setup-kit
/setup-kit
Set up or update AI-agent files (optionally with a minimal, team or strict profile), agent guards, lint/CI checks and repository settings in the current project, in any language. Use when the user runs /setup-kit, asks to "set up the kit", "add AGENTS.md and checks", "configure agent hooks / CI / Sonar / Dependabot for this repo", or wants to update or remove an item the kit installed earlier. Supports `/setup-kit`, `/setup-kit update`, `/setup-kit remove <item>`, `/setup-kit remove --all`, `/setup-kit status` and `/setup-kit doctor`.
- Author
- mihaitaranu88 ↗
- Repository
- project-setup-kit ↗
- Licence
- MIT
- Last repository change
- 12.11.2026
Assessment and security
- Files in this package
- 121
- Scripts
- 28
- HTML comments
- 2
- URLs in SKILL.md
- 0
No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.
MIT, open: use and modify while retaining author attribution.
- 28 scripts in this package. Review them before installing, as Claude may execute them.
- SKILL.md zawiera komentarze HTML (2). Nie widać ich w podglądzie Markdown, a model je czyta.
Scripts: scripts/kit.mjs, scripts/outdated.mjs, templates/agent-env/agent-setup.sh, templates/checks/agent-docs-consistency.mjs, templates/checks/branch-name.mjs, templates/checks/commit-msg.mjs, templates/checks/guard-remote-db.mjs, templates/checks/lib.mjs, templates/checks/lint-ratchet.mjs, templates/checks/lockfile-check.mjs, templates/checks/migration-check.mjs, templates/checks/new-deps.mjs, template
What it does and when to use it
Set up or update AI-agent files (optionally with a minimal, team or strict profile), agent guards, lint/CI checks and repository settings in the current project, in any language. Use when the user runs /setup-kit, asks to "set up the kit", "add AGENTS.md and checks", "configure agent hooks / CI / Sonar / Dependabot for this repo", or wants to update or remove an item the kit installed earlier. Supports `/setup-kit`, `/setup-kit update`, `/setup-kit remove <item>`, `/setup-kit remove --all`, `/setup-kit status` and `/setup-kit doctor`.
Triggers from the description
Original author description
Set up or update AI-agent files (optionally with a minimal, team or strict profile), agent guards, lint/CI checks and repository settings in the current project, in any language. Use when the user runs /setup-kit, asks to "set up the kit", "add AGENTS.md and checks", "configure agent hooks / CI / Sonar / Dependabot for this repo", or wants to update or remove an item the kit installed earlier. Supports `/setup-kit`, `/setup-kit update`, `/setup-kit remove <item>`, `/setup-kit remove --all`, `/setup-kit status` and `/setup-kit doctor`.
How it works
- Files you work from
- Modes
- Non-negotiable rules
- Finishing
SKILL.md preview
--- name: setup-kit description: Set up or update AI-agent files (optionally with a minimal, team or strict profile), agent guards, lint/CI checks and repository settings in the current project, in any language. Use when the user runs /setup-kit, asks to "set up the kit", "add AGENTS.md and checks", "configure agent hooks / CI / Sonar / Dependabot for this repo", or wants to update or remove an item the kit installed earlier. Supports `/setup-kit`, `/setup-kit update`, `/setup-kit remove <item>`, `/setup-kit remove --all`, `/setup-kit status` and `/setup-kit doctor`. --- # Project setup kit You configure the current repository item by item from a catalog of 51 items. You never add anything without explicit consent, and you never ask about something the project already has. Speak to the user in their language. Keep each message short: one item, your recommendation, the reason, the question. ## Files you work from All paths are relative to this skill's folder. | File | Use it for | |---|---| | `references/flow.md` | The exact conversation flow and wording. Follow it step by step. | | `references/catalog.md` | The 51 items: id, group, when it applies, dependencies, detection signals, recommendation, files it writes. | | `references/detection.md` | How to detect language, stack, platform, AI tools, monorepo parts, and items that already exist. | | `references/languages.md` | Tool mapping per language (linters, type checkers, formatters, test runners, audit tools, dead-code tools). | | `references/questionnaire.md` | The rule question bank used by item 1 to fill `AGENTS.md`. | | `references/state-file.md` | Schema of `.ai-kit.json` and how update/remove work. | | `references/upgrades.md` | What changed in each kit version and what update mode offers for it. | | `references/platforms.md` | GitHub (full support) and other platforms (guide-only). | | `references/profiles.md` | Presets (minimal, team, strict) that answer most items in advance; the user still reviews the plan and every file. | | `references/troubleshooting.md` | What to do when a command, tool, hook, CI run or repository setting fails. Read it on the first error. | | `references/guides/*.md` | Step-by-step guides shown when the user says yes to an item that needs external setup. | | `templates/**` | Files you copy into the project and adapt. Never copy blindly: adapt to the detected language, paths and answers. | | `scripts/kit.mjs` | The only way to change kit state: `install`/`record` files (hashes), `state`/`context`/`setting` in `.ai-kit.json`, `config add/set` sections of `.ai-kit/config.json`, `pending