| name | loopify-codex-sdk |
| description | Use when adding Codex SDK scripts or templates as loop actuators for repair, review, context building, trace summarization, or meta-loop proposal steps. |
loopify-codex-sdk
Add Codex SDK actuators to loops.
When To Use
- A loop should call Codex SDK programmatically.
- A project needs a repair, review, evaluator, or trace-summarizer actuator.
- The loop needs resumable thread IDs, sandbox assumptions, or structured
outputs.
Do Not Use
- Do not treat a Codex SDK call as the whole loop.
- Do not write concrete SDK API calls from memory if current docs were not
verified.
- Do not hide sandbox, model, cost, or authority decisions.
Workflow
- Read the Loop Contract.
- Identify the actuator role: repair, review, evaluator, context builder,
trace summarizer, or meta-loop proposer.
- Verify current official SDK docs before concrete API calls.
- If verification is unavailable, use clearly marked placeholders.
- Scaffold the actuator from
templates/codex-repair-actuator.py or
templates/codex-review-actuator.md.
- Encode authority boundaries and structured output.
- Document that Codex SDK is the actuator, not the loop.
Output
Codex SDK actuator script/template and integration notes.
Validation
- The surrounding loop still owns checks, state, stop rules, and trace.
- Outputs are structured enough for the next loop step.
- Sandbox and authority assumptions are explicit.
Example
Use loopify-codex-sdk to scaffold a repair actuator for the failing check in .loopify/loops/001-papertrail/loop-contract.md.
