core/docs/ARCHITECTURE.md

Architecture — Chat Switchboard v0.17

Deployment Modes

Three Docker images support different deployment scenarios:

Image	Dockerfile	Contents	Use Case
Unified	Dockerfile	nginx + Go backend	Dev, docker-compose, single-node
Backend	server/Dockerfile	Go binary only	K8s — scale API independently
Frontend	Dockerfile.frontend	nginx + static files + CM6 bundle	K8s — scale FE independently

Unified bundles everything in one container (4-stage Docker build: vendor libs → CM6 bundle → Go backend → nginx runtime). Nginx serves static files and proxies /api/* and /ws to the Go backend running on :8080. Good for development and small deployments.

Split (Backend + Frontend) for production K8s: Ingress routes /api/* and /ws to the backend Service, everything else to the frontend Service. The frontend entrypoint (docker-entrypoint-fe.sh) handles BASE_PATH injection into index.html and dynamic nginx config generation at startup. Supports branding volume mounts at /branding/.

                    ┌─────────────────────────┐
                    │  Ingress / Traefik       │
                    │  ├─ /api/* → be-svc:8080 │
                    │  ├─ /ws   → be-svc:8080  │
                    │  └─ /*    → fe-svc:80    │
                    └─────────────────────────┘
                         │              │
              ┌──────────▼──┐   ┌───────▼────────┐
              │  Backend    │   │  Frontend       │
              │  (Go :8080) │   │  (nginx :80)    │
              │  replicas:N │   │  replicas:M     │
              └──────┬──────┘   └────────────────┘
                     │
              ┌──────▼──────┐
              │ PostgreSQL  │   (or SQLite for
              └─────────────┘    single-node)

Design Principles

1. Persona-as-Trust-Boundary: A persona (model + config + prompt) is the unit of access control. Users interact with personas, not raw provider configs. Admins control which models are visible; team admins control which personas their team can use.

2. Roles vs Teams: Clean separation between vertical permissions (Roles: admin, user) and horizontal visibility (Teams: organizational units). A user's role determines what they can do; their team membership determines what they can see.

3. Store Layer: All database access goes through typed Go interfaces (store.Stores). Handlers never write raw SQL. The store layer has two implementations — store/postgres/ and store/sqlite/ — selected at startup via DB_DRIVER. This enables Postgres for production and SQLite for single-node or air-gapped deployments.

4. Scope Model: Provider configs, personas, and model settings all use a three-value scope column: global (admin-managed, visible to all), team (team-admin-managed, visible to team), personal (user-managed, visible to owner). The owner_id column points to the owning user or team depending on scope.

5. Capabilities Resolution: Model capabilities (vision, tool calling, thinking, context window) are resolved through a priority chain: catalog DB (provider API sync, per-provider authoritative) → heuristic inference (regex patterns on model ID). No static model table — the same model can have different capabilities on different providers. The catalog is populated by auto-fetch on provider creation and manual refresh.

6. Channels as Execution Context: Channels are the universal container for conversation state — messages, tool activity, notes, and artifacts all hang off a channel. Today channels are single-user direct chats. The architecture anticipates multi-participant channels (team members, anonymous visitors, AI personas) for workflow execution, without requiring changes to the message tree, tool framework, or streaming infrastructure.

Workflow Architecture (Future — v0.21.0+)

The platform's existing primitives (teams, personas, channels, tools, notes) compose into a workflow engine where the channel is the execution context that moves through defined stages.

Conceptual Model

Team
 └─ Workflow (team-admin defined)
      ├─ name, description
      ├─ entry conditions (public link, team-internal, API trigger)
      └─ Stages[]
           ├─ persona_id (which AI drives this stage)
           ├─ assignment_team_id (who can be assigned)
           ├─ form_template (structured note schema, optional)
           └─ transitions[] (conditions → next stage)

Channel (workflow instance)
 ├─ workflow_id + current_stage
 ├─ participants[]
 │    ├─ anonymous visitor (mTLS fingerprint / session token)
 │    ├─ AI persona (per-stage, from workflow definition)
 │    └─ assigned team member (claimed or auto-routed)
 ├─ messages (existing tree structure)
 ├─ notes (channel-scoped artifacts, intake forms)
 └─ tool activity (existing execution framework)

How Existing Primitives Map

Existing Primitive	Workflow Role
Persona (system prompt + model)	Drives a workflow stage — the AI knows what to collect, when to route
Team (members + roles)	Owns the workflow definition; members are assignable to channels
Channel (messages + tree)	Execution instance of a workflow; full conversation history
Notes + Tools	Structured data collection; the persona's system prompt is the form definition, the note is the filled form
EventBus + WebSocket	Real-time notifications for assignment, stage transitions, new messages

