agentleFS
andymaiandymai/claude-agent-team/skills/codex-write/SKILL.md

codex-write

Skill4 starsChanged 56 days agoMITView on GitHub

Write outward-facing prose through the OpenAI Codex CLI instead of directly. Use when the deliverable is text another person will read: a design or shaping doc, a proposal, a wiki or Notion page, a Slack message, a PR title, body, or reply, a review reply, an issue or ticket, an email. You gather the facts and decisions into a brief, Codex writes, you fact-check the draft against the brief and run the house-rule checker, then show the draft before anything is published. Not for code, tests, prompt files, commit messages, or replies to the user in the session.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. eprouveze/claude-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the checker fails, add the rule to the brief and rerun rather than hand-editing.

    line 57
  2. Distinctive:

    Read the draft against the brief one sentence at a time.

    line 61
  3. Distinctive:

    Every sentence in the draft traces to a line in the brief.

    line 74
  4. Distinctive:

    No content in the brief or draft is confidential to a project the venue cannot see.

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Codex Write

The point is the model's writing. You own the facts, the decisions, and the fact-check. Codex owns the sentences, so the brief constrains content, not style, beyond the user's own rules. Nothing is published until the user has seen the draft and said so.

Preconditions

  • codex is on PATH and logged in. Check with codex login status. If it is not, stop and tell the user; do not write the piece yourself as a fallback unless they say to.
  • The user's writing rules are wherever this project keeps them (CLAUDE.md, AGENTS.md, memory). Read them before writing the brief; they go into the brief verbatim.

Steps

1. Research

Collect every fact the piece needs, each with a source you observed this session: a file and line you opened, a query you ran, a URL you fetched, a message you read. Make the judgment calls yourself and write down the reason for each. Do not proceed until every fact has a source. Codex gets no credit for anything you did not give it.

2. Brief

Fill in ${CLAUDE_PLUGIN_ROOT}/skills/codex-write/reference/brief-template.md. State the audience, the purpose, the venue's structure if it has one, the facts with sources, the decisions with reasons, the user's own writing rules copied verbatim, and a word target. Do not add style rules of your own. End the brief by asking for the finished text only, no preamble. Save it to a temp file, for example /tmp/codex-write/<slug>.brief.md.

3. Run

${CLAUDE_PLUGIN_ROOT}/skills/codex-write/scripts/run.sh BRIEF.md OUT.md --max N

run.sh calls codex exec non-interactively, writes only the final message to OUT.md, then runs scripts/check.py on it. The checker fails only on em or en dashes. It prints the word count and notes when the target is exceeded, without failing. It never prompts, so no tmux is needed. Add -C <dir> after the targets only when Codex must read project files itself.

If the checker fails, add the rule to the brief and rerun rather than hand-editing.

4. Fact-check

Read the draft against the brief one sentence at a time. Any fact, number, name, or claim that is not in the brief is removed, or verified the same way as in step 1 and added to the brief. Any fact from the brief that the draft changed is corrected. Quote the offending sentence when you report a fix.

5. Show, then publish

Show the user the draft in a quoted block with the checker's output line. Publish only when they say so. Publishing means posting, committing, or saving anywhere another person will read it.

Self-check before reporting

  • Every sentence in the draft traces to a line in the brief.
  • check.py printed PASS on the exact file you are showing.
  • The user's own writing rules appear in the brief verbatim, and no style rules of yours do.
  • No content in the brief or draft is confidential to a project the venue cannot see.
  • You have not published anything yet.

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.