- Extract CSS and JavaScript into separate static files
- Created nextcloud_mcp_server/auth/static/vector-viz.css
- Created nextcloud_mcp_server/auth/static/vector-viz.js
- Updated templates to reference external assets
- Fix vector visualization issues:
- Normalize vectors before PCA to match Qdrant's cosine distance
- Add zero-norm and NaN detection/handling for large datasets
- Enable responsive Plotly sizing (autosize + responsive config)
- Widen plot area to full viewport width with minimized margins
- Improve visualization accuracy:
- Query point now positioned correctly relative to documents
- Handles 200+ points without JSON serialization errors
- Full-width plot maximizes screen space utilization
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Migrates from custom word-based chunking to LangChain's MarkdownTextSplitter
for better semantic search quality. This implements the chunking portion of
ADR-011.
Changes:
- Replace custom regex word chunker with MarkdownTextSplitter
- Optimized for Markdown content (headers, code blocks, lists)
- Convert from word-based (512 words) to character-based (2048 chars) chunking
- Maintain backward-compatible ChunkWithPosition interface
- Update configuration defaults and validation
- Update all unit tests (12/12 passing)
Benefits:
- Respects markdown structure boundaries
- Never breaks code blocks or headers mid-chunk
- Preserves semantic coherence within chunks
- Expected 20-30% improvement in recall quality
- Industry-standard approach (used by production RAG systems)
Note: Full reindex required to apply new chunking to existing documents.
Current vector database still contains old word-based chunks.
Related: ADR-011 (Improving Semantic Search Quality)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added support for two fusion algorithms (RRF and DBSF) to combine dense
semantic and sparse BM25 search results, with comprehensive documentation
and unit tests.
Changes:
- Added fusion parameter to nc_semantic_search and nc_semantic_search_answer tools
- Updated ADR-014 with detailed comparison of RRF vs DBSF fusion algorithms
- Added unit tests for fusion algorithm initialization and validation
- Updated search_method in responses to include fusion type (e.g., "bm25_hybrid_rrf")
Fusion Algorithms:
- RRF (Reciprocal Rank Fusion): Default, rank-based, general-purpose
- DBSF (Distribution-Based Score Fusion): Score normalization using statistics
RRF is recommended for most use cases due to its robustness and established
track record. DBSF may provide better results when retrieval systems have
very different score distributions.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Refactored LLM provider infrastructure to support sustainable additions of new providers with both embedding and text generation capabilities.
## Major Changes
### Unified Provider Architecture (ADR-015)
- Created `nextcloud_mcp_server/providers/` with unified Provider ABC
- Providers now support optional capabilities (embeddings and/or generation)
- Auto-detection registry with priority: Bedrock → Ollama → Simple
- Backward compatible - existing code continues to work
### New Providers
- **BedrockProvider**: Full Amazon Bedrock integration
- Embeddings: Titan Embed, Cohere Embed models
- Generation: Claude, Llama, Titan Text, Mistral models
- Model-specific request/response handling
- AWS credential chain integration
- **OllamaProvider**: Migrated with both capabilities support
- **AnthropicProvider**: Moved from test code to production providers
- **SimpleProvider**: Migrated in-memory fallback provider
### Breaking Changes
None - full backward compatibility maintained:
- `embedding.get_embedding_service()` still works
- RAG evaluation tests updated to use unified providers
- All existing tests pass (127 unit tests)
### Testing
- Added 9 comprehensive Bedrock unit tests with mocked boto3
- All existing unit tests pass
- Type checking (ty) and linting (ruff) pass
- Verified backward compatibility
### Documentation
- `docs/ADR-015-unified-provider-architecture.md`: Comprehensive ADR
- `docs/bedrock-setup.md`: AWS setup guide with IAM permissions
- `CLAUDE.md`: Updated with provider architecture section
### Dependencies
- Added `boto3>=1.35.0` to dev dependencies (optional)
## Environment Variables
### Bedrock
- `AWS_REGION`: AWS region (e.g., "us-east-1")
- `BEDROCK_EMBEDDING_MODEL`: Model ID for embeddings
- `BEDROCK_GENERATION_MODEL`: Model ID for generation
- `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`: Optional credentials
### Ollama
- `OLLAMA_BASE_URL`: API URL
- `OLLAMA_EMBEDDING_MODEL`: Embedding model (default: "nomic-embed-text")
- `OLLAMA_GENERATION_MODEL`: Generation model
## AWS Bedrock Permissions Required
Minimal IAM policy:
```json
{
"Effect": "Allow",
"Action": ["bedrock:InvokeModel"],
"Resource": ["arn:aws:bedrock:*::foundation-model/*"]
}
```
See `docs/bedrock-setup.md` for detailed setup instructions.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add ADR-013 documenting RAG evaluation architecture
- Implement two-part evaluation: Context Recall (retrieval) + Answer Correctness (generation)
- Create Click CLI for ground truth generation and corpus upload
- Add pytest fixtures and tests for retrieval/generation quality
- Use BeIR/nfcorpus dataset with 5 selected test queries
- Support Ollama and Anthropic LLM providers
- Generate synthetic ground truth answers offline
- Add comprehensive documentation in tests/rag_evaluation/README.md
The framework separates one-time setup (generate/upload) from test execution,
making tests much faster (~6-12 min vs ~15-25 min per run).
