decklet
CLAUDE.md5 starsChanged 3 months agoMITView on GitHub
A slide engine for agents: the deck is a JSON model, the renderer is the editor, and create writes one self-contained deck.html a person can drag, retype, present and print. It ships as the @grunion/decklet package, a Claude Code plugin and an agent skill (SKILL.md).
- Installs packages
The file
We split this CLAUDE.md into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 28
Every behaviour change comes with a test.
- Distinctive: line 41
Verify the deploy: open both URLs and confirm the change is there.
≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md 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.
Files
decklet
A slide engine for agents: the deck is a JSON model, the renderer is the editor, and create writes one self-contained deck.html a person can drag, retype, present and print. It ships as the @grunion/decklet package, a Claude Code plugin and an agent skill (SKILL.md).
Build and run
- Node 22 or newer. Nothing to install for the engine or the CLI.
- Optional browser tooling for
verify,pdfandimport-html:npm i -D playwright && npx playwright install chromium - The authoring loop, from the repo root:
node bin/new.mjs --out model.json --slides 8 --density readingnode bin/finalize.mjs model.json --style style.json --out deck.htmlruns the three stages below with one style and writesdeck.finalize.jsonnode bin/validate.mjs model.json --style style.json --strictnode bin/create.mjs --model model.json --style style.json --out deck.html --format slidesnode bin/verify.mjs deck.html --strict
- Rebuild the explainer deck:
npm run build:deck - Rebuild the template pack:
npm run build:library - Host a deck so any browser can save edits:
node bin/serve.mjs deck.html - Regenerate the brand assets:
node brand/build-logos.mjs
Test
- Suite:
npm test(runsnode --test test/**/*.test.mjs; live browser tests skip when Playwright is absent) - Live proofs:
npm run test:live - Narrow the browser projects for a local run:
DECKLET_PROJECTS=chromium,firefox - Brand assets:
node --test brand/build-logos.test.mjs - CI gate:
.github/workflows/test.yml, jobgate, on every pull request and every push tomain. It runsnpm i -D playwright && npx playwright install --with-deps chromium webkit firefox, thennode --test test/**/*.test.mjs. - Rules the gate holds (CONTRIBUTING.md):
- Every behaviour change comes with a test.
- The eight text roles stay the only type scale.
- Motion never changes what print, the contact sheet, the PDF or
verifymeasure. - One file, zero network: no webfonts, CDNs or remote images.
template.htmlanddeck.htmlstay in sync.
Ship
- One change per pull request. The description says what changed, why, and how it was verified, with screenshots for anything visual (CONTRIBUTING.md).
- Open an issue before a large change. A runtime dependency, a build step, a network request or a second file next to
deck.htmlis declined on shape. deck.htmlandlibrary.htmlare generated under byte-identity gates (test/kpi.test.mjs,test/library.test.mjs). A change totemplate.html,bin/create.mjs,lib/,templates/orexamples/explainerrebuilds both withnpm run build:deckandnpm run build:libraryin the same PR, and again onmainafter the merge. Never resolve a conflict in either file by hand: take either side and rebuild.- A version bump moves
package.json,.claude-plugin/plugin.json,.claude-plugin/marketplace.jsonand aCHANGELOG.mdheading together (ROADMAP.md, "Release lockstep"). - Deploy: GitHub Pages serves the repo root from
main, so a merge publishes https://grunion-ai.github.io/decklet/deck.html and https://grunion-ai.github.io/decklet/library.html. - Verify the deploy: open both URLs and confirm the change is there.
- Vendored archify updates:
docs/archify.md. The weeklyarchifyworkflow pushesarchify/<tag>, and a person opens the PR by hand.
Pointers
- Authoring contract (model, style, thresholds, anti-patterns):
SKILL.md. Machine summary and file map:llms.txt. - Where each kind of change goes:
CONTRIBUTING.md, "Where things go". - The build loop and the second pass:
docs/building.md. - Editor, HUD, persistence, PDF:
docs/editor.md. - Charts, connectors, figures, logos, assets, verify, examples:
docs/. - Plan and sequencing:
ROADMAP.md. - Visual system:
DESIGN.md. - Purpose and positioning:
SOUL.md. - Security reports:
SECURITY.md.
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.

