agentleFS
KbediakoKbediako/evergreen-codex-skills/skills/consult-chatgpt-pro/SKILL.md

consult-chatgpt-pro

Skill5 starsChanged 2 months agoMITView on GitHub

Consult ChatGPT Pro when requested or when an unresolved, consequential question needs an external reasoning review. Reconcile its advice with local evidence.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use Pro as a consultant. Codex owns the decision, implementation, local validation, and completion of the user's task.

    line 8
  2. Very specific:

    Read context packets when selecting evidence or building and verifying the ZIP.

    line 14
  3. Distinctive:

    Read reconciliation and closure when acting on advice, preparing re-review, or deciding closure.

    line 16
  4. Distinctive:

    Verify material recommendations locally.

    line 22

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Consult ChatGPT Pro

Use Pro as a consultant. Codex owns the decision, implementation, local validation, and completion of the user's task. Pro can assess only the material supplied in its current conversation; do not imply it inspected omitted files or ran local checks.

Prepare and consult

Define the question, scope, success criteria, available evidence, and requested verdict. Keep optional suggestions outside the task unless the user expands it.

  • Read context packets when selecting evidence or building and verifying the ZIP. Include only authorized task evidence. For rollout-derived evidence, use the typed exporter in native-agent-evals; never attach a source rollout or conversation export.
  • Read ChatGPT browser execution before selecting a browser or mode, transmitting evidence, waiting, or collecting a response. Follow the current Browser control skill and live runtime documentation, open a fresh tab and conversation for each substantive review, and verify the composer surface mode and exact visible model label before sending. When Pro is required, Pro must be visibly selected; do not infer it from the request, profile, or account UI.
  • Read reconciliation and closure when acting on advice, preparing re-review, or deciding closure. It owns finding dispositions and recurring-failure review. Isolated findings permit focused re-review; an activated recurring-failure gate requires full-scope re-review.

Send the smallest complete verified packet. Wait for a substantive response without forcing an early answer. If required evidence cannot be supplied, narrow the verdict explicitly or report the consultation as BLOCKED.

Reconcile and finish

Verify material recommendations locally. Implement changes already authorized by the task, validate the affected result, and return current evidence for re-review while material findings remain. An internal review checkpoint does not require renewed user permission.

Keep a compact ledger of the ZIP SHA-256, the pre-send page URL and resulting durable conversation URL, fresh-tab status, selected surface mode (Chat or Work), exact visible model label, included evidence and omissions, findings and dispositions, local validation, and current state. Treat the surface mode and model label as separate fields. Update it on material changes rather than every wait.

Use CLOSED only when the latest substantive Pro verdict covers the declared scope, explicitly leaves no unresolved material finding, and every material finding has an evidence-backed resolved disposition. Missing evidence or an ambiguous verdict remains BLOCKED; use ABANDONED only when the user ends the consultation. A time limit or passing tests cannot substitute for the verdict.

Report the outcome and ledger, then complete any remaining authorized task work. Review closure alone does not finish the user's objective.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.