Tests are manual only (not in CI) and require external LLM access.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Updates ADR-012 to clarify that all search and filtering operations
must happen server-side, not in the browser.
Key changes:
- Enhanced viz pane data flow showing server-side processing
- Added performance benefits section (384x bandwidth reduction)
- Detailed server-side filtering approach:
* Query execution via search/algorithms.py
* User ID filtering (multi-tenant security)
* Document type filtering
* PCA reduction (768-dim → 2D) on server
* Only 2D coordinates + metadata sent to client
- Updated Phase 3 implementation plan:
* Remove ALL client-side search logic
* Implement /app/vector-viz server endpoint
* htmx form submission for queries
* Performance optimizations (caching, streaming)
This ensures:
- Minimal bandwidth usage (only 2 floats per doc vs 768)
- Client handles only visualization, not computation
- Can visualize 10,000+ documents without client lag
- Raw vectors never leave server (security)
- Same search logic as MCP tool (consistency)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Enhances ADR-012 with detailed architecture visualization and UI mockup
for the vector visualization pane.
Added sections:
- Architecture diagram showing MCP tool and viz pane integration
- Data flow diagrams for both MCP requests and viz pane interactions
- Detailed UI mockup with ASCII art showing:
* Search configuration controls
* Algorithm selector with weight sliders
* Interactive 2D scatter plot (Plotly.js)
* Results panel with scores
* Performance comparison table
- Technology stack details (htmx, Alpine.js, Plotly.js, Tailwind CSS)
The diagrams illustrate how the viz pane and MCP tool share the same
search algorithm implementations from search/algorithms.py, ensuring
consistency between user testing interface and programmatic API.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
After comprehensive research, the hybrid OAuth + AppAPI architecture is NOT
being implemented due to fundamental architectural incompatibilities.
Key updates:
- Status: Proposed → Not Planned
- Added validation from Nextcloud Context Agent project
- Context Agent (official NC ExApp with MCP) faces IDENTICAL limitations
- Proves constraints are architectural, not implementation-specific
Context Agent findings:
- ExApp with MCP server endpoint (~28 tools exposed)
- Uses Task Processing API for confirmations (NOT MCP elicitation)
- Works around AppAPI proxy limitations by changing protocol
- MCP endpoint is secondary feature with documented constraints
- Primary use: In-app Assistant integration, not external MCP clients
Critical features impossible through AppAPI proxy:
- ❌ MCP sampling (eliminates RAG/LLM features)
- ❌ MCP elicitation (user prompts)
- ❌ Real-time progress updates
- ❌ Bidirectional streaming
- Validated by Context Agent facing same limitations
Decision rationale:
- MCP requires multi-turn nested interactions
- AppAPI provides stateless request/response proxy only
- No implementation effort can bridge this fundamental gap
- Would require complete AppAPI redesign (WebSocket, message routing)
- Even official Nextcloud projects work around these limitations
Alternative considered for future:
- Register as Task Processing provider (different product)
- Use Nextcloud Assistant UI (not external MCP clients)
- Accept different capabilities (no sampling, custom flows)
OAuth mode remains sole solution for external MCP client integration.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This ADR documents the architectural decision to support both OAuth and
AppAPI (ExApp) deployment modes in a single codebase with 90%+ code sharing.
Key additions:
- Comprehensive analysis of AppAPI limitations and challenges
- Feature parity matrix comparing OAuth vs AppAPI modes
- Resolution of critical open questions via research:
* Non-browser client authentication (app passwords/OAuth)
* Streaming transport compatibility (buffered, not real-time)
* Callbacks/webhooks (MCP notifications not possible in AppAPI)
- Detailed implementation plan with 4 phases (10 days)
- Mode-aware architecture with abstraction layer
Critical findings:
- AppAPI mode does NOT support MCP sampling (RAG features)
- No real-time progress updates (use Nextcloud notifications)
- Buffered streaming only (Streamable HTTP works, WebSocket doesn't)
- Requires app password support in AppAPI proxy
Deployment mode selection:
- OAuth: Multi-tenant, external clients, sampling/RAG, real-time updates
- AppAPI: Single-tenant, simplified install, native UI, admin-controlled
Related to investigation of ~/Software/app_api/ and ~/Software/nc_py_api/
for AppAPI integration patterns.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Simplified the webapp routing structure by consolidating the admin UI
to a single clean endpoint.
