chidori
Chidori's premise is that agents are non-deterministic, expensive and long-running, which makes them hard to debug and fragile in production. Its answer is a single boundary: agents are written as plain async TypeScript, and every LLM call, tool call and HTTP request goes through the runtime as a recorded host call.
Because the runtime sees every side effect, a run can be checkpointed at each safepoint and resumed in a new process after a crash, replayed byte-identically with zero LLM calls for debugging, or suspended to disk by chidori.input() or a named signal until a human answers, minutes or days later. A recorded run can be committed to Git as an integration test that costs nothing to run. Determinism is enforced with a fixed clock and seeded randomness.
It ships as one Rust binary with an embedded pure-Rust JavaScript engine, so there is no Node, Deno or V8 to install, and TypeScript and Python SDKs talk to it over HTTP. The trade-off is a new runtime with its own constraints: agent code has to go through chidori.* calls to get these guarantees.
Features
Why choose it
Trade-offs
Maintenance check
- Last commit
- 2026-10-22
- Commits, last 90 days
- 16
- Releases, last 90 days
- 1
- Latest release
- v3.8.1 (2026-08-30)
- Open issues
- 16
- Recent issues with at least one comment
- 0/3
Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.
Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.
Most discussed open issues
- #195 User memory as a replay-safe Chidori host call0 comments · 2026-09-29
- #192 Consider rusty_alloc in place of mimalloc0 comments · 2026-09-04
- #182 Fleet-mode resume of a paused session fails with `reading resume source ./__no_default_agent__.ts` when the persisted run state cannot be se0 comments · 2026-08-22
- #38 🌟 Chidori x WisePick - Replayable Routing & Deterministic Agent Semantics0 comments · 2026-05-25
Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.
Compatibility
Quick start
Use cases
Alternatives
Comments
No comments yet. Be the first!