Open-source continuity for serious AI work

rate-limit-handoff

Turn usage limits into clean checkpoints. Preserve the next action, critical context, and verification state across Claude Code, Codex, Cursor, Grok, and Antigravity.

View on GitHub Install from source

One living handoff. Claude Code / OpenAI Codex / Cursor / Grok Build / Antigravity / Hermes

The checkpoint model

Keep the operating state, not just the transcript.

Rate limits are unavoidable. Lost context is not. rate-limit-handoff captures what the next capable session needs to continue without rediscovering the work.

01 / DECIDE

Choose deliberately

Continue on an approved lighter model or schedule the remaining work for the next full-capacity window.

02 / WRITE

Snapshot the state

Record the exact next action, decisions, constraints, changed files, test evidence, and unresolved risks.

03 / RESUME

Return with full context

Open the living handoff.md, confirm the workspace state, and continue from the named next move.

Public beta / MIT

Pure Python. No cloud account. No telemetry.

Install directly from GitHub, initialize a workspace, and keep the handoff with the project. Python 3.10 or later.

Read the documentation

terminal

$ pip install git+https://github.com/PurpleOrangeAI/rate-limit-handoff.git
$ rate-limit-handoff --init

$ rate-limit-handoff --schedule \
    --reset-at "04:00" \
    --model codex \
    --summary "Resume from the verified checkpoint"