Design Constraints for Current Development

These invariants keep the workflow path open without building it prematurely:

• Channels: Don't assume single-owner. If touching channel queries, keep room for team_id, type (beyond direct), and multi-participant access patterns alongside user_id.
• Notes: User-scoped with [[wikilink]] bi-directional linking (v0.17.3). The note_links junction table tracks directed edges between notes, with target_note_id nullable for dangling links (unresolved references). Links are extracted from content on save via regex, resolved by title match, and re-resolved when new notes are created. The graph endpoint returns all nodes, edges, and unresolved links for Canvas-based force-directed visualization. Future channel-scoped notes (attached to a conversation) need a channel_id FK option. source_message_id enables jump-to-source provenance from notes back to the originating chat message.
• Personas: Already team-scoped. Don't couple to "user picks from dropdown" — a workflow stage references a persona programmatically.
• Tool ExecutionContext: Already carries UserID + ChannelID. Will need TeamID and WorkflowID — the struct is easily extended.
• Auth: mTLS anonymous users need identity to participate in channels. Lightest version: a participants table that can reference a user_id or an opaque session identifier (cert fingerprint). Don't assume every channel participant has a row in users.

Package Structure

server/
├── main.go                    # Wiring: stores → handlers → routes
├── config/config.go           # Env-based configuration
├── database/
│   ├── database.go            # Connection management (PG + SQLite)
│   ├── migrate.go             # Auto-migration on startup
│   └── migrations/
│       ├── 001_v016_schema.sql  # Consolidated schema (PG)
│       └── 002_v017_persona_kb.sql
├── store/
│   ├── interfaces.go          # Store interfaces + shared types
│   ├── postgres/              # Postgres implementations (19 stores)
│   │   ├── stores.go          # NewStores() constructor
│   │   ├── provider.go        # ProviderStore
│   │   ├── catalog.go         # CatalogStore
│   │   ├── persona.go         # PersonaStore
│   │   ├── channel.go         # ChannelStore
│   │   ├── message.go         # MessageStore
│   │   ├── user.go            # UserStore
│   │   ├── team.go            # TeamStore
│   │   ├── note.go            # NoteStore
│   │   ├── knowledge.go       # KnowledgeStore
│   │   └── ...
│   └── sqlite/                # SQLite implementations (19 stores)
│       ├── stores.go          # NewStores() constructor
│       └── ...                # Mirror of postgres/ with dialect adaptations
├── models/models.go           # Shared domain types
├── capabilities/
│   ├── intrinsic.go           # Heuristic detection + resolution
│   └── resolver.go            # ModelsForUser() unified resolver
├── compaction/                # Conversation summarization engine
├── crypto/                    # AES-256-GCM API key encryption
├── events/                    # EventBus + WebSocket hub
├── extraction/                # Document text extraction pipeline
├── handlers/                  # HTTP handlers (Gin)
│   ├── auth.go                # Login, register, refresh, logout
│   ├── admin.go               # User/config/model management
│   ├── channels.go            # Channel CRUD
│   ├── messages.go            # Message CRUD + forking
│   ├── completion.go          # Chat completions (SSE streaming)
│   ├── stream_loop.go         # Shared streaming + tool execution loop
│   ├── capabilities.go        # Model list + ResolveModelCaps
│   ├── presets.go             # Persona CRUD (all scopes)
│   ├── apiconfigs.go          # User provider config CRUD (BYOK)
│   ├── teams.go               # Team management
│   ├── notes.go               # Notes CRUD + search + graph + backlinks
│   ├── knowledge.go           # Knowledge base management
│   └── ...
├── notelinks/                 # Wikilink extraction (regex → NoteLink structs)
├── knowledge/                 # KB chunking, embedding, search
├── providers/                 # LLM provider adapters
│   ├── provider.go            # Provider interface
│   ├── anthropic.go
│   ├── openai.go
│   ├── openrouter.go
│   └── venice.go
├── middleware/                 # Auth, admin, CORS, rate limiting
├── storage/                   # Blob storage (PVC + S3)
└── tools/                     # Built-in tool definitions

Store Layer Pattern

Every store follows the same pattern:

// Interface in store/interfaces.go
type FooStore interface {
    Create(ctx context.Context, f *models.Foo) error
    GetByID(ctx context.Context, id string) (*models.Foo, error)
    Update(ctx context.Context, id string, patch models.FooPatch) error
    Delete(ctx context.Context, id string) error
    List(ctx context.Context, opts ListOptions) ([]models.Foo, int, error)
}

