litopencode
AGENTS.md2 starsChanged 45 days agoMITView on GitHub
Family-wide rules live in the parent ../AGENTS.md. Read that first if you started here; an agent launched inside this directory does not see it automatically. This file covers only what differs in this repository. LitOpenCode — the OpenCode port of the lit workflow family. npm package @litfamily/litopencode (executable/native ID litopencode). TypeScript ESM, tests via node --test. Git root is this directory, branch master. Do not pipe a suite through | tail — $? then reports tail's exit code, not the…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,843 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository45 other files it gives its agents.
- agent-rosterSkill
- autoconferenceSkill
- autoresearchSkill
- browser-driveSkill
- comment-checkerSkill
- debuggingSkill
- deep-interviewSkill
- doctor-installerSkill
- durable-litgoalSkill
- frontend-ui-uxSkill
- lit-burnoff-fileSkill
- lit-burnoffSkill
- lit-codeSkill
- lit-commitSkill
- lit-comprehendSkill
- lit-crucibleSkill
- lit-diagram-drawerSkill
- lit-docxSkill
- lit-fetchSkill
- lit-handoffSkill
- lit-humanizerSkill
- lit-initSkill
- lit-planSkill
- lit-pptxSkill
- lit-recapSkill
- litresearchSkill
- lit-scientific-visualizationSkill
- lit-typographic-motionSkill
- litworkSkill
- lsp-setupSkill
- lspSkill
- native-goal-verdictSkill
- readme-studioSkill
- refactorSkill
- reference-benchmark-claimsSkill
- release-guardrailsSkill
- review-workSkill
- rulesSkill
- search-workflow-ideasSkill
- start-workSkill
- structural-searchSkill
- tool-guardsSkill
- visual-qaSkill
- wikifySkill
- workflow-loopSkill
The file
We split this AGENTS.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
- Very specific: line 36
Do not pipe a suite through | tail — $?
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.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
AGENTS.md — LitOpenCode
Family-wide rules live in the parent ../AGENTS.md. Read that first if you
started here; an agent launched inside this directory does not see it automatically.
This file covers only what differs in this repository.
What this is
LitOpenCode — the OpenCode port of the lit workflow family. npm package @litfamily/litopencode (executable/native ID litopencode).
TypeScript ESM, tests via node --test. Git root is this directory, branch master.
Entry points
| Surface | Path |
| Activation and routing | src/activation.ts, src/activation-routing.ts, src/activation-prompt-utils.ts |
| Command and feature registries | src/commands.ts, src/features.ts, src/skills.ts |
| Installer asset handling | src/cli/managed-skill-assets.ts |
| Automatic handoff (opt-in) | src/auto-handoff.ts (settings, route, marker, digest), src/auto-handoff-hooks.ts (event and system-prompt hooks) |
| TUI mission panel | src/tui.ts (./tui export, slots and toasts), src/tui/ (pure model, layout, motion), src/cli/tui-config.ts (tui.json edits) |
| Skills corpus | skills/<id>/SKILL.md |
Verification
npm test # node --test; ~649 pass as of 2026-08-01
npm run typecheck
npm run build
npm run scan:legacy-tokens
npm run check:version
npm run check:pack-payload
npm run check:managed-skill-manifestDo not pipe a suite through | tail — $? then reports tail's exit code, not the suite's,
and the failure detail is truncated away. Redirect to a file instead. Running several repos'
suites concurrently has produced phantom failures here that do not reproduce serially.
Local conventions
- Relative imports carry explicit
.tsextensions (rewriteRelativeImportExtensions). Preserve that — dropping the extension breaks the build, not just style. - The
lit-planagent is planning-only by design: edit and bash are denied even in yolo mode. Never grant it write access to make a task easier. - Some test filenames intentionally keep an older prefix (e.g.
test/comprehend-command.test.mjsfor thelit-comprehendskill). A SKILL contract'sverification:list names that exact path, so the pair is internally consistent. Renaming means editing the contract; do not "fix" it casually.
Do not touch
HANDOFF.mdin this directory is deprecated. The authoritative handoff is../HANDOFF.mdat the family root.vendor/holds exact vendored sources; treat those corpus files as read-only authority rather than editing them to change behavior.
Packaging
Publication is governed by .npmignore (its presence disables the .gitignore fallback).
AGENTS.md is listed there, so this file is tracked in git but never published — a bare
npm pack --dry-run did include it before that entry was added.
The active Ignition vector cover (docs/assets/cover.svg) and docs/release-checklist.md stay in Git and are excluded from npm. The WebP motion cover and its still frame ship because the package README references them. The static robot cover (docs/assets/cover.webp) no longer appears in either README, but it still ships and the pack guard still requires it. The SVG uses explicit geometric paths and outlined glyphs with no embedded raster or external font dependency. Obsolete cover.png and generate_cover.py have been archived outside this product; keep their npm exclusions and pack-guard negative controls so those files cannot re-enter the package.
README.md and README-Ko-KR.md are the short GitHub landing pages and load assets by relative ./docs/... paths. What moved off them lives in the GitHub-only pages docs/guide.md, docs/skills.md, docs/screens.md and docs/brand.md (each with a -Ko-KR twin, images relative to docs/); those pages and docs/assets/social-preview.png are excluded from npm, check-pack-payload.mjs rejects them in a tarball and requires the shipped images they show. The npm package pages are README-npm.md and README-npm-Ko-KR.md (a hyphen, because npm always packs root README.* files); tools/readme-for-npm.mjs copies them over the GitHub pages in prepack and restores them in postpack, and a publish with --ignore-scripts must run its apply and restore by hand (see docs/release-checklist.md). Every file referenced by the npm pages must ship in the package and use a version-pinned https://cdn.jsdelivr.net/npm/@litfamily/litopencode@<version>/... URL. The exact README asset allowlist in .npmignore includes the outlined ASCII display, badges, wordmark and clay mark, licensed Lucide icons and license, font license, family illustration, and Ignition poster/MP4/GIF. The four Jev snapshot WebP files, the eight dark and light on-screen pictures (install, doctor, ignition toast, planner permissions), the three status card pictures, the six dark and light automatic handoff pictures and the promo film in English and Korean (animated preview, poster, MP4, a shared reduced-motion still, and the sources and Pretendard license under docs/assets/readme/promo-source/) are shown only by the GitHub pages, so they stay in Git and out of the package; the pack guard rejects them and does not require them. check-pack-payload.mjs requires the landing README targets and rejects unapproved presentation files. Keep the copyable ASCII in all four README pages and update tools/version-manifests.json when a pinned URL is added or removed. docs/release-checklist.md remains repository-only: the GitHub pages may link it, the npm pages must not. Runtime icons elsewhere remain supported. Keep README entry paths concise; detailed operational contracts live in docs/reference.md and docs/reference-Ko-KR.md and remain packaged. Preserve native runtime payloads and supported small icons. New documentation assets must resolve locally before committing.
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.

