OpenArmature

A workflow framework for LLM pipelines and tool-calling agents. Typed state, structural graph checks, and observability that doesn't require buy-in from every node.

Get started View on GitHub

---

•   Typed, frozen state

  ---

  State schemas are Pydantic models with frozen=True and extra="forbid". Nodes can't mutate state; they return partial updates and the engine merges via per-field reducers.

•   Compile-time checks

  ---

  Bad graph shapes (dangling edges, unreachable nodes, conflicting reducers, missing entry) fail at .compile(), not at run time.

•   Observable, opt-in

  ---

  Attach an Observer to see every node boundary. Drop in the optional OTel mapping for spans + log correlation; logs carry trace_id / span_id / correlation_id automatically.

•   Checkpointable

  ---

  In-memory and SQLite Checkpointer backends ship in core. Crash at node N+1, resume from node N's saved state on the next invocation.

•   First-class fan-out

  ---

  Per-instance fan-out with bounded concurrency, error-policy choice, and observability events that attribute correctly per instance.

•   Async-first, LLM-agnostic

  ---

  The engine has no concept of LLMs or tools; those live at the node boundary. Use any provider, any model, any external system.

---

Open specification

OpenArmature is defined by a public, language-agnostic specification, not a Python-shaped opinion exported to other languages. Reference implementations share conformance fixtures, so behavior stays identical across languages, runtimes, and tooling stacks.

Read the spec →