// Implementation in store/postgres/foo.go
type fooStore struct{ db *sql.DB }
func (s *fooStore) Create(ctx context.Context, f *models.Foo) error { ... }

Handlers receive store.Stores (a bundle of all store interfaces):

type AdminHandler struct { stores store.Stores }

func (h *AdminHandler) CreateUser(c *gin.Context) {
    // ...
    err := h.stores.Users.Create(c.Request.Context(), user)
}

Capabilities Resolution Chain

When the system needs to know what a model can do (vision? tools? thinking?):

1. model_catalog DB (exact match: model_id + provider_config_id)
   ↓ miss
2. model_catalog DB (any provider: same model_id)
   ↓ miss
3. Heuristic inference (name-based: "gpt-4-vision" → vision=true)

The capabilities.ModelsForUser() function combines catalog entries, team personas, and user preferences into a single unified model list for the frontend.

Scope / Ownership Model

scope='global'   → owner_id=NULL     → Admin-managed, visible to all
scope='team'     → owner_id=team.id  → Team-admin-managed, visible to team
scope='personal' → owner_id=user.id  → User-managed, visible to owner

Used by: provider_configs, personas, model_catalog (visibility column adds enabled/disabled/team states on top).

Schema Migration

Migrations are handled by the backend on startup (no separate migration job):

1. Creates schema_migrations table if absent
2. Checks which migration files have been applied
3. Applies any new .sql files in order

Postgres uses consolidated schemas (001_v016_schema.sql + incremental). SQLite uses application-generated UUIDs and datetime() instead of gen_random_uuid() and now(). Both drivers share the same migration tracking table.

Frontend Architecture

Vanilla JavaScript, no framework. The frontend ships as static files served by nginx. The only build step is the CM6 editor bundle, compiled at Docker build time via esbuild (IIFE output, no runtime bundler).

File Structure

src/
├── index.html                  # Single-page app shell
├── sw.js                       # Service worker (offline cache)
├── manifest.json               # PWA manifest
├── css/styles.css              # All styles (CSS variables, light/dark themes)
├── js/
│   ├── api.js                  # HTTP client with token refresh
│   ├── app.js                  # Application state machine, startup, routing
│   ├── chat.js                 # Chat input abstraction, message send/receive
│   ├── events.js               # Labeled event bus + WebSocket bridge
│   ├── debug.js                # Debug panel, diagnostics, state export
│   ├── ui-core.js              # DOM rendering, sidebar, modals, panels
│   ├── ui-settings.js          # Settings tabs, theme, appearance, teams
│   ├── ui-admin.js             # Admin panel rendering
│   ├── admin-handlers.js       # Admin CRUD operations + extension editors
│   ├── settings-handlers.js    # User settings CRUD, command palette
│   ├── tokens.js               # Input token counting + context budget
│   ├── attachments.js          # File upload, paste-to-file, lightbox
│   ├── notes.js                # Notes panel CRUD + graph + daily notes
│   ├── note-graph.js           # Canvas force-directed graph visualization
│   ├── knowledge.js            # Knowledge base UI
│   └── __tests__/              # Node.js test suite (node --test)
├── editor/                     # CM6 source (compiled at build time)
│   ├── package.json            # CM6 + language mode dependencies
│   ├── package-lock.json       # Lockfile for npm ci
│   ├── build.mjs               # esbuild script → IIFE bundle
│   ├── index.mjs               # Bundle entrypoint (window.CM)
│   ├── chat-input.mjs          # Chat input factory (markdown, code blocks)
│   ├── code-editor.mjs         # Code editor factory (syntax, keybindings)
│   ├── note-editor.mjs         # Note editor factory (markdown + wikilinks)
│   ├── wikilink.mjs            # CM6 wikilink extension (parse, decorate, autocomplete)
│   └── theme.mjs               # Switchboard + chat input + note editor themes
└── vendor/                     # Vendored libraries (local + CDN fallback)
    ├── marked/                 # Markdown renderer
    ├── purify/                 # DOMPurify (XSS protection)
    ├── mermaid/                # Diagram renderer
    ├── katex/                  # Math renderer
    └── codemirror/             # CM6 bundle (built by Docker)
        └── codemirror.bundle.js  # ~295KB min, ~90KB gzip

CM6 Integration

The CodeMirror 6 bundle provides three factory functions exposed on window.CM:

CM.chatInput(target, opts) — Markdown-mode editor for the chat input area. Features: auto-growing height, Enter=send / Shift+Enter=newline, WYSIWYG fenced code block decorations (visual container with monospace font and accent border), inline code shortcut (Ctrl/Cmd+E), spell check. Always uses standard keybindings.

