duct / cli
AGENTS.md11 starsChanged 4 days agoMITView on GitHub
npx getduct check <url> is the deterministic half of the launch-check skill (skills/launch-check/): it runs every live check of references/checklist.md with its fixed thresholds and the skill's score arithmetic, and npx getduct card <result.json> draws the 1200x630 score card people post. The skill's ids, areas and weights are mirrored in src/catalog.js and the result shape is defined once in schema/launch-check.schema.json (src/schema.js interprets it, no second copy); test/catalog.test.js fails when either drifts from the skill, so a checklist change lands in…
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- app/AGENTS.mdAGENTS.md
- backend/AGENTS.mdAGENTS.md
- desktop/AGENTS.mdAGENTS.md
- site/AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/landing-pages.mdcCursor rule
- site/llms.txtllms.txt
- add-blog-postSkill
- add-changelog-entrySkill
- add-connectorSkill
- component-previewSkill
- featureSkill
- launch-readiness-auditSkill
- mosaic-panelSkill
- new-pageSkill
- prioritizeSkill
- security-auditSkill
- session-auditSkill
- working-backwardsSkill
- duct-sprint-refreshSkill
- launch-checkSkill
The file
Files
cli/ — the getduct npm package
npx getduct check <url> is the deterministic half of the launch-check skill (skills/launch-check/): it runs every live check of references/checklist.md with its fixed thresholds and the skill's score arithmetic, and npx getduct card <result.json> draws the 1200x630 score card people post. The skill's ids, areas and weights are mirrored in src/catalog.js and the result shape is defined once in schema/launch-check.schema.json (src/schema.js interprets it, no second copy); test/catalog.test.js fails when either drifts from the skill, so a checklist change lands in both places in the same PR. Zero runtime dependencies is a rule, not a preference: global fetch and node: built-ins only, Node 20+, and the card is drawn by our own rasterizer (src/card/raster.js) from glyph outlines generated by scripts/build-fonts.mjs out of DM Sans and Gelasio (OFL 1.1, licences beside the output in src/card/fonts/; re-run the script only when a face or the character set changes, and never hand-edit its output). Checks are pure functions of a fetched snapshot (src/check/checks.js), so every threshold is tested against fixture files through a fake fetch (test/helpers.js) and no test touches the network; a check the live site cannot settle returns UNVERIFIED with the evidence an agent needs, never FAIL. The fetcher stays polite: the getduct/<version> (+repo) User-Agent, two requests at a time, a timeout, capped bodies. make check-cli runs the tests (.github/workflows/cli.yml in CI). Releases are a cli-vX.Y.Z tag after bumping package.json and the getduct@X.Y.Z pins in skills/ (the skill and the plugin README pin an exact version because the Claude and OpenAI plugin directories refuse a launcher that can change after review; test/catalog.test.js fails until all three agree): .github/workflows/cli-release.yml publishes from CI through npm trusted publishing with provenance, so there is no npm token anywhere and none should be added.
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
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.