Changes:
- Moved webapp from /user/page to /app (root of mount)
- Removed /user JSON endpoint (no longer needed)
- Updated mount point from /user to /app in app.py
- Updated all route path checks (3 locations)
- Updated OAuth redirects to point to /app
- Updated all HTMX endpoint references
- Updated documentation (ADR-007, CHANGELOG)
- Added redirect from /app to /app/ for trailing slash handling
New Route Structure:
- /app - Main webapp (HTML UI with tabs)
- /app/revoke - Revoke background access
- /app/webhooks - Webhook management UI
- /app/webhooks/enable/{preset_id} - Enable webhook preset
- /app/webhooks/disable/{preset_id} - Disable webhook preset
Breaking Change: Existing bookmarks to /user or /user/page will no longer work.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Manual testing of Nextcloud webhook_listeners app to validate webhook
payloads against ADR-010 expected schemas and document implementation
requirements for webhook-based vector synchronization.
## Changes
- Add test webhook endpoint at /webhooks/nextcloud in app.py
- Captures and logs webhook payloads for analysis
- Returns 200 OK immediately for webhook delivery confirmation
- Create webhook-testing-findings.md with comprehensive test results
- Captured payloads for 5/6 webhook event types
- Critical findings: missing node.id in deletions, type mismatches
- Implementation recommendations with code examples
- Update ADR-010 with Appendix A: Manual Webhook Testing Results
- Document actual vs expected webhook behavior
- Update event mapping table with tested webhook status
- Add 6 specific implementation recommendations
- Include testing implications for future development
## Testing Results
✅ NodeCreatedEvent - fires correctly, includes node.id (integer)
✅ NodeWrittenEvent - fires correctly, includes node.id (integer)
✅ NodeDeletedEvent - fires but missing node.id field (path only)
✅ CalendarObjectCreatedEvent - fires correctly with full iCal
✅ CalendarObjectUpdatedEvent - fires correctly with full iCal
❌ CalendarObjectDeletedEvent - does not fire (potential NC bug)
## Key Findings
1. NodeDeletedEvent missing node.id field - requires path-based fallback
2. node.id returns integer not string - needs casting for consistency
3. Multiple webhooks fire per operation - needs deduplication logic
4. Calendar deletion webhooks don't fire - reported as issue #53497
5. Calendar webhooks include full iCal content - enables rich parsing
## GitHub Issues
- Created issue #56371: NodeDeletedEvent missing node.id field
- Commented on issue #53497: CalendarObjectDeletedEvent not firing
Closes#283
---
_This commit was generated with the help of AI, and reviewed by a Human_
Simplifies the OpenTelemetry tracing setup by removing the redundant
OTEL_ENABLED flag and using the presence of OTEL_EXPORTER_OTLP_ENDPOINT
to determine if tracing should be enabled. This follows the standard
OpenTelemetry environment variable conventions more closely.
Changes:
- Remove OTEL_ENABLED/tracing_enabled flag in favor of checking if
OTEL_EXPORTER_OTLP_ENDPOINT is set
- Add OTEL_EXPORTER_VERIFY_SSL configuration option for OTLP endpoints
with self-signed certificates (defaults to false for development)
- Move HTTPXClientInstrumentor initialization to module level to ensure
httpx calls are traced across all Nextcloud API requests
- Add tracing spans to vector sync operations (scan_user_documents)
- Fix authorization header logging to only warn about missing headers
in OAuth mode (BasicAuth mode doesn't use Authorization headers)
- Update observability documentation to reflect simplified configuration
- Refactor Dockerfile to use --no-editable flag for uv sync
Breaking changes:
- OTEL_ENABLED environment variable is removed
- Tracing is now automatically enabled when OTEL_EXPORTER_OTLP_ENDPOINT
is set
Migration guide:
- Remove OTEL_ENABLED=true from environment configuration
- Tracing will be enabled automatically if OTEL_EXPORTER_OTLP_ENDPOINT
is configured
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add architecture decision record for integrating Nextcloud webhooks
into the vector database synchronization system.
Key features:
- Webhook endpoint at /webhooks/nextcloud receives push notifications
- Complements existing polling (ADR-007) without replacing it
- Optional authentication via WEBHOOK_SECRET
- Simple architecture: webhooks are just another DocumentTask producer
- Administrators can reduce polling frequency when webhooks are configured
Benefits:
- Reduced latency: seconds to minutes instead of up to 1 hour
- Lower API load: ~95% reduction when polling frequency is increased
- Better scalability: only process changed documents
- No changes required to scanner or processor components
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This PR enables safe switching between embedding models and multi-server
deployments by implementing auto-generated Qdrant collection names based on
deployment ID and model name.
## Problem
Previously, all deployments used a single hardcoded collection name
"nextcloud_content", which caused two critical issues:
1. **Dimension mismatches when switching models**: Changing
OLLAMA_EMBEDDING_MODEL (e.g., nomic-embed-text at 768D → all-minilm at
384D) would cause runtime errors as vectors couldn't be inserted into a
collection with incompatible dimensions.
2. **Collection collisions in multi-server setups**: Multiple MCP servers
sharing a single Qdrant instance would overwrite each other's data,
making horizontal scaling impossible.
