agentleFS
kemalcalakkemalcalak/design-editor/packages/CLAUDE.md

design-editor / packages

CLAUDE.md16 starsChanged 26 days agoMITView on GitHub

The JS that remains after the TypeScript retirement: the Chrome extension and the build tooling around it. This directory is the Bun project root (packages/package.json); the repository root is a pure Cargo workspace with no package.json. The pen-* packages (pen-types, pen-core, pen-engine, pen-renderer, pen-figma, pen-mcp, pen-ai-skills, pen-sdk, pen-react, pen-acp) and pen-codegen were retired along with apps/*. Their functionality now lives in the Rust crates/ (see crates/CLAUDE.md). Nothing here depends on them. The agent-native Zig runtime was also removed — the…

The file

We split this CLAUDE.md into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run these from packages/:

    line 9

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

Packages

The JS that remains after the TypeScript retirement: the Chrome extension and the build tooling around it. This directory is the Bun project root (packages/package.json); the repository root is a pure Cargo workspace with no package.json.

The pen-* packages (pen-types, pen-core, pen-engine, pen-renderer, pen-figma, pen-mcp, pen-ai-skills, pen-sdk, pen-react, pen-acp) and pen-codegen were retired along with apps/*. Their functionality now lives in the Rust crates/ (see crates/CLAUDE.md). Nothing here depends on them. The agent-native Zig runtime was also removed — the built-in agent runtime is now the Rust agent crate (vendor/agent, shared with Zode).

Tooling

Run these from packages/:

  • Lint / format: bun run lint (oxlint, plus the extension's extractor drift check) / bun run format (oxfmt).
  • Iconify catalog (Rust assets): bun run generate-iconify-catalog — scripts/generate-iconify-catalog.mjs reads @iconify-json/* and writes crates/de-editor-ui/assets/iconify-catalog-{core,brands}.json (the icon catalog embedded in / served by the Rust web target) plus the gzipped extended families iconify-catalog-{material-symbols,phosphor}.json.gz. Regenerating also rewrites core/brands from whatever @iconify-json versions are installed; commit only the files you meant to change.
  • Versions: no package.json carries a version. The Chrome manifest keeps a 0.0.0 placeholder that package-extension.sh stamps. Bump with scripts/bump-version.sh at the repo root (its header lists every artifact that reads the version).

de-chrome-extension (de-chrome-extension/)

Manifest V3 Chrome extension that captures the rendered active tab and imports it into a running Design Editor. No dependencies — load it unpacked from chrome://extensions.

  • The logic is Rust. crates/de-chrome-extension-core (endpoint rules, chunked-transfer integrity, /mcp envelope + reply classification, download-name sanitisation) compiles to wasm; packages/de-chrome-extension/scripts/build-wasm.sh runs cargo build --target wasm32-unknown-unknown + wasm-bindgen --target web and installs the shim + module into de-chrome-extension/wasm/, which is gitignored like every other wasm-bindgen output in the repo. Build it before "Load unpacked" — the popup reports an actionable error otherwise. Unit-test the logic natively with cargo test -p de-chrome-extension-core.
  • The JS that remains is glue: chrome.* calls, fetch, popup DOM, and the functions injected into the captured tab (which must be JS — they run in the page's process). The manifest declares 'wasm-unsafe-eval' in content_security_policy.extension_pages, which is what MV3 requires to instantiate a bundled .wasm; no code is fetched from outside the package.
  • The capture engine is vendor/snapshot-extractor.js, a byte-identical copy of crates/de-html/assets/snapshot-extractor.js (the contract de_html::import_snapshot parses). bun run lint runs de-chrome-extension/scripts/check-extractor-sync.sh, which fails on drift; --fix re-copies the canonical asset. Never edit the copy.
  • The live MCP endpoint exposes two extension-scoped REST capabilities. Snapshot import uses POST /api/import/web-snapshot (insert-only and open to a well-formed chrome-extension:// origin unless explicitly pinned), with a plain /mcp tools/call import_web_snapshot fallback for unmanaged --serve-web. Intelligent design.md extraction uses a stricter asynchronous POST /api/generate/design-md + GET/DELETE job flow: it accepts only bounded style evidence and queues model work only for an extension id explicitly paired through DESIGN_EDITOR_EXTENSION_ALLOWED_IDS and a cancellable, evidence-only built-in API provider; tool-capable CLI/ACP adapters are not eligible. The extension generates a deterministic local guide when pairing or compatible model generation is unavailable. See snapshot_ingest.rs and design_md_route.rs under crates/de-host-services/src/mcp_live/. The extension's offline path is Download .de, which converts the capture locally into a ready-to-open document.
  • Store packaging. bun run package-extension (or de-chrome-extension/scripts/package-extension.sh) builds the wasm, runs the tests and all four guards, stages only the runtime files, copies pt to pt_BR/pt_PT for Chrome's manifest-locale lookup, and writes de-chrome-extension/dist/de-chrome-extension-<version>.zip (gitignored). The listing's privacy policy is de-chrome-extension/docs/privacy-policy.md.

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.