agentleFS
Sign inStart free
zoranmarticzoranmartic/coord-wright/skills/coord-review/SKILL.md

coord-review

Skill0 starsChanged 47 days agoApache-2.0View on GitHub

Validate a shaped coord task against shaping quality checks (plan, acceptance, subtask metadata, model allowlist, high-risk keywords). Use after shaping a task or before promoting it to pending. Accepts an optional task id; infers the latest shaped task if omitted.

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

8 rules. About 0 are common, 4 are distinctive, and 4 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.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If a task id was provided, find it in the project task directory.

    line 15
  2. Distinctive:

    If not, list the shaping ids:

    line 16
  3. Very specific:

    If yes, apply targeted fixes via coord update (never edit the task file directly), then re-run coord-review and confirm exit 0 before …

    line 33
  4. Very specific:

    If reasoning_effort: xhigh is on the task, also ask:

    line 36
  5. Very specific:

    If LGTM, offer: "Ready to promote — run coord promote <id> to move to pending."

    line 42
  6. Distinctive:

    Never mutate the task to bypass a check.

    line 46
  7. Very specific:

    Do not run coord-review on tasks in pending, claude-working, or terminal statuses — those have already entered the work queue.

    line 47
  8. Distinctive:

    If the task file is not found (wrong id or task in a different project), report the error and ask the user to confirm the task id.

    line 48

≈ “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

Goal: catch shaping defects before a task enters the runnable queue. Run bin/coord-review on the task file and present any objections to the user with an inline fix offer. Print LGTM on a clean pass.

Input

Optional: task id (e.g. 2026-05-13-some-task-id). If omitted, infer the most recently updated task in shaping status.

Workflow

  1. Resolve the task file.
    • If a task id was provided, find it in the project task directory.
    • If not, list the shaping ids:
      python3 "${COORD_TOOLS:-$HOME/Projects/coord-wright}/bin/coord" list --status=shaping --format=ids
      Ids come oldest-first (date-prefixed filenames); take the last one as the latest shaped task. If several are in shaping, name the one chosen so the user can correct it.
    • Resolve the full path to the task file using the project task directory.
  2. Run the validator.
    python3 "${COORD_TOOLS:-$HOME/Projects/coord-wright}/bin/coord-review" <task-file-path>
  3. Interpret results.
    • Before reporting LGTM, read all four subtraction answers: do they address the actual deletion alternative, resulting orphans, budget, and retirement or justified growth? Confirm the third answer agrees with the frontmatter band. The validator checks structure, not whether prose is credible. Bare "none", repeated policy slogans, or unanswered questions need task-specific reasoning. A justified decision to retire nothing is valid; do not invent cleanup or relabel a task to clear the status advisory. Apply this semantic check even when coord-shape-review is skipped for simple work.
    • Exit 0: print LGTM — no shaping objections found for <task-id>.
    • Exit 1: present each numbered objection from stdout and ask:

      "I can fix these inline — want me to apply the fixes and re-run the check?" If yes, apply targeted fixes via coord update (never edit the task file directly), then re-run coord-review and confirm exit 0 before reporting done.

  4. xhigh task extra prompt.
    • If reasoning_effort: xhigh is on the task, also ask:

      "This task is xhigh — do you want to run a Codex loop pass before promoting?"

    • Relay the user's answer; do not run Codex automatically.
  5. Report.
    • State the final check result (LGTM or remaining objections).
    • If LGTM, offer: "Ready to promote — run coord promote <id> to move to pending."

Notes

  • Never mutate the task to bypass a check. Only fix real shaping gaps.
  • Do not run coord-review on tasks in pending, claude-working, or terminal statuses — those have already entered the work queue.
  • If the task file is not found (wrong id or task in a different project), report the error and ask the user to confirm the task id.

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.

Check your setup, free