## Solution
### Auto-Generated Collection Naming
Collections are now automatically named using the pattern:
\`{deployment-id}-{model-name}\`
**Deployment ID**: Uses \`OTEL_SERVICE_NAME\` if configured (and not default
value), otherwise falls back to \`hostname\` for simple Docker deployments.
**Model Name**: From \`OLLAMA_EMBEDDING_MODEL\` with path separators sanitized.
**Examples**:
- \`my-mcp-server-nomic-embed-text\` (with OTEL_SERVICE_NAME=my-mcp-server)
- \`mcp-container-all-minilm\` (simple Docker, hostname=mcp-container)
**Override**: Users can still set \`QDRANT_COLLECTION\` explicitly to bypass
auto-generation for backward compatibility.
### Dimension Validation
Added startup validation that checks collection dimensions match the
embedding service. If a mismatch is detected, the server fails fast with a
clear error message explaining:
- Expected vs actual dimensions
- Likely cause (model change)
- Solutions (delete collection, use different name, or revert model)
### Improved Sampling Error Handling
Enhanced MCP sampling rejection handling to treat user rejections as normal
behavior rather than errors:
- **User rejections** ("rejected", "denied") → INFO log, no traceback
- **Unsupported clients** → INFO log, no traceback
- **Other MCP errors** → WARNING log, no traceback
- **Unexpected errors** → ERROR log WITH traceback
This aligns with the MCP specification where clients SHOULD prompt users for
approval/denial of sampling requests.
## Changes
### Core Implementation
- **nextcloud_mcp_server/config.py**: Added \`get_collection_name()\` method
with deployment ID detection and model name sanitization
- **nextcloud_mcp_server/vector/qdrant_client.py**: Dimension validation on
collection open with helpful error messages
- **nextcloud_mcp_server/vector/{scanner,processor}.py**: Updated to use
\`get_collection_name()\`
- **nextcloud_mcp_server/auth/userinfo_routes.py**: Vector sync status uses
\`get_collection_name()\`
- **nextcloud_mcp_server/server/semantic.py**:
- Updated semantic search tools to use \`get_collection_name()\`
- Improved sampling rejection error handling (McpError vs Exception)
### Documentation
- **docs/semantic-search-architecture.md**: New comprehensive architecture
document (557 lines) covering background sync, semantic search flow, RAG
implementation, and deployment modes
- **docs/configuration.md**: Added detailed "Qdrant Collection Naming"
section with examples and multi-server deployment guidance
- **docker-compose.yml**: Added comments explaining collection naming behavior
- **README.md**: Updated semantic search descriptions to clarify
experimental status, Notes-only support, and infrastructure requirements
## Migration Guide
**For existing single-server deployments:**
Option 1 (Recommended): Use explicit collection name for continuity
\`\`\`bash
QDRANT_COLLECTION=nextcloud_content # Keep existing collection
\`\`\`
Option 2: Allow auto-generation and re-embed
\`\`\`bash
# Remove QDRANT_COLLECTION override
# New collection will be created based on deployment ID + model
# Requires re-embedding all documents (may take time)
\`\`\`
**For new multi-server deployments:**
Set unique OTEL service names per server:
\`\`\`bash
# Server 1
OTEL_SERVICE_NAME=mcp-prod
OLLAMA_EMBEDDING_MODEL=nomic-embed-text
# → Collection: "mcp-prod-nomic-embed-text"
# Server 2
OTEL_SERVICE_NAME=mcp-staging
OLLAMA_EMBEDDING_MODEL=nomic-embed-text
# → Collection: "mcp-staging-nomic-embed-text"
\`\`\`
## Benefits
✅ **Safe model switching**: Each model gets its own collection, preventing
dimension mismatch errors
✅ **Multi-server support**: Multiple MCP servers can share one Qdrant
instance without conflicts
✅ **Clear ownership**: Collection names show which deployment and model owns
the data
✅ **Better error messages**: Dimension validation provides actionable
guidance
✅ **Backward compatible**: Existing deployments can continue using
\`QDRANT_COLLECTION\` override
## Testing
Validated with:
- Single-server deployments (default hostname-based naming)
- Multi-server deployments (OTEL service name-based naming)
- Model switching scenarios (dimension validation)
- Collection override scenarios (backward compatibility)
Next steps: Testing various Ollama embedding models to investigate optimal
chunk sizes and performance characteristics.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add Prometheus metrics for HTTP, MCP tools, Nextcloud API, OAuth, vector sync, and DB operations
- Add OpenTelemetry distributed tracing with OTLP export
- Add structured JSON logging with trace context correlation
- Add ObservabilityMiddleware for automatic HTTP instrumentation
- Add app_name attribute to all client classes for per-app metrics
- Add configuration for metrics, tracing, and logging via environment variables
- Add documentation in docs/observability.md
- Fix graceful degradation when tracing is disabled (default state)
- Fix uvicorn logging configuration to use observability formatters
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Adds flexible Qdrant deployment modes to reduce infrastructure requirements
