Recall

Recall is a local-first note-taking app for small, easy-to-forget notes.

The first implementation uses:

• Python and FastAPI for the API and web UI
• SQLite for local persistence
• A scheduled worker that enriches pending notes with elaboration and tags
• Docker Compose for local deployment

Services

• API server: CRUD, search, and bot control endpoints
• Web interface: browser-based note management
• Worker: scheduled enrichment of unprocessed notes
• Tags page: manage predefined tags used during enrichment
• Context page: edit the shared AI context file used during enrichment

Run locally

1. Build and start the stack with Docker Compose.
2. Open the web interface on port 8001.
3. Use the API on port 8000 for JSON access.

Docker Compose commands

Start all services in the background:

docker compose up -d

Stop all services:

docker compose down

Rebuild images and restart services:

docker compose up -d --build

Rebuild and restart a single service (example: worker):

docker compose up -d --build worker

Notes

• The original note is preserved.
• Enrichment output is stored separately from the source note.
• Search scans the original note, elaboration, and tags.

Worker LLM Configuration

The worker reads AI settings from a runtime YAML file at recall_data/llm_config.yaml (mounted in containers as /data/llm_config.yaml).

• If the file does not exist, the worker creates it on startup.
• Default provider is ollama.
• You can point to your network Ollama server by setting ollama.base_url.

Example:

provider: ollama
ollama:
	base_url: http://192.168.1.50:11434
	model: llama3.1:8b
	timeout_seconds: 30

If Ollama is unreachable or returns invalid JSON, enrichment is marked as error for that note with the failure reason in last_enrichment_error.

AI context file

The shared context file lives at recall_data/context.md and is mounted inside containers as /data/context.md.

• Use the Context page to add personal facts and preferences.
• The file is sent with every enrichment request.