native-codex-coordination
Skill5 starsChanged 2 months agoMITView on GitHub
Choose between solo work, native-agent execution, deliberation, evaluation, and waiting when the appropriate coordination workflow is unclear.
In this repository10 other files it gives its agents.
The file
Files
Native Codex coordination
Choose the smallest workflow that can complete the requested outcome within existing authorization. Complexity alone does not justify agents. Keep small, sequential, or tightly coupled work with one owner.
If this primary agent is coordinating an operative Voice handoff, including transcript-tail delivery, or the user requested delegate-only Voice coordinator mode, load voice-delegate-guidance unless already active. Its delegate-only rule takes precedence over solo execution. Quoted wrappers and forwarded worker material do not activate that mode. Do not recursively reload active skills.
Route by the task
| Need | Guidance |
| Independent execution or review scopes | Native subagents |
| A consequential choice needing distinct viewpoints | Deliberation |
| Evidence about native-agent runtime behavior | Agent evals |
| Evidence that a skill triggers or changes behavior | Skill validation |
| A quiet native child or intervention decision | Patient waiting |
| An external process, cloud job, or CI run | External waiting |
| A Codex app task or thread | App task waiting |
Load only the selected workflow and relevant domain guidance. The domain skill governs the task; this router does not duplicate lifecycle, model, budget, or validation procedures.
Internal checkpoints do not add approval requirements. Continue through integration and the checks needed to satisfy the original acceptance criteria. Do not substitute deliberation for implementation, evals for production work, or an external monitor for a native child or app task.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