for local development and smaller deployments:
**Configuration Changes:**
- Add QDRANT_LOCATION environment variable (mutually exclusive with QDRANT_URL)
- Three modes: network (URL), in-memory (:memory:, default), persistent (file path)
- Settings dataclass validation via __post_init__ ensures mutual exclusivity
- API key warning when set in local mode (ignored, only for network mode)
**Client Initialization:**
- Auto-detect mode: network (url + api_key) vs local (:memory: or path=)
- In-memory: AsyncQdrantClient(":memory:") - zero config default
- Persistent: AsyncQdrantClient(path="/app/data/qdrant") - file storage
- Network: AsyncQdrantClient(url, api_key) - production mode
**Docker Compose Updates:**
- Qdrant service moved to optional profile (--profile qdrant)
- MCP service uses QDRANT_LOCATION=:memory: by default
- Added mcp-data volume for persistent storage (/app/data)
- No hard dependency on qdrant service
**Documentation:**
- Comprehensive configuration guide in docs/configuration.md
- All three modes documented with pros/cons
- Docker Compose examples for each mode
- Environment variable reference table
**Tests:**
- 13 new config validation tests (mutual exclusivity, defaults, warnings)
- Persistent mode integration test (create, close, reopen, verify persistence)
- All 82 unit tests + 5 smoke tests pass
**Breaking Change:**
- Default changed from QDRANT_URL=http://qdrant:6333 to QDRANT_LOCATION=:memory:
- Simplifies local development (no external service needed)
- Production deployments: explicitly set QDRANT_URL or QDRANT_LOCATION
Related: ADR-007 background vector sync implementation
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Replace static VECTOR_SYNC_ENABLED_APPS environment variable with per-user
database storage for which apps to index. This allows each user to control
their own indexing preferences (e.g., enable notes and calendar but not
deck or files).
Rationale:
- Nextcloud doesn't support granular OAuth scopes at the app level
- Per-user settings provide flexibility for multi-user deployments
- Users control app enablement via nc_enable_vector_sync MCP tool
- Aligns with OAuth architecture where users manage their own settings
Changes:
- ADR-007: Remove VECTOR_SYNC_ENABLED_APPS from configuration section
- ADR-007: Update scanner implementation to read from database
- ADR-007: Add explanation of per-user app enablement mechanism
- ADR-007: Clarify that nc_enable_vector_sync tool manages this setting
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Update ADRs to reflect that vector database and semantic search support
multiple Nextcloud apps (notes, calendar, deck, files, contacts) rather
than being notes-specific. Introduce semantic:read/write OAuth scopes
to replace app-specific scope requirements for cross-app search.
Changes:
- ADR-007: Add plugin architecture (DocumentScanner, DocumentProcessor,
DocumentVerifier) for multi-app vector sync
- ADR-008: Rename tools from nc_notes_semantic_* to nc_semantic_*, update
scope from notes:read to semantic:read
- ADR-009: NEW - Document decision to use generic semantic:read scope
with dual-phase authorization instead of requiring all app scopes
- oauth-architecture.md: Add semantic:read/write scope documentation
- README.md: Move semantic search to dedicated section separate from Notes
This is a breaking change that correctly positions semantic search as a
cross-app capability before broader adoption. Existing deployments will
need to re-authenticate with the new semantic:read scope.
Relates to user request to decouple vector database from notes-only model
and establish proper OAuth scope boundaries for multi-app semantic search.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add nc_notes_semantic_search_answer tool that combines semantic search
with MCP sampling to generate natural language answers from retrieved
Nextcloud Notes. This enables Retrieval-Augmented Generation (RAG)
patterns without requiring a server-side LLM.
Key features:
- Client-side LLM generation via ctx.session.create_message()
- Graceful fallback when sampling unavailable
- Proper source citations in generated answers
- No results optimization (skips sampling when no docs found)
- Comprehensive unit and integration tests
Implementation details:
- SamplingSearchResponse model with generated_answer and sources
- Fixed prompt template with document context and citation instructions
- Model preferences hint Claude Sonnet for balanced performance
- Falls back to returning documents without answer on sampling failure
Updates:
- Add ADR-008 documenting sampling architecture decision
- Add MCP sampling pattern guidance to CLAUDE.md
- Update README.md and docs/notes.md (7 → 9 tools)
- Add 4 unit tests and 6 integration tests
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Add comprehensive ADR-007 documenting background vector database
synchronization architecture using anyio TaskGroups for in-process
concurrency. This supersedes ADR-003's conceptual background worker.
