Back to directory

Chidori

An agent framework in which every side effect an agent performs runs through the runtime as a recorded host call, so any run can be checkpointed, replayed with zero LLM calls, and resumed after a crash.

Visit website (opens in a new tab)thousandbirdsinc.github.io
Chidori

Chidori is an agent framework where every LLM call, tool call, and HTTP request an agent makes flows through the runtime as a recorded host call, rather than reaching the world directly. Because the runtime sees and records every side effect, it can log, cache, replay, pause on, and resume from any of them.

This gives runs several properties: a completed run can be replayed with zero LLM calls for identical output, useful for debugging and tests; runs are checkpointed at every host safepoint so a crash can be resumed in a new process; and a run can pause for a human or another agent to answer, without holding a process open, then continue later.

Agents are written as plain async TypeScript. The runtime is a single Rust binary with an embedded pure-Rust JavaScript engine and TypeScript and Python SDKs, with no Node, DSL, or native bindings required. It is Apache-2.0 licensed.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.