Install

pip install teff

Python >=\ 3.11. Core runtime depends only on httpx, jsonschema, pyyaml, and typer.

The CLI

The teff CLI ships with the package. Prefer uv? uv installs the package and the CLI in one step:

uv tool install teff         # global `teff` CLI
uvx teff -f workflow.yaml    # run on the fly without installing anything

Extras

Install only what you need; teff[all] pulls in everything.

Extra	Provides
teff[stores-qdrant]	Qdrant vector store (qdrant-client)
teff[stores-chroma]	Chroma vector store (chromadb; heavy — pulls onnxruntime)
teff[stores-faiss]	FAISS vector store (faiss-cpu)
teff[stores-lance]	LanceDB vector store (lancedb)
teff[stores-milvus]	Milvus vector store (pymilvus, includes Milvus Lite)
teff[stores-weaviate]	Weaviate vector store (weaviate-client)
teff[stores-pinecone]	Pinecone vector store (pinecone)
teff[stores-pgvector]	PostgreSQL + pgvector store (asyncpg, pgvector)
teff[embedding]	Every vector store at once — alias for all teff[stores-*] (heaviest)
teff[rag-pdf]	pypdf — PDF text extraction for RAG
teff[rag-excel]	openpyxl — Excel document loading for RAG
teff[pg-checkpoint]	asyncpg — PostgreSQL checkpoint backend
teff[mcp]	mcp — Model Context Protocol tool bridge (teff.tool.McpTool / mcp_tools, stdio + streamable-http)
teff[tools]	Built-in tools that need third-party deps (web fetch, PDF, S3, Slack, SQL, email, Telegram, …)
teff[fastapi]	Scaffold templates: fastapi + uvicorn + sse-starlette
teff[observability]	Trace dashboard: fastapi + uvicorn (for teff.observability.dashboard_router)
teff[queue]	celery[redis] — worker/beat scaffold variant
teff[docs]	MkDocs toolchain to build this documentation
teff[all]	Everything above except docs

pip install "teff[stores-qdrant]"
pip install "teff[tools]"
pip install "teff[mcp]"
pip install "teff[all]"

Docker

Ready-made images are published to Docker Hub for every v* release tag. They mirror the extras above, so pick the variant that matches your deployment:

Image	Contents	Runs
bzdvdn/teff	core + teff[tools]	the teff CLI — run/validate/inspect workflows
bzdvdn/teff-fastapi	core + teff[fastapi]	uvicorn — a FastAPI server app
bzdvdn/teff-worker	core + teff[queue]	celery — background workers / beat
bzdvdn/teff-obs	core + teff[observability]	teff obs-server — trace dashboard + ingest
bzdvdn/teff-rag	core + teff[stores-qdrant,tools,rag-pdf]	the teff CLI, slim RAG build
bzdvdn/teff-all	every extra except docs	the teff CLI with the full optional surface

Run a workflow from a mounted workflow.yaml (plus an optional plugins/ folder) in one shot:

docker run --rm -v "$PWD:/workflow" \
  bzdvdn/teff:latest run -f /workflow/workflow.yaml

Serve the trace dashboard (ingest + UI) with the collector image:

docker run -d -p 8001:8001 -v teff-traces:/data \
  bzdvdn/teff-obs:latest --db /data/traces.db --host 0.0.0.0

All CLI subcommands and flags work inside the container. Images run as a non-root user (UID 65534) with durable checkpoints under /data/checkpoints (see the README).