agentleFS
tstraub89tstraub89/canon-ai/.claude/skills/canon-status/SKILL.md

canon-status

Skill3 starsChanged 9 days agoMITView on GitHub

Use when the human asks where canon tasks stand — "where are we", "pipeline status", "what's in flight", "what should I work on next", "what's blocked", or explicit `/canon-status` invocation. Also use at session start when several canon tasks may be in progress and a map of what needs attention vs. what's still running is needed. Read-only — surfaces task phase, blockers, and recommended next action across all in-progress tasks.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 3 are distinctive, and 4 are 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.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Not just "awaiting approval" — tell them how to approve and what happens next.

    line 50
  2. Very specific:

    If a task has been in_progress for the same phase across multiple /canon-status checks (infer from iteration counts and last-updated), note …

    line 67
  3. Very specific:

    If code_review.verdict = "changes_requested" and implement.status = "in_progress", note: "Implementer iterating (round N)."

    line 69
  4. Very specific:

    If any done task doesn't have a PR open yet, say so explicitly: "No PR yet — run canon run <id> --pr to push and open the draft."

    line 81
  5. Distinctive:

    If all tasks are clean, say so: "Everything's moving — no action needed."

    line 92
  6. Very specific:

    If there's nothing in flight and nothing needs attention: "Pipeline is idle.

    line 93
  7. Distinctive:

    Don't pad. If there's nothing to say, don't say it.

    line 95

≈ “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

Pipeline Status

echo "Branch:"
git branch --show-current 2>/dev/null
echo "Unpushed commits:"
git rev-list --count HEAD --not --remotes 2>/dev/null
echo ""
canon task list 2>/dev/null || echo "(no tasks)"

Instructions

Use Glob on tasks/*/status.json only to discover task IDs — skip tasks/_archive/. Then get each task's state with canon task status <id>, not by reading the REPO_ROOT file: for worktree tasks past plan, the REPO_ROOT copy is the stale pre-implement scaffold, and canon task status routes to the live worktree copy automatically.

Produce a status report with the sections below. Omit any section that has no entries. Don't show empty headers.

Be direct and opinionated. Don't just describe state — tell the operator what to do next and flag anything that looks wrong or worth watching.


🔴 Needs you

A task belongs here if any of these are true:

ConditionMeaning
spec.status = "done" AND spec_review.status = "pending" AND human_spec_gate = true (fast tier, or not yet run)Spec is ready for review
spec_review.status = "done" AND plan.status = "pending" (full tier halted at the spec gate; human_spec_gate reads false because the halt consumed it)Spec reviewed, waiting on you: read spec.md and spec-review.md including the Author Judgment Calls rows, then re-run canon run <id>
qa.status = "done" AND human_review.status = "pending"QA is done — ready to open a PR
escalations[] is non-emptyLoop cap hit or blocked — needs a call
Any phase has status = "blocked"Something is stuck

For each entry, be specific about what to do:

**task-id** — Title
→ [exact next action — e.g. "Read tasks/<id>/spec.md and say 'approved' to kick off the pipeline"]

Not just "awaiting approval" — tell them how to approve and what happens next.


🟡 In flight

Tasks where any phase has status = "in_progress" and no action is needed yet.

**task-id** — Title
→ Phase: [phase]  |  [iteration count or verdict if relevant]

Flag warning signs inline:

  • code_review.iterations_current_loop ≥ 2: "Review looping — N rounds this cycle. Default cap is 3 for XS/S/M, 5 for L/XL."
  • spec_review.iterations_current_loop ≥ 2: "Spec review looping — worth watching."
  • auto_block_count > 0: "Previously auto-blocked [N] times — escalation history in tasks/<id>/status.json."
  • If a task has been in_progress for the same phase across multiple /canon-status checks (infer from iteration counts and last-updated), note it: "Phase hasn't advanced — may be stalled."

If code_review.verdict = "changes_requested" and implement.status = "in_progress", note: "Implementer iterating (round N)."


🟢 Done this branch

Tasks where human_review.status = "done".

**task-id** — Title  [→ open PR if not yet pushed]

If any done task doesn't have a PR open yet, say so explicitly: "No PR yet — run canon run <id> --pr to push and open the draft."


Claude's take

End with 2–5 sentences that are actually useful — not a recap of what's above.

Things worth saying:

  • What's the single most important thing to do right now?
  • Is there anything that looks off (a task stuck in a phase, a review loop that's getting expensive, a worktree that might have diverged)?
  • If all tasks are clean, say so: "Everything's moving — no action needed."
  • If there's nothing in flight and nothing needs attention: "Pipeline is idle. Create a new task with canon task new <id> "Title"."

Don't pad. If there's nothing to say, don't say it.


Related

  • /canon-spec — author a new task when status shows the pipeline idle or "needs a next action."
  • /canon-pipeline — drive an in-flight task forward.
  • /canon-spec-review — pre-flight a spec before invoking the pipeline.
  • /canon-changelog — draft release notes when status shows shipped tasks.

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.