Key decisions:
- In-process architecture using anyio TaskGroups (not Celery)
- Scanner task runs hourly, detects changes via timestamp comparison
- In-memory asyncio.Queue for pending documents
- Pool of 3 concurrent processor tasks for I/O-bound embedding workloads
- Qdrant metadata as single source of truth for indexing state
- Simple user controls: enable/disable with status visibility
Benefits:
- Single container deployment (was 3: mcp, celery-worker, celery-beat)
- No distributed task queue infrastructure
- Shared process state (no volume coordination)
- Sufficient throughput for I/O-bound embedding APIs
- Simpler debugging and deployment
Update ADR-003 status to "Superseded by ADR-007" with reference link.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This PR fixes multiple OAuth-related issues:
## Unified OAuth Callback
- Consolidated `/oauth/callback-nextcloud` and `/oauth/login-callback` into single `/oauth/callback` endpoint
- Flow type determined by session lookup via state parameter (no query params in redirect_uri)
- Fixes redirect_uri validation issues with IdPs requiring exact match
- Legacy endpoints kept as aliases for backwards compatibility
## PKCE Implementation
- Implemented PKCE (RFC 7636) for Flow 2 (resource provisioning)
- Generate code_verifier and code_challenge
- Store code_verifier in session storage
- Retrieve and use in token exchange
- Fixed PKCE for browser login (integrated mode)
- Previously only worked for external IdP (Keycloak)
- Now works for both Nextcloud OIDC and external IdP
## Login Elicitation Fixes (ADR-006)
- Fixed elicitation URL to route through MCP server endpoint
- Changed from direct Nextcloud URL to `/oauth/authorize-nextcloud`
- Ensures PKCE is properly handled by server
- Fixed login detection after OAuth flow completes
- Look up refresh token by state parameter instead of user_id
- Works even when Flow 1 token not present
- Added `get_refresh_token_by_provisioning_client_id()` method
## Session Authentication
- Fixed `/user/page` redirect loop
- Shared oauth_context with mounted browser_app
- SessionAuthBackend can now validate sessions correctly
## Tests
- Added comprehensive login elicitation test suite
- Updated scope authorization test expectations
- All 43 OAuth tests passing
## Files Changed
- `app.py`: Shared oauth_context, unified callback route
- `oauth_routes.py`: Unified callback, PKCE for Flow 2
- `browser_oauth_routes.py`: PKCE for integrated mode
- `oauth_tools.py`: Fixed elicitation URL generation
- `refresh_token_storage.py`: Added lookup by provisioning_client_id
- `test_login_elicitation.py`: New test suite
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This commit completes the OAuth audience validation implementation per RFC 7519,
RFC 8707 (Resource Indicators), and RFC 9728 (Protected Resource Metadata).
## Key Changes
### OAuth Resource Parameters (RFC 8707)
- Add `resource` parameter to Flow 1 (MCP client auth) with MCP server audience
- Add `resource` parameter to Flow 2 (Nextcloud access) with Nextcloud audience
- Add `nextcloud_resource_uri` to oauth_context configuration
- Fix undefined variable error in starlette_lifespan
### PRM-Based Resource Discovery (RFC 9728)
- Update tests to fetch resource identifier from PRM endpoint
- Add fallback to hardcoded value if PRM fetch fails
- Demonstrate correct OAuth client implementation pattern
### ADR-005 Documentation Updates
- Update to reflect simplified RFC 7519 compliant implementation
- Document that MCP validates only its own audience (not Nextcloud's)
- Add section on OAuth resource parameters and PRM discovery
- Update implementation checklist to show completed items
- Mark status as "Implemented" with update date
## Implementation Details
The solution follows RFC 7519 Section 4.1.3: resource servers validate only
their own presence in the audience claim. This simplifies the logic while
maintaining security:
- MCP server validates MCP audience only
- Nextcloud independently validates its own audience
- No dual validation required at MCP layer
- Token reuse is allowed per RFC 8707 for multi-audience tokens
## Test Results
✅ test_mcp_oauth_server_connection - PASSED
✅ test_deck_board_view_permissions - PASSED
✅ test_prm_endpoint - PASSED
All OAuth flows now properly specify target resources, resulting in correct
audience validation throughout the system.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Per RFC 7519 Section 4.1.3, resource servers should only validate their
own presence in the audience claim, not check for other resource servers.
Changes:
- UnifiedTokenVerifier now validates only MCP audience (not Nextcloud's)
- Nextcloud independently validates its own audience when receiving API calls
- This is NOT token passthrough (we validate tokens before use)
- This IS token reuse which is explicitly allowed by RFC 8707
Updates:
- Simplified _validate_multi_audience() to follow OAuth spec
- Updated docstrings and comments to clarify RFC 7519 compliance
- Fixed unit tests that expected dual-audience validation
- Updated ADR-005 to document the correct OAuth interpretation
- All tests pass: unit (65), smoke (5), OAuth integration
This makes the implementation simpler, more maintainable, and properly
aligned with OAuth 2.0 specifications while maintaining security.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This ADR addresses the critical token passthrough vulnerability identified
in Issue #261 by proposing a unified token verifier that eliminates the
security issue while maintaining flexibility.
Key changes:
- Consolidates two non-compliant verifiers into single UnifiedTokenVerifier
- Implements two-layer architecture (verification + exchange)
- Supports multi-audience mode (default) and token exchange mode (opt-in)
- Removes all token passthrough paths to comply with MCP security spec
- Works within python-sdk constraints using proper separation of concerns
The solution provides:
- Single source of truth for token validation
- MCP specification compliance
- Minimal performance impact (1-2% of LLM request time)
- Clear migration path for existing deployments
BREAKING CHANGE: All OAuth deployments must be reconfigured to specify
resource URIs (NEXTCLOUD_MCP_SERVER_URL and NEXTCLOUD_RESOURCE_URI) and
choose between multi-audience or token exchange mode.
