Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/ewebdzine/canonify/doctornpx skills add ewebdzine/canonify --skill doctorgit clone --depth 1 https://github.com/ewebdzine/canonifyWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ewebdzine/canonify/doctor)<a href="https://agentmods.dev/skills/ewebdzine/canonify/doctor"><img src="https://agentmods.dev/badge/skills/ewebdzine/canonify/doctor.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00117 | $0.01870 |
| Opus 5 | $0.00059 | $0.00935 |
| Sonnet 5 | $0.00023 | $0.00374 |
| Haiku 4.5 | $0.00012 | $0.00187 |
Grade A, and why
doctor scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Doctor - keep the canon honest
The health gate. It runs two checks over every canon - plus an optional third check on whether the
installed Canonify plugin itself is behind its GitHub source - and reports a worklist; it does not
auto-fix. Run it as a full-repo sweep (manually, or on a schedule set by /canonify:kickoff); the
incremental version of the staleness check also runs inside /canonify:commit on the diff.
A canon is one canonical-pattern doc;
CANONIFY.mdindexes them. This gate checks that the canons still point at real code (reference integrity) and still match it (staleness).
How to run
-
Load the manifest. Read
CANONIFY.mdand enumerate every canon it lists, plus each categorydocs/<area>/INDEX.md. -
Check 1 - reference integrity (structural). For each canon:
- Every
file:linepath it cites still exists in the repo (flag broken paths). - It is registered: a one-line block in
CANONIFY.mdAND a row in its categoryINDEX.md. - No orphans (a canon file with no manifest entry) and no dangling entries (a manifest line pointing at a canon file that no longer exists).
- Every
-
Check 2 - staleness (temporal). Git holds the dates - do not hand-log them. For each canon:
baseline= the canon'sverified: <git-sha>frontmatter marker (the commit at which it was last confirmed against the code). If absent, fall back to the canon's own last-commit SHA and flag it as unmarked.refs= the source files the canon cites (itsfile:linereferences + its References section).- For each
ref: rungit log <baseline>..HEAD -- <ref>. If non-empty, flag - the canon references code that moved since it was verified. - Use git, never file mtime (checkout resets mtime). Use the platform git the repo requires
(read-only
git log); on a Windows working tree, the Windows git, not a WSL git.
-
Check 3 - plugin freshness (optional, best-effort - about the tooling, not the canons). Also surface when the installed Canonify plugin is behind its GitHub source, so a scheduled sweep catches an out-of-date toolchain too. Skip cleanly when it does not apply:
- Installed version: read Claude Code's plugin state (
~/.claude/plugins/installed_plugins.json) for this plugin's entry (canonify@<marketplace>) and note its recordedversion(andgitCommitSha). If there is no install record - the plugin is loaded locally for development (e.g. you are running inside its own source repo) - SKIP; a local checkout is its own latest. - Source version (resolve it the way Claude Code does): a plugin's effective version is the
versionin its.claude-plugin/plugin.json, else the marketplace entry'sversion, else the commit SHA. Resolve the marketplace's git URL (from~/.claude/plugins/known_marketplaces.jsonor the plugin'smarketplace.jsonsource - do not hardcode a repo), then read the source's current effective version: fetch.claude-plugin/plugin.jsonversionfrom the default branch (raw file fetch, orgit ls-remote <url> HEADfor the tip SHA when noversionis published). If offline or unresolvable, SKIP and say so. - Compare effective versions, not raw commits: flag only when the source's effective version
differs from the installed one -
Canonify plugin behind source: installed v<installed>, latest v<latest> - run /plugin update canonify@<marketplace>, then /reload-plugins. Comparing versions (not rawHEAD) means an unreleased commit - a doc tweak with no version bump - does NOT false-flag. - Caveat to state:
/plugin updateonly pulls when the publishedversionmoves (or, for a SHA-tracked plugin with noversionfield, on any new commit). If a maintainer ships changes without bumping the publishedversion, neither this check nor/plugin updatewill see them - that is maintainer discipline, not a consumer problem. Never run the update yourself - report only.
- Installed version: read Claude Code's plugin state (
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 110 lines · 117 tokens per session scan A 576b3bc42168
doctor is a skill published in the GitHub repository ewebdzine/canonify (23 stars, last pushed 2mo ago), licensed MIT. It adds 117 tokens to every session and 1,870 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…
chronicle
Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…
imagegen
Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…