list/apply/clean`, `verify`, `templates`, `guard-test`, `pin-actions`, `outdated`, `uninstall-plan`. It refuses tool-protected destinations. Run `node <skill-dir>/scripts/kit.mjs` without arguments for the full usage. | ## Modes - `/setup-kit` with no state file: full setup (flow.md, all phases). - `/setup-kit minimal|team|strict`: full setup with that profile preselected (`profiles.md`). - `/setup-kit` with an existing `.ai-kit.json`: same as `update`. - `/setup-kit update`: read `references/upgrades.md` and apply every section newer than `kitVersion` in `.ai-kit.json`, oldest first; compare the files in `.ai-kit.json` with the current templates; show what changed; never overwrite a file whose current hash differs from the recorded hash without showing the diff and asking. Then offer items that were never answered. - `/setup-kit remove <id>`: run `node <skill-dir>/scripts/kit.mjs uninstall-plan <id>`, show the plan, ask for confirmation, then remove the files, update `AGENTS.md` and `.ai-kit.json` (state-file.md, "remove"). - `/setup-kit remove --all`: the same for the whole kit with `uninstall-plan --all`. Order: hook wiring and git hooks first (so no hook points at a deleted script), then kit blocks in merged files (`AGENTS.md`, `.claude/settings.json`, `.mcp.json`, manifests), then the files in "Delete", then each file in "Ask first" one by one, then `.ai-kit/` and finally `.ai-kit.json`. List the repository settings to undo by hand; never change them yourself. - `/setup-kit status`: print the items with their state (installed, declined, not applicable, pending), the repository settings still pending, and the files waiting in `.ai-kit/pending/` (`kit.mjs pending list`). Change nothing. - `/setup-kit doctor`: check that every installed item still works and change nothing: Node version, `.ai-kit/config.json` parses, each recorded file exists and its hash matches, every guard answers a sample payload correctly, hook wiring points to existing scripts, git hooks are installed, the last CI runs of kit workflows passed (`gh run list`), and pinned versions are current (`node <skill-dir>/scripts/kit.mjs outdated`: actions, tool images such as gitleaks, actionlint, zizmor, trivy, hadolint, and MCP packages; for a MAJOR change read the release notes before proposing it). Report each problem with the fix from `troubleshooting.md` and ask before applying any. ## Non-negotiable rules 1. Analysis first, questions after. Read the repo before asking anything (flow.md, phase 1). 2. Skip what already exists. If an item is fully configured, do not ask about it; list it under "already configured". If it is partially configured, ask only about the missing part. 3. Skip what does not apply. An item whose `applies-when` condition is false (no DB, no FE, no API, single service) is not asked. 4. Respect dependencies. Ask items in the order given by `depends-on` in catalog.md. 5. One item at a time. Never batch several items into one question, except when the user explicitly accepts a whole group with its recommendations (flow.md, "group accept"). 6. Every question carries your recommendation and a one-line reason. 7. No change without consent. Write files only after an explicit yes for that item. 8. After every change: list each file created or modified, show the diff for modified files, ask the user to confirm they reviewed the files, then ask "Do you want changes, or shall we move to the next item?". Do not continue until they answer. 9. Complete, never replace. Existing configs (linters, CI, `AGENTS.md`, hooks) are extended in place. Show the diff first. In `AGENTS.md` the kit writes only inside `<!-- ai-kit:begin … -->` / `<!-- ai-kit:end … -->` blocks. 10. Every local check gets a CI counterpart. Hooks can be skipped with `--no-verify`; CI runs the same command from the same config. 11. Hooks stay fast. Pre-commit work must finish in about 10 seconds. Full type-check and full test runs go to CI. 12. External actions need their own consent. Repository settings, branch protection, secrets and anything sent to GitHub or another service are collected in the final "repository settings" step (item 12/21/22/23 settings) and executed only after the user approves that exact list. 13. Never commit, push or open PRs yourself unless the user explicitly asks for it in this session. 14. Never read real `.env` files. Use `.env.example` or ask. 15. Record everything in `.ai-kit.json` as you go, so an interrupted run can resume. 16. Keep `AGENTS.md` current. Each item that adds a rule appends it to `AGENTS.md` right away; a final pass at the end tidies it and checks the size budget (item 3). 17. Never invent facts. Commands, paths and conventions written into project files must exist in the repo; mark anything unconfirmed as `TODO:` for the user. 18. Once the protect guard (item 9) is installed, edits to the kit's own files trigger an approval prompt. That is expected; never try to bypass or disable it. 19. On any error, follow `references/troubleshooting.md`: stop, show the error, explain the likely cause, offer options with a recommendation, and continue only after the user chooses. Never get past an error by weakening a check or disabling a guard unless the user picks that option. ## Finishing At the end print: items installed, items declined, items not applicable, the repository settings still to do manually (if any), and how to re-run (`/setup-kit update`). Then stop.
