after completion. Fixed: display is now truncated to the parent message before streaming starts, so the new response streams in-place as a clean branch. Backend context was already correct (excludes old response).
• Regenerate loses tools, reasoning, and tool execution — Regen handler was a stripped-down copy of the completion handler missing tool definitions, tool execution loop, reasoning_content forwarding, and tool_calls persistence. Model lost access to note tools on regen and fell back to generic capabilities. Refactored: extracted streamWithToolLoop() into stream_loop.go as the single canonical streaming implementation. Both streamCompletion (normal chat) and Regenerate now call the shared function — only persistence differs. Future streaming features (new event types, tool capabilities) automatically apply to all code paths.
• OpenRouter free models crash — Free models with nil pricing caused pq: invalid input syntax for type json on catalog sync. Nil pricing now routes through ToJSON() producing "{}" instead of nil []byte.
• API key appears unsaved — ListGlobalConfigs returned raw ProviderConfig structs where APIKeyEnc is json:"-" (never serialized), so has_key was always undefined. Now returns computed has_key field.
• Case-insensitive usernames — Login, registration, and all user lookups use LOWER() in SQL. All user creation paths normalize to lowercase. New migration 002_ci_username.sql adds LOWER() unique indexes and normalizes existing rows.

[0.9.2] — 2026-02-23

Added

• Collapsible code blocks: Code blocks over 15 lines auto-collapse into <details> with language and line count summary. Language label shown in top-left corner of all code blocks.
• HTML preview: "Preview" button on HTML code blocks opens a sandboxed iframe (allow-scripts, no allow-same-origin). Auto-detected for untagged blocks via heuristic.
• Token count estimate: Live token counter below input area showing approximate tokens and context usage percentage when model has max_context.
• Context length warning: Dismissable banner above input at 75% (yellow) and 90% (red) context usage with guidance to start a new chat.
• Proxy interception detection: _parseJSON() checks Content-Type before .json() on all API paths including streaming. Typed proxyBlocked errors with proxy page title extraction and actionable splash messages.
• Environment injection: window.__ENV__ wired through entrypoint, k8s, and index.html for dev/test/production gating.
• Enhanced diagnostics: NET:PROXY log type, Content-Type capture in fetch interceptor, environment info in export header and state snapshot.
• Team admin audit scoping: New GET /api/v1/teams/:teamId/audit and /audit/actions endpoints scoped to team members. Activity Log section in team manage panel with filter dropdown and pagination.
• New favicon: Switchboard panel design with provider-colored jacks. Animated SVG (rotating plugs), 32px PNG, 256px PNG, and ICO.
• Seed users (dev/test only): SEED_USERS=user:pass:role,... env var pre-creates active users on startup. Ignored in production. K8s secret switchboard-seed-users wired in backend manifest.

[0.9.1] — 2026-02-23

Removed

• Static known model table: Deleted knownModels map from backend and KNOWN_MODELS from frontend. The same model ID can have different capabilities depending on the provider (e.g. DeepSeek has tool_calling on OpenRouter but not on Venice). A hardcoded table can't represent this.
• Frontend lookupKnownCaps(): Removed client-side capability guessing. Backend is the sole source of truth via catalog → heuristic chain.

Changed

• Resolution chain simplified: catalog (provider API sync) → heuristic inference. No intermediate known table. Providers that report capabilities via API are authoritative; heuristics are best-effort for unsynced models.
• All providers updated: OpenAI, OpenRouter, Anthropic, Venice now call InferCapabilities() directly instead of the dead known table lookup.
• Frontend resolveCapabilities(): Now passes through backend caps as-is. No client-side merging with a static table.
• EXTENSIONS.md recovered into repo, updated with Appendix A (Custom Renderers) and Appendix B (Model Roles with utility/embedding/generation slots)
• ROADMAP.md restructured: extension foundation pulled to v0.11.0, model roles to v0.10.0, dependency graph, TBD replaces post-1.0, removed v0.8→v0.9 migration (OBE — no public release, no test path)

Added

• Heuristic patterns: Updated to detect qwen3, gpt-5, grok, kimi, minimax, glm-5, gemma-3 model families. Vision expanded to claude-opus/sonnet (not just claude-3). Reasoning expanded for thinking, grok, glm patterns.

Fixed

• Preset capability pills: Presets with auto-resolve (no provider_config_id) now inherit base model capabilities via GetByModelIDAny catalog fallback.
• Venice optimizedForCode: Added mapping to CodeOptimized capability.
• CI test stability: BYOK journey tests use unreachable endpoints so auto-fetch doesn't race with simulated data injection.

[0.9.0] — 2026-02-22

Added

• Schema consolidation: 21 migrations collapsed to single 001_initial.sql
• Store layer: All database access through typed interfaces (no raw SQL in handlers)
• Persona model: Trust-boundary model replacing old presets; scoped global/team/personal
• Capabilities resolver: Three-tier chain — catalog → known table → heuristic inference
• Three-state model visibility: enabled / disabled / team-only
• BYOK auto-fetch: Creating a personal provider triggers model discovery from provider API
• User model refresh: POST /api-configs/:id/models/fetch endpoint + UI button
• Composite model IDs: configId:modelId format prevents cross-provider collisions
• Audit log foundation: All admin operations logged with actor, action, resource
• Journey integration tests: API-driven test suite replacing fake-data tests
• Frontend test suite: 107 tests, 27 suites validating model processing pipeline
• Live Venice API test: Proves real BYOK → auto-fetch → models visible flow

Fixed

• API key storage: json:"-" tag on ProviderConfig.APIKeyEnc silently dropped keys during admin create/update. Fixed with wrapper structs that bypass the tag.
• NULL model_default scan: scanProviders() crashed on NULL model_default column, silently hiding all team and BYOK models. Fixed with sql.NullString.
• Nil slice serialization: Go nil slices serialized as JSON null instead of [], breaking frontend fallback chains. Fixed with make([]T, 0).
• Frontend error swallowing: API responses with errors field were silently ignored.

Changed

• Backward-compatible API routes with v0.8 field name aliases
• User model preferences table (user_model_settings)