agentsclimarketplace
Sign in

Kickoff

Skill ewebdzine/canonify/skills/kickoff

One-time onboarding wizard that sets Canonify up in a fresh repo - a short survey (scan-and-seed? scheduled doctor?), then write CANONIFY.md + the docs/ skeleton and optionally bulk-seed canons by scanning the codebase. Run when the user types /canonify:kickoff, or says "set up Canonify", "kick off Canonify", "onboard this repo", "initialize the canon", "start canonify here".From its SKILL.md

Install
npx -y skills add ewebdzine/canonify --skill kickoff

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 23 stars23 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.2 KB, as published. Nobody here has run it

Kickoff - set Canonify up in a repo

The onboarding gate. Run it ONCE in a fresh repo to stand up Canonify: a routing manifest (CANONIFY.md), a docs/ corpus of canons, and - if you want - an initial set of canons drafted from the codebase. It is an interactive wizard, not a silent file-creator.

Vocabulary: a canon is one canonical-pattern doc; CANONIFY.md is the manifest the gates (/canonify:plan / :build / :commit) route through. This gate creates both. The templates ship with the plugin at ${CLAUDE_PLUGIN_ROOT}/templates/.

How to run

  1. Detect.

    • Confirm this is a git repo (Canonify's staleness checks rely on git).
    • Already set up? If CANONIFY.md exists at the repo root, do NOT re-run the survey - report what is there and offer to reconfigure (add categories, seed more canons) instead.
    • Detect the stack to pre-fill sensible category names: *.csproj/*.sln -> .NET; pyproject.toml/ requirements.txt -> Python; package.json -> Node; go.mod -> Go; and so on. Categories are the docs/<area>/ folders (e.g. integrations, services, ui) - the project's own, not a fixed set.

  2. Survey (ask a few; auto-detect the rest - keep it short):

    • Auto-seed by scanning?
      • Scan and propose (recommended): discover the surface (third-party integrations, internal services, key subsystems, UI patterns) and propose a list of canons to confirm before writing any.
      • Start empty: just the skeleton; author canons later with /canonify:create-canon.
    • Scheduled doctor backstop? Yes (weekly) / no, run it manually / later. A periodic /canonify:doctor sweep catches drift that accumulates between commits.
    • Confirm the stack + category folders you detected.

  3. Act.

    • Write CANONIFY.md at the repo root from ${CLAUDE_PLUGIN_ROOT}/templates/CANONIFY.md.template, filling in the detected categories.
    • Create the docs/<category>/ skeleton with a short INDEX.md per category.
    • Optionally copy ${CLAUDE_PLUGIN_ROOT}/templates/doc-style.md into the repo (e.g. docs/doc-style.md) so authors have the house-style locally.
    • If scanning: discover -> propose the canon list -> on confirm, run /canonify:create-canon in bulk over each target (index the code, draft in the house-style, register in CANONIFY.md + the category INDEX, stamp verified). Show progress; never silently generate a large batch.
    • If scheduled doctor: set up a periodic /canonify:doctor run via the host's scheduling (a scheduled agent, or a repo cron) at the chosen cadence.

  4. Report. What was created (CANONIFY.md, the category skeleton, any seeded canons + their verified markers), and the next steps: author more with /canonify:create-canon, orient before planning with /canonify:plan.

Don't do

  • Don't re-run on an already-set-up repo. Detect CANONIFY.md and offer reconfigure instead.
  • Don't silently mass-generate canons. Propose the list, get confirmation, then write + register.
  • Don't impose a fixed category set. Use the categories that fit THIS repo's stack and domain.
  • Don't invent canon content during the scan. Each seeded canon goes through /canonify:create-canon (evidence-based, every claim file:line-cited), not a guess.

Output template

## /canonify:kickoff - <repo>

Detected: <stack>, git repo. Categories: <area-a>, <area-b>, ...

Survey:
- Auto-seed: <scan-and-propose | empty>
- Scheduled doctor: <weekly | manual | later>

Created:
- CANONIFY.md (manifest, <N> category blocks)
- docs/<area>/INDEX.md  (x N)
[- seeded canons: <doc-a>.md, <doc-b>.md, ... (each registered + verified)]

Next: /canonify:create-canon to add canons; /canonify:plan to orient before planning.

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.