Related: #261
Supersedes: Token passthrough mode in ADR-004
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Resolves the token exchange implementation gap where get_session_client()
was implemented but never used by tools. Unifies token acquisition into a
single async get_client() method that handles both pass-through and token
exchange modes transparently.
Core Changes:
- Make get_client() async and merge token exchange logic into it
- Remove scopes parameter from token exchange (Nextcloud doesn't support OAuth scopes)
- Update all 8 tool modules to use await get_client(ctx)
- Fix provisioning decorator to skip checks in BasicAuth mode
Token Acquisition Modes:
1. BasicAuth: Returns shared client (no token operations)
2. OAuth pass-through (default): Verifies and passes Flow 1 token to Nextcloud
3. OAuth token exchange (opt-in): Exchanges Flow 1 token for ephemeral token via RFC 8693
Key Architectural Clarifications:
- Progressive Consent (Flow 1/2) = Authorization architecture
- Token Exchange = Token acquisition pattern during tool execution
- Refresh tokens from Flow 2 are NEVER used for tool calls (only background jobs)
- Nextcloud scopes are "soft-scopes" enforced by MCP server, not IdP
Documentation Updates:
- ADR-004: Added comprehensive token acquisition patterns section
- CRITICAL-TOKEN-EXCHANGE-PATTERN.md: Updated to reflect implementation status
- CLAUDE.md: Updated architectural patterns with async get_client()
Testing:
- All 36 unit tests passing
- All 4 smoke tests passing (BasicAuth mode)
- Linting issues fixed (ruff)
Configuration:
ENABLE_TOKEN_EXCHANGE=false (default) - pass-through mode
ENABLE_TOKEN_EXCHANGE=true (opt-in) - token exchange mode
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Documents the architectural flaw in current implementation where
session tokens and background tokens are not properly separated.
Key issues identified:
- Session tokens should be exchanged on-demand (RFC 8693)
- Background tokens should use separate refresh token grant
- Current implementation reuses refresh tokens incorrectly
- No separation between foreground and background operations
This is a P0 blocker that must be fixed before production use.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Replace hybrid flow model with true progressive consent where MCP client authenticates directly to IdP (Flow 1) and server requests separate explicit provisioning for Nextcloud access (Flow 2). This separates client authentication from resource authorization, uses distinct client_id for each flow, and keeps server stateless by default until user explicitly grants offline access via provision_nextcloud_access tool.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Refactor ADR-004 to document the proper OAuth architecture where MCP
clients are registered at the IdP level (not with MCP server) and use
a progressive consent pattern with dual OAuth flows.
## Key Changes
### MCP Client Registration
- Document that MCP clients (Claude Desktop, etc.) register at IdP level
- Show DCR and pre-registration options
- Clarify client validation happens against IdP registry
### Progressive Consent Architecture
Replace single "Hybrid Flow" with three-phase progressive consent:
**Phase 1: MCP Client Authentication** (Always)
- MCP client uses own client_id (e.g., "claude-desktop")
- User consents to "Claude Desktop accessing MCP Server"
- MCP server validates client exists at IdP
- Stores MCP client access token
**Phase 2: Nextcloud Consent** (Conditional)
- Only if MCP server doesn't have refresh token for user
- MCP server uses own client_id ("nextcloud-mcp-server")
- User consents to "MCP Server accessing Nextcloud offline"
- MCP server stores master refresh token
- SSO: If already authenticated, only consent needed
**Phase 3: Token Exchange** (Standard PKCE)
- Client exchanges MCP authorization code
- Validates PKCE code_verifier
- Returns access token (aud: mcp-server)
- Client never sees master refresh token
### Implementation Status Section
- Document current implementation as "simplified hybrid flow"
- List what's implemented vs what needs refactoring
- Clarify current tests use simplified version
- Note progressive consent is target architecture
## Benefits of Progressive Consent
✅ Standards-compliant: Proper OAuth clients at IdP level
✅ Secure: Client validation against IdP registry
✅ Efficient: Nextcloud consent only once per user
✅ Transparent: Users understand each authorization step
✅ SSO-friendly: Minimal re-authentication in Phase 2
## Implementation Tracking
The refactoring from simplified hybrid flow to progressive consent will
be tracked in a separate issue. Current implementation demonstrates:
- MCP server can intercept OAuth callbacks
- Refresh tokens stored securely
- PKCE flow works end-to-end
- Tool execution succeeds
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Critical architectural corrections to properly implement secure token brokering:
## Key Changes:
1. **Removed Dual Token Concept**: MCP server no longer generates its own JWTs.
Instead, it acts as a token broker using IdP-issued tokens with proper
audience validation.