Files in this package (121)
- SKILL.md 8 KB
- references/catalog.md 44 KB
- references/detection.md 7 KB
- references/flow.md 11 KB
- references/languages.md 6 KB
- references/platforms.md 1 KB
- references/profiles.md 4 KB
- references/questionnaire.md 32 KB
- references/state-file.md 4 KB
- references/troubleshooting.md 10 KB
- references/upgrades.md 8 KB
- references/guides/ai-review.md 2 KB
- references/guides/github-settings.md 2 KB
- references/guides/preview.md 2 KB
- references/guides/sonar.md 2 KB
- references/guides/task-apps.md 1 KB
- scripts/kit.mjs 26 KB
- scripts/outdated.mjs 7 KB
- templates/agent-env/README.md 2 KB
- templates/agent-env/agent-setup.sh 1 KB
- templates/agent-env/claude-web-session-start.json 1 KB
- templates/agent-env/copilot-setup-steps.yml 1 KB
- templates/agent-env/cursor-environment.json 1 KB
- templates/agent-env/devcontainer.json 1 KB
- templates/agents/AGENTS.md.tmpl 2 KB
- templates/agents/CLAUDE.md 1 KB
- templates/agents/copilot-instructions.md 1 KB
- templates/agents/cursor-agents.mdc 1 KB
- templates/checks/agent-docs-consistency.mjs 5 KB
- templates/checks/branch-name.mjs 1 KB
- templates/checks/commit-msg.mjs 2 KB
- templates/checks/config.json 6 KB
- templates/checks/guard-remote-db.mjs 4 KB
- templates/checks/lib.mjs 5 KB
- templates/checks/lint-ratchet.mjs 3 KB
- templates/checks/lockfile-check.mjs 2 KB
- templates/checks/migration-check.mjs 4 KB
- templates/checks/new-deps.mjs 10 KB
- templates/checks/precheck.mjs 8 KB
- templates/checks/prefix-budget.mjs 2 KB
- templates/checks/verify.mjs 3 KB
- templates/checks/warning-budget.mjs 3 KB
- templates/checks/worktree.mjs 3 KB
- templates/ci/.size-limit.json 1 KB
- templates/ci/knip.json 1 KB
- templates/ci/release-please-config.json 1 KB
- templates/ci/sonar-project.properties 1 KB
- templates/ci/github/README.md 8 KB
- templates/ci/github/agent-files.yml 1 KB
- templates/ci/github/ai-review.yml 3 KB
- templates/ci/github/bundle-size.yml 1 KB
- templates/ci/github/ci.dotnet.yml 1 KB
- templates/ci/github/ci.go.yml 1 KB
- templates/ci/github/ci.java.yml 1 KB
- templates/ci/github/ci.node.yml 1 KB
- templates/ci/github/ci.php.yml 1 KB
- templates/ci/github/ci.python.yml 1 KB
- templates/ci/github/ci.ruby.yml 1 KB
- templates/ci/github/ci.rust.yml 1 KB
- templates/ci/github/ci.yml 1 KB
- templates/ci/github/coverage.yml 1 KB
- templates/ci/github/duplication.yml 1 KB
- templates/ci/github/infra-scan.yml 1 KB
- templates/ci/github/lint-ratchet.yml 1 KB
- templates/ci/github/migrations.yml 1 KB
- templates/ci/github/pr-conventions.yml 3 KB
- templates/ci/github/pr-size.yml 2 KB
- templates/ci/github/preview-fe.yml 3 KB
- templates/ci/github/release.yml 1 KB
- templates/ci/github/security.yml 2 KB
- templates/ci/github/sonar.yml 1 KB
- templates/ci/github/workflow-security.yml 1 KB
- templates/commands/README.md 1 KB
- templates/commands/address-review-findings.md 1 KB
- templates/commands/bugfix.md 1 KB
- templates/commands/feature.md 1 KB
- templates/commands/ship.md 1 KB
- templates/env-validation/README.md 3 KB
- templates/git-hooks/.pre-commit-config.yaml 1 KB
- templates/git-hooks/README.md 3 KB
- templates/git-hooks/lefthook.yml 1 KB
- templates/github/CODEOWNERS 1 KB
- templates/github/dependabot.yml 1 KB
- templates/github/pull_request_template.md 1 KB
- templates/github/renovate.json 1 KB
