feat(providers): runtime provider management with CRUD, presets, and multi-auth

CLAUDE.md

@@ -114,7 +114,7 @@ curl http://localhost:3000/api/v1/health   # backend (via web proxy)
 
 ```text
 src/synthorg/
-  api/            # Litestar REST + WebSocket API (controllers, guards, channels, JWT + API key + WS ticket auth, approval gate integration, coordination endpoint, collaboration endpoint, settings endpoint, RFC 9457 structured errors (ErrorCategory, ErrorCode, ErrorDetail, ProblemDetail, CATEGORY_TITLES, category_title, category_type_uri, content negotiation)), AppState hot-reload slots (provider_registry, model_router with swap methods), settings dispatcher lifecycle
+  api/            # Litestar REST + WebSocket API (controllers, guards, channels, JWT + API key + WS ticket auth, approval gate integration, coordination endpoint, collaboration endpoint, settings endpoint, provider management endpoint (CRUD + test + presets), RFC 9457 structured errors (ErrorCategory, ErrorCode, ErrorDetail, ProblemDetail, CATEGORY_TITLES, category_title, category_type_uri, content negotiation)), AppState hot-reload slots (provider_registry, model_router with swap methods, provider_management), settings dispatcher lifecycle
     auth/         # Authentication subpackage (controller, service, middleware, JWT + API key + WS ticket store, models, config)
   budget/         # Cost tracking, budget enforcement (pre-flight/in-flight checks, auto-downgrade), billing periods, cost tiers, quota/subscription tracking, CFO cost optimization (anomaly detection, efficiency analysis, downgrade recommendations, approval decisions), spending reports, budget errors (BudgetExhaustedError, DailyLimitExceededError, QuotaExhaustedError)
   cli/            # Python CLI module (superseded by top-level cli/ Go binary)
@@ -127,7 +127,7 @@ src/synthorg/
   memory/         # Persistent agent memory (pluggable MemoryBackend protocol), backends/ (Mem0 adapter: backends/mem0/), retrieval pipeline (ranking, RRF fusion, injection, context formatting, non-inferable filtering), shared org memory (org/), consolidation/archival (consolidation/, dual-mode density-aware archival: DensityClassifier, AbstractiveSummarizer, ExtractivePreserver, DualModeConsolidationStrategy)
   persistence/    # Operational data persistence — pluggable PersistenceBackend protocol, SQLite initial, SettingsRepository (namespaced settings CRUD) (see Memory & Persistence design page)
   observability/  # Structured logging, correlation tracking, log sinks
-  providers/      # LLM provider abstraction (LiteLLM adapter)
+  providers/      # LLM provider abstraction (LiteLLM adapter), auth types (AuthType enum: api_key/oauth/custom_header/none), presets (ProviderPreset, PROVIDER_PRESETS for Ollama/LM Studio/OpenRouter/vLLM), runtime CRUD (management/ -- ProviderManagementService, asyncio.Lock-serialized create/update/delete/test, hot-reload of ProviderRegistry + ModelRouter via AppState swap)
   settings/       # Runtime-editable settings persistence (DB > env > YAML > code defaults), typed definitions (9 namespaces, including JSON type for structural data), Fernet encryption for sensitive values, config bridge (JSON serialization for Pydantic models/collections), ConfigResolver (typed scalar + structural data accessors for controllers — get_agents, get_departments, get_provider_configs with validation fallbacks to YAML), validation, registry, change notifications via message bus, SettingsSubscriber protocol (subscriber.py), SettingsChangeDispatcher (dispatcher.py, polls #settings channel, routes to subscribers, restart_required filtering)
     definitions/  # Per-namespace setting definitions (api, company, providers, memory, budget, security, coordination, observability, backup)
     subscribers/  # Concrete settings subscribers (ProviderSettingsSubscriber — rebuilds ModelRouter on strategy change, MemorySettingsSubscriber — advisory logging for memory config)
@@ -138,7 +138,7 @@ src/synthorg/
 web/              # Vue 3 + PrimeVue + Tailwind CSS dashboard
   src/
     api/          # Axios client, endpoint modules, TypeScript types (mirrors backend Pydantic models)
-    components/   # Vue components organized by feature (agents/, approvals/, budget/, common/, dashboard/, layout/, messages/, org-chart/, tasks/)
+    components/   # Vue components organized by feature (agents/, approvals/, budget/, common/, dashboard/, layout/, messages/, org-chart/, providers/, tasks/)
     composables/  # Reusable composition functions (useAuth, useLoginLockout, usePolling, useOptimisticUpdate, useWebSocketSubscription)
     router/       # Vue Router config with auth guards
     stores/       # Pinia stores (auth, agents, tasks, budget, messages, meetings, approvals, websocket, analytics, company, providers)

docs/design/operations.md

@@ -43,6 +43,7 @@ whether the backend is a cloud API, OpenRouter, Ollama, or a custom endpoint.