2. **Strict Audience Isolation**:
- Tokens with `aud: "mcp-server"` can ONLY authenticate to MCP server
- Tokens with `aud: "nextcloud"` can ONLY access Nextcloud APIs
- No tokens have multiple audiences (security boundary violation)
- Compromised MCP tokens cannot access Nextcloud directly
3. **Linked Authorization Pattern**: Single OAuth flow obtains a master
refresh token capable of minting tokens for different audiences as needed.
This solves the challenge of needing both MCP authentication and Nextcloud
access from a single user authorization.
4. **Token Broker Implementation**:
- Validates incoming tokens have `audience: "mcp-server"`
- Uses stored refresh tokens to obtain `audience: "nextcloud"` tokens
- Never exposes Nextcloud tokens to MCP clients
- Maintains short-lived cache for performance
5. **PKCE and Native Client Updates**:
- Proper 302 redirects (no HTML pages)
- Complete PKCE verification in token endpoint
- IdP tokens returned directly (not MCP-generated)
6. **Security Enhancements**:
- Comprehensive audience validation examples
- Token exchange pattern documentation
- Keycloak configuration for audience mapping
- Trust boundary diagrams
This architecture maintains strict security boundaries while enabling the
MCP server to act on behalf of users for both authentication and resource
access, following OAuth best practices and enterprise security standards.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Major rewrite of ADR-004 to reflect federated authentication pattern with
shared identity provider (IdP) instead of direct Nextcloud authentication.
Key changes:
- Replaced "Sign-in with Nextcloud" with "Federated Authentication"
- Added shared IdP (Keycloak, Okta, Azure AD) as central auth provider
- MCP server now acts as OAuth client to shared IdP, not Nextcloud
- Single user authentication grants both identity and Nextcloud access
- Updated all diagrams to show 4-party architecture
- Removed authorize_nextcloud tool - uses standard 401 flow
- Added proper token rotation with reuse detection
- Clarified Pattern 3 vs Pattern 4 differences in comparison doc
- Pattern 3 can use external IdPs via user_oidc (not limited to NC)
Architecture benefits:
- True single sign-on with enterprise IdP support
- OAuth-compliant on-behalf-of pattern
- Supports SAML/LDAP backends through IdP
- Nextcloud validates IdP tokens, not MCP-specific tokens
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
- Supersedes ADR-002 which fundamentally misunderstood MCP protocol constraints
- Introduces "Sign-in with Nextcloud" architecture pattern
- MCP server becomes OAuth client to enable offline/background operations
- Implements full token rotation with reuse detection for security
- Includes comprehensive implementation details and migration strategy
Key architectural shift:
- From: Pass-through authentication (stateless, no offline access)
- To: MCP server as OAuth client (stateful, full offline capabilities)
The solution enables background workers to operate independently of MCP
sessions by storing and rotating refresh tokens securely.
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
This commit implements and documents both RFC 8693 token exchange tiers
from ADR-002, enabling both production-ready delegation and advanced
impersonation capabilities.
- Enable Keycloak preview features (`--features=preview`) to support
both Standard V2 and Legacy V1 token exchange modes
- Update Tier 1 status from "NOT IMPLEMENTED" to "IMPLEMENTED (Legacy V1)"
- Add detailed empirical testing results showing:
- Standard V2 rejects `requested_subject` parameter
- Legacy V1 accepts parameter but requires impersonation permissions
- Complete configuration steps for enabling impersonation
- Add comparison table showing when to use each tier
- Add "When to Use" guidance for both tiers
- Document that Tier 2 (Delegation) is the recommended default
- Update docstring to document both Tier 1 and Tier 2 support
- Add tier-specific logging (shows which tier is being used)
- Document permission requirements for Tier 1 impersonation
**tests/integration/auth/test_token_exchange_standard_v2.py**:
- Test delegation without impersonation (Tier 2)
- Verify sub claim remains unchanged (service account identity)
- Verify no special permissions required
- Test exchanged tokens work with Nextcloud APIs
- All tests PASS ✅
**tests/integration/auth/test_token_exchange_legacy_v1.py**:
- Test impersonation with `requested_subject` (Tier 1)
- Verify sub claim changes to target user
- Auto-skip if impersonation permissions not configured
- Document permission requirements in test docstrings
- Test exchanged tokens work with Nextcloud APIs
**tests/manual/test_impersonation.py**:
- Comprehensive impersonation validation script
- Tests both Standard V2 and Legacy V1 behavior
- Decodes JWT tokens to verify sub claim changes
- Validates tokens against Nextcloud APIs
**tests/manual/configure_impersonation.py**:
- Automated permission configuration helper
- Documents manual Keycloak CLI configuration steps
Both token exchange tiers are now fully implemented and tested:
- **Tier 2 (Delegation)** - ✅ RECOMMENDED
- Standard V2 (production-ready)
- No special permissions required
- Service account identity preserved
- **Tier 1 (Impersonation)** - ✅ Advanced use only
- Legacy V1 (--features=preview required)
- Requires manual permission grant via Keycloak CLI
- Subject claim changes to target user
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Chris Coutinho