init
Skill4 starsChanged 45 days agoMITView on GitHub
Scaffolds the Web Studio in a project (the studio's init, not Claude Code's): asks the conversation language and review mode, creates/updates CLAUDE.md sections, seeds .claude/docs, .claude/rules, docs/ and production/, merges settings; a re-run adds only what is missing and hands off to /update or /adopt. Run first in plugin mode (copy mode: to set the language), for 'set up the studio'.
137 other projects use a version of this.
Adopted by
Most recently taken up by
- naderelewa/Product-to-Prodtheir own version, first seen Nov 8, 2026
- ShahriarBijoy/show-me-howtheir own version, first seen Nov 8, 2026
- MMACABRE/SKILLStheir own version, first seen Nov 8, 2026
- BSSE23047/nauraiz-devtheir own version, first seen Nov 8, 2026
- edwinidrus/dragonhoardtheir own version, first seen Nov 8, 2026
- NilsHellwig/acl-paper-skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use daily-arxiv, exp-eval and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
LambenthanSkilldaily-arxiv
每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新
Also in 5 projects that use this
LambenthanSkillexp-eval
实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges
Also in 5 projects that use this
skyllwtSkillexp-pilot-eval
Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.
Also in 4 projects that use this
skyllwtSkillexp-pilot-run
Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.
Also in 4 projects that use this
gaebalaiSkillcodex-system
PROACTIVELY consult Codex CLI, your highly capable supporter with exceptional reasoning and task completion abilities. Codex is a trusted expert you should ALWAYS consult BEFORE making decisions on: design choices, implementation approaches, debugging strategies, refactoring plans, or any non-trivial problem. When uncertain, consult Codex. Don't hesitate - Codex provides better analysis. Explicit triggers: "think deeper", "analyze", "second opinion", "consult codex".
Also in 3 projects that use this
- +6 more from Lambenthan
- +1 more from skyllwt
Other versions137 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 137
- microsoft/vscode★ 195ksame name
- mcollina/skills★ 2ksame name
- skyllwt/OmegaWiki★ 1.7ksame name
- DAIR-AutoResearch/AutoSci★ 1.7ksame name
- PKU-DAIR/AutoSci★ 1.7ksame name
- skyllwt/AutoSci★ 1.7ksame name
- TokenRollAI/cc-plugin★ 483same name
- pipecat-ai/gradient-bang★ 420same name
- AlexPEClub/ai-coding-starter-kit★ 382same name
- dannysmith/tauri-template★ 312same name
127 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository52 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- testing/CLAUDE.mdCLAUDE.md
- a11y-auditSkill
- adoptSkill
- api-contractSkill
- architecture-decisionSkill
- architecture-reviewSkill
- backlogSkill
- brainstormSkill
- changelogSkill
- code-reviewSkill
- collaboration-protocolSkill
- create-storiesSkill
- data-modelSkill
- dependency-auditSkill
- deploySkill
- design-systemSkill
- dev-storySkill
- docsSkill
- feature-specSkill
- game-conceptSkill
- hardenSkill
- helpSkill
- hotfixSkill
- impactSkill
- incidentSkill
- migrateSkill
- pentestSkill
- perf-auditSkill
- product-specSkill
- qa-planSkill
- refactorSkill
- release-checklistSkill
- retrospectiveSkill
- security-auditSkill
- setup-stackSkill
- skill-improveSkill
- skill-testSkill
- sprint-planSkill
- sprint-statusSkill
- stack-updateSkill
- startSkill
- story-doneSkill
- team-featureSkill
- team-gameSkill
- team-releaseSkill
- team-securitySkill
- tech-debtSkill
- test-setupSkill
- threat-modelSkill
- updateSkill
- ux-specSkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 0 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 125 edited versions we read keep these.
0
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 37
After the "write" answer: touch .claude/.write-consent (§ Gates).
- Very specific: line 47
Ask, one AskUserQuestion: commit (Recommended) · leave uncommitted.
- Very specific: line 58
After INITIALISED: /adopt full (Recommended for brownfield: audits stack, artefacts and settings, fills technical-preferences.md from the …
- Very specific: line 59
After ALREADY INITIALISED: read references/rerun-handoff.md — its ladder (/update · /adopt full · /help) marks exactly one option …
- Very specific: line 60
After NOT INITIALISED: /init again (--plugin-root <path> when no root was found.
“Common” is measured: kept by at least 6 in 10 of the 125 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Init — scaffold the studio in this project
Language, <hooks>, gate mechanics (consent marker, Gate record) and the documents-lane commit gate: docs/coordination-rules.md § Skill conventions. Until CLAUDE.md exists, the language chosen in Phase 2 binds instead.
Writes files only after "May I write …?" → "yes".
Glossary: <root> — the seed root, plugin mode only (the install path found in Phase 1). Copy mode has no <root>: install.sh seeded docs, rules and hooks, and its stamp .claude/.web-studio-version is the only version source.
Phase 1: Locate the studio files
- Mode and seed root, in this order; the first that holds decides:
--plugin-root→ plugin mode,<root>is that path;.claude/.web-studio-versionpresent (install.sh's stamp) → copy mode, nothing to seed — before the "Plugin root:" line and the plugin list: a plugin installed at user scope prints that line into every project's session and never turns a copy-mode project into plugin mode;- the "Plugin root:" line printed by the session-start hook → plugin mode;
claude plugin list --json: theweb-studiorow whoseprojectPathis this project's root (git rev-parse --show-toplevel), else itsuser-scope row → its install path; the list covers every project on this machine, never take the first match (/updatePhase 1);.claude/docs/stack-reference/index.mdalready present — copy mode, nothing to seed. None found → ask the user for the path to the plugin/kit directory. Write nothing until a root is known; no path given →NOT INITIALISED (no plugin root)(Phase 6).