CM.codeEditor(target, opts) — Full-featured code editor for the admin extension panel (JSON manifests, JavaScript scripts). Features: line numbers, bracket matching, search/replace, fold gutter, 10 bundled language modes. Supports Vim and Emacs keybindings via user preference (reconfigurable at runtime via compartments).

CM.noteEditor(target, opts) — Rich markdown editor for the notes panel. Features: heading size rendering (h1–h3), blockquote styling, fenced code block decorations, [[wikilink]] autocomplete (triggered by [[), wikilink chip rendering (clickable, styled by link/transclusion type), search/replace. onLink callback for navigating to linked notes; linkCompleter async callback for autocomplete results.

Both factories return a clean API: getValue(), setValue(), focus(), destroy(). The ChatInput abstraction in chat.js wraps the CM6 instance with a textarea fallback — all callsites use ChatInput.getValue() etc., never raw DOM access.

Graceful degradation: Every integration point checks window.CM before using CM6. If the bundle fails to load (air-gapped without the build step, CDN issues), the app falls back to native <textarea> with zero breakage.

Theme System

CSS variables define the color palette in :root (dark, default) and [data-theme="light"] (light override). All UI components — including CM6 editors — reference these variables, so theme switches propagate instantly without editor reconfiguration.

The appearance settings offer three modes: Light, Dark, System. System mode listens to prefers-color-scheme and re-evaluates on OS theme change. Theme changes emit theme.changed on the EventBus; CM6 code editors toggle the oneDark syntax theme via compartment reconfiguration.

Communication Pattern

app.js calls API.* methods, updates state, then calls UI.* methods to render. events.js handles real-time updates via WebSocket. No framework, no virtual DOM, no reactive bindings.

Security Model

• Auth: JWT access tokens (short-lived) + refresh tokens (DB-stored, revocable)
• Admin Bootstrap: SWITCHBOARD_ADMIN_USERNAME/PASSWORD env vars create/update admin on every startup (K8s secret pattern)
• API Key Storage: Provider API keys encrypted with AES-256-GCM. Two-tier: org keys use ENCRYPTION_KEY env var, personal BYOK keys use per-user encryption keys (admins cannot recover)
• Policies: Boolean flags in global_settings table control registration, BYOK, team providers, etc.
• Audit: All admin operations logged to audit_log with actor, action, resource type/ID, and diff
• Banner: Environment classification banner configurable via admin settings (text, color, position)
• CORS: Configurable allowed origins via env var
• No sensitive terminology: Banner system avoids classification-related terms for security compliance

File Storage

Blob storage for attachments, extracted text, and (future) knowledge base documents.

Interface: storage.ObjectStore — Put, Get, Delete, DeletePrefix, Exists, Healthy, Stats, Backend. All handlers use the interface; backend is selected at startup.

Backends:

Backend	Config	Use Case
PVC	STORAGE_BACKEND=pvc + STORAGE_PATH	Single-node, dev, docker-compose. Local filesystem with atomic writes (temp+rename).
S3	STORAGE_BACKEND=s3 + S3_ENDPOINT, S3_BUCKET, credentials	Multi-node production. MinIO, Ceph RGW, AWS S3. Uses minio-go v7.
Auto	STORAGE_BACKEND= (empty)	Tries PVC at STORAGE_PATH; disables if not writable.

Key layout: attachments/{channel_id}/{attachment_id}_{filename}. S3 prefix (S3_PREFIX) prepended for shared buckets.

PVC always mounted: Even with S3 backend, the PVC mount at STORAGE_PATH is needed for the extraction queue's local scratch directory (processing/{id}/status.json). With S3, the PVC can be small (1Gi) — only transient coordination state, not blobs.

Admin panel: Settings → Storage tab shows backend type, health status, file count, total size, endpoint/bucket (S3) or path (PVC), orphan detection and cleanup.

Backward Compatibility

v0.9 maintains backward-compatible API routes:

Old Route	New Handler	Notes
/api/v1/presets	PersonaHandler	Returns both personas and presets keys
/api/v1/api-configs	ProviderConfigHandler	Unchanged route, new implementation
/api/v1/models	ModelHandler.ListEnabledModels	Alias for /models/enabled

JSON field rename: api_config_id → provider_config_id in channel and completion request/response bodies. Frontend updated to match.