     ```yaml
     providers:
       example-provider:
+        auth_type: api_key             # api_key | oauth | custom_header | none
         api_key: "${PROVIDER_API_KEY}"
         models:                        # example entries -- real list loaded from provider
           - id: "example-large-001"
@@ -65,6 +66,7 @@ whether the backend is a cloud API, OpenRouter, Ollama, or a custom endpoint.
             estimated_latency_ms: 200
 
       openrouter:
+        auth_type: api_key           # api_key | oauth | custom_header | none
         api_key: "${OPENROUTER_API_KEY}"
         base_url: "https://openrouter.ai/api/v1"
         models:                        # example entries
@@ -76,6 +78,7 @@ whether the backend is a cloud API, OpenRouter, Ollama, or a custom endpoint.
             alias: "or-reasoning"
 
       ollama:
+        auth_type: none
         base_url: "http://localhost:11434"
         models:                        # example entries
           - id: "llama3.3:70b"
@@ -98,6 +101,17 @@ The framework uses **LiteLLM** as the provider abstraction layer:
 - Load balancing across providers
 - OpenAI-compatible interface (all providers normalized)
 
+### Provider Management
+
+Providers can be managed at runtime through the API without restarting:
+
+- **CRUD**: Create, update, delete providers via `POST/PUT/DELETE /api/v1/providers`
+- **Connection test**: `POST /api/v1/providers/{name}/test` -- sends a minimal probe and reports latency
+- **Presets**: `GET /api/v1/providers/presets` lists built-in templates (Ollama, LM Studio, OpenRouter, vLLM); `POST /api/v1/providers/from-preset` creates from a template
+- **Hot-reload**: On mutation, `ProviderManagementService` rebuilds `ProviderRegistry` + `ModelRouter` and atomically swaps them in `AppState` -- no downtime
+- **Auth types**: `api_key` (default), `oauth` (stores credentials, MVP uses pre-fetched token), `custom_header`, `none` (local providers)
+- **Credential safety**: Secrets are Fernet-encrypted at rest via the `providers.configs` sensitive setting; API responses use `ProviderResponse` DTO that strips all secrets and provides `has_api_key`/`has_oauth_credentials`/`has_custom_header` boolean indicators
+
 ### Model Routing Strategy
 
 Model routing determines which LLM handles a given request. Six strategies are available,
@@ -981,7 +995,7 @@ future CLI tool are thin clients that call the API -- they contain no business l
 | `/api/v1/approvals` | Pending human approvals queue |
 | `/api/v1/analytics` | Performance metrics, dashboards |
 | `/api/v1/settings` | Runtime-editable configuration (9 namespaces), schema discovery |
-| `/api/v1/providers` | Model provider status, config |
+| `/api/v1/providers` | Provider CRUD (create, update, delete), connection testing, presets, 4 auth types (api_key, oauth, custom_header, none) |
 | `/api/v1/ws` | WebSocket for real-time updates (ticket auth via `?ticket=`) |
 | `POST /api/v1/auth/ws-ticket` | Exchange JWT for one-time WebSocket connection ticket |
 
@@ -1049,7 +1063,7 @@ and retry guidance.
 - **Budget Panel**: Spending charts, per-agent breakdown (projections/alerts planned)
 - **Meeting Logs**: Placeholder — coming soon
 - **Artifact Browser**: Placeholder — coming soon
-- **Settings**: Runtime-editable configuration via DB-backed settings persistence (9 namespaces: api, company, providers, memory, budget, security, coordination, observability, backup). Setting types: STRING, INTEGER, FLOAT, BOOLEAN, ENUM, JSON. 4-layer resolution (DB > env > YAML > code defaults), Fernet encryption for sensitive values, REST API (GET/PUT/DELETE + schema endpoints for dynamic UI generation), change notifications via message bus. `ConfigResolver` provides typed scalar accessors and structural data accessors for API controllers: scalar reads assemble full Pydantic config models from individually resolved settings (using `asyncio.TaskGroup` for parallel resolution); structural reads (`get_agents`, `get_departments`, `get_provider_configs`) resolve JSON-typed settings with Pydantic schema validation and graceful fallback to `RootConfig` defaults on invalid data. **Hot-reload**: `SettingsChangeDispatcher` polls the `#settings` bus channel and routes change notifications to registered `SettingsSubscriber` implementations. Settings marked `restart_required=True` are filtered (logged as WARNING, not dispatched). Concrete subscribers: `ProviderSettingsSubscriber` (rebuilds `ModelRouter` on `routing_strategy` change via `AppState.swap_model_router`), `MemorySettingsSubscriber` (advisory logging for non-restart memory settings).
+- **Settings**: Provider management tab (add/edit/delete providers, connection test, preset-based creation) integrated alongside runtime-editable configuration via DB-backed settings persistence (9 namespaces: api, company, providers, memory, budget, security, coordination, observability, backup). Setting types: STRING, INTEGER, FLOAT, BOOLEAN, ENUM, JSON. 4-layer resolution (DB > env > YAML > code defaults), Fernet encryption for sensitive values, REST API (GET/PUT/DELETE + schema endpoints for dynamic UI generation), change notifications via message bus. `ConfigResolver` provides typed scalar accessors and structural data accessors for API controllers: scalar reads assemble full Pydantic config models from individually resolved settings (using `asyncio.TaskGroup` for parallel resolution); structural reads (`get_agents`, `get_departments`, `get_provider_configs`) resolve JSON-typed settings with Pydantic schema validation and graceful fallback to `RootConfig` defaults on invalid data. **Hot-reload**: `SettingsChangeDispatcher` polls the `#settings` bus channel and routes change notifications to registered `SettingsSubscriber` implementations. Settings marked `restart_required=True` are filtered (logged as WARNING, not dispatched). Concrete subscribers: `ProviderSettingsSubscriber` (rebuilds `ModelRouter` on `routing_strategy` change via `AppState.swap_model_router`), `MemorySettingsSubscriber` (advisory logging for non-restart memory settings).
 
 ### Human Roles