- Brownfield or empty — code exists on the evidence of
references/seed-manifest.md§ Stage (a manifest with sources, deploy files, a git history beyond a handful of commits); the result decides the stage (Phase 3) and the hand-off (Phase 6).
Phase 2: Language and review mode
--language / --review skip the matching question.
- Language, one
AskUserQuestion: "Which language should we use for conversation and documents?" Options: English plus the language of the user's own words when it differs (this session's messages; on a bare/initwith no prose, a non-English README/CLAUDE.md or the user's global~/.claude/CLAUDE.md— aReadoutside the project, allowed here). Always at least two named options — one option is a railroad, not a choice; nothing inferable → the two or three languages most plausible for this user's environment. - Switch language now. From the answer on (an option,
--language, or a language stated in the reply text) every question, the plan, the write summary and the hand-off are in that language — CLAUDE.md not existing yet is no excuse; the session that follows inherits the tone/initsets. - Review mode, one
AskUserQuestion:lean(Recommended for a single developer: lead plus security on sensitive work) ·full(every significant artefact reviewed) ·solo(reviews on request). It scopes reviews only: no mode skips/start//adopt, the specs or the catalog's required steps (review-workflow.md).
Phase 3: Plan
- Decide the stage (
production/stage.txt) from Phase 1 step 2:discoveryfor an empty project; brownfield →build(code, no release) oroperate(release files, compose.prod, a deploy skill in the repository —references/seed-manifest.md§ Stage; a README's live URL only prompts a question), proposed from the facts and confirmed in the "May I write?" question. Never writediscoveryover a running project. - Show the plan from
references/seed-manifest.md: every path it lists as created or merged (settings template per mode as it says), marking existing files that differ; an existingCLAUDE.mdgets the exact insertion shown. - Write gate: "May I write these files?" — one
AskUserQuestion: write (Recommended) · show the draft/diff first · not now. "Not now" → nothing is written,NOT INITIALISED (declined). - After the "write" answer:
touch .claude/.write-consent(§ Gates).
Phase 4: Write and verify
- Apply the plan; print the tree of created files.
- Smoke-check the statusline:
echo '{"cwd":"'"$PWD"'"}' | bash .claude/statusline.sh. - Version stamp — copy mode only. Keep install.sh's
.claude/.web-studio-version; missing → record the version of<kit>/.claude-plugin/plugin.jsononly when--kit <path>names the kit repository (the argument/updatetakes;--plugin-rootmeans plugin mode), else nameinstall.shas the fix. Plugin mode writes none:claude plugin list --jsonholds the version, and a stamp would claim files the project does not hold (/updatedeletes it).
Phase 5: Commit (documents lane)
Right after the write, one commit gate (§ Documents-lane commit gate; .claude/docs/git-workflow.md § Documents).
- Record it:
<hooks>session-state.sh set Gate "/init Phase 5: commit?". - Ask, one
AskUserQuestion: commit (Recommended) · leave uncommitted; on a story branch, name it and offer the § Documents-lane fork (switch to<default>Recommended — a pipeline-wide document · commit here · leave uncommitted). After the answer:<hooks>session-state.sh set Gate "—". - On "commit":
docs: initialise web studio, staging exactly the documents this run wrote —CLAUDE.md,.claude/docs/,.claude/rules/,docs/web-studio/README.md,production/stage.txt,production/review-mode.txt,.claude/.web-studio-version(copy mode). The seededdocs/*andproduction/*folders are not listed: git stages no empty folder. .claude/settings.json,.claude/statusline.shand the.gitignorelines are toolchain work, not documents: never in thedocs:commit; name them in the result and offer the chore lane (git-workflow.md § Chore / infra).- Nothing is committed without the answer. Nothing written (
ALREADY INITIALISEDwith 0 files differing,NOT INITIALISED) → no commit gate.
Phase 6: Verdict and hand-off
- Verdict, one line:
INITIALISED— neither the## Studio (Web Studio)block inCLAUDE.mdnor a seeded.claude/docs/existed before this run (Phase 1 items 1.3 and 1.5 did not apply).ALREADY INITIALISED (N files differ)— one of them did; N = seeded files whose content differs from the seed root (listed in Phase 3; recipe and ignore list inreferences/rerun-handoff.md); the run only added what was missing.NOT INITIALISED (declined)— "not now" at the write gate;NOT INITIALISED (no plugin root)— no root found, no path given. Nothing was written.
- Next step — one
AskUserQuestion, never a plain text line, every command namespaced (§ Skill conventions → Paths and names):- After
INITIALISED:/adopt full(Recommended for brownfield: audits stack, artefacts and settings, fillstechnical-preferences.mdfrom the facts) ·/start(Recommended for an empty project) ·/help· stop here. - After
ALREADY INITIALISED: readreferences/rerun-handoff.md— its ladder (/update·/adopt full·/help) marks exactly one option Recommended by the first fact that holds, offers the others, and never recommends "do nothing" on a project with code. - After
NOT INITIALISED:/initagain (--plugin-root <path>when no root was found; Recommended) ·/help· stop here.
- After
/initscaffolds the studio and stops there. It never writes product code (index.html, amain, a component), however trivial the goal: a greenfield project's first product artifact is/start's, whose stack and spec steps may pick the shortest path — never init inline. A hand-off answered with anything but a pipeline command is the user's call; init does not pre-empt it with code.
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.