- templates/guards/danger-guard.mjs 7 KB
- templates/guards/deps-guard.mjs 1 KB
- templates/guards/env-guard-allowlist.txt 1 KB
- templates/guards/env-guard.mjs 3 KB
- templates/guards/git-guard.mjs 2 KB
- templates/guards/hook-io.mjs 4 KB
- templates/guards/package-registry.mjs 19 KB
- templates/guards/post-edit-check.mjs 2 KB
- templates/guards/protect-guard.mjs 4 KB
- templates/guards/stop-check.mjs 2 KB
- templates/lint/README.md 3 KB
- templates/lint/eslint-ai-slop.config.mjs 2 KB
- templates/lint/golangci-ai-slop.yml 1 KB
- templates/lint/phpstan-ai-slop.neon 1 KB
- templates/lint/ruff-ai-slop.toml 1 KB
- templates/lint/eslint-rules/no-barrel-exports.mjs 1 KB
- templates/memory/README.md 1 KB
- templates/memory/_decision.md 1 KB
- templates/memory/_lesson.md 1 KB
- templates/repo/.editorconfig 1 KB
- templates/repo/.gitattributes 1 KB
- templates/repo/.mcp.json 1 KB
- templates/roles/architect.md 1 KB
- templates/roles/bug-fixer.md 1 KB
- templates/roles/feature-dev.md 1 KB
- templates/roles/qa-engineer.md 1 KB
- templates/roles/reviewer.md 1 KB
- templates/skills/SKILL.md.tmpl 1 KB
- templates/tool-config/README.md 2 KB
- templates/tool-config/claude/settings.json 2 KB
- templates/tool-config/copilot/danger-guard.json 1 KB
- templates/tool-config/copilot/deps-guard.json 1 KB
- templates/tool-config/copilot/env-guard.json 1 KB
- templates/tool-config/copilot/git-guard.json 1 KB
- templates/tool-config/copilot/protect-guard.json 1 KB
- templates/tool-config/cursor/hooks.json 1 KB
Versions
The database has no other skills named setup-kit.
Database snapshot: 13.11.2026 · package version 0ba5bc85066b (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100kit-setup
「Claude Code × 完全自動化キット」の初期設定をするスキル。道具の点検(OS・Python・ffmpeg・Pillow・Chrome/Edge)→足りないものを公式の入れ方で案内→作業フォルダ作成→キットのスキルをスキル置き場へコピー→各スキルの試し運転→「次にやること」の順で進める。 「キットの初期設定をして」「キットのセットアップをして」「キットを使えるようにして」「スキルを入れて」「環境チェックして」「動くか確かめて」と言われたときに使う。 Use when the user says "set up the kit",… - 96setup
First-run setup of the conductor kit in the current repository. Creates the .conductor/ state folder and its config, detects the deploy branch and test command, asks only for what cannot be detected, and checks the repository is ready for isolated worktrees. Use once per… - 100kit-doctor
Use when the user greets you or opens a session with no task ("hi", "hello", "hey", "yo", "what can you do", "where do I start"), asks what they need installed ("set up my pc", "what do I need", "check my setup", "is my toolchain ok"), or asks whether a game repo meets the bar… - 100update-kit
Update the kit's own files to the latest release without touching the creator's files. Use when the creator says "update the kit", "get the latest version", "is there a new version", "check for updates", "upgrade the kit" or "apply the update". - 100audit
Check an existing GitHub repo against the kit conventions (merge settings, protection of main, labels, labelled issues, kit files, actions pinned by SHA, ROADMAP in sync with issues, English only, CI and Dependabot) and report pass, warn and fail with fixes. Use when the user… - 100setup
In T3 Code only, install, configure, update or remove t3-orchestration for a project. Use when the user asks to set up orchestration on a project, add another project, update the plugin, or uninstall it.