一键在 Manus 中运行任何 Skill

setup-mrbaeksang-skills

星标0

分支0

更新时间2026年6月21日 01:37

Configure the mrbaeksang rails in a repo — record its deployment level and confirm the enforcing hooks are active. Run this once after installing the plugin. Use when the user runs /setup-mrbaeksang-skills or asks to set up / configure the rails.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

在 Manus 中运行

来源

Mrbaeksang

Mrbaeksang/skills

打开 GitHub 仓库查看创作者相关仓库

下载

在 Manus 中运行

SKILL.md

readonly

name	setup-mrbaeksang-skills
description	Configure the mrbaeksang rails in a repo — record its deployment level and confirm the enforcing hooks are active. Run this once after installing the plugin. Use when the user runs /setup-mrbaeksang-skills or asks to set up / configure the rails.
disable-model-invocation	true

Setup mrbaeksang skills

The rails calibrate everything off one number: the repo's deployment level. This skill records it and confirms the enforcing hooks are live. Run it once per repo.

Steps

Explain, then ask the deployment level. It's the master dial — the user may not know the term. Levels:
- L0 — local / prototype. Just making it work. Rails off; lightest.
- L1 — staging / internal. Real but limited users. Research full, smoke-gate auto, hooks warn.
- L2 — production. External/paying users, real responsibility. Full enforcement; hooks block.
Recommend L2 unless the user says it's a throwaway. Never assume — always ask.
Write docs/agents/deployment-level.md with two machine-readable markers, each on its own line:
```


```
current = what hooks enforce now (block/warn/off); target = what you design toward (seams). Then the human-readable dial table. The markers are the single source of truth hooks grep.
Seed .claude/usage-rules.tsv from the plugin's rules/starter-usage-rules.tsv (common deps' official-usage rules). The generic usage-guard hook reads this file; /onboard-dependency appends a line per new dependency. The hook code never changes — only this data grows.

Write the ## Engineering rails section into CLAUDE.md (or AGENTS.md if that's the repo's file) so a fresh agent reads the conventions before hooks ever fire:

## Engineering rails (mrbaeksang)
- Deployment level: docs/agents/deployment-level.md (current/target). Build for target, run at current.
- External dependency? Onboard first (/onboard-dependency): research official usage grounded in the
  installed source → vendor docs/official/<dep>/CONSIDERATIONS.md → append usage rules. Don't use a
  library you haven't onboarded.
- No dodge: never ship "overkill / blocked-on-credential / not sure" — check level / use a test
  double or local instance / research first.
- Mechanism lives in code: ERD = the migration, API = generated OpenAPI. Don't hand-author
  erd.md/api.md. Decisions → docs/adr/, terms → CONTEXT.md.
- Hooks (.claude, level-calibrated): onboard-dependency-guard, doc-discipline-guard, usage-guard.

Confirm rails are active. /reload-plugins should report the mrbaeksang-rails hooks. Smoke-test: add an un-onboarded dependency to a manifest (blocked at L2 / warned at L1), and ensure docs/official/INDEX.md exists for vendored sources.

Which skills now read this

/onboard-dependency — gates new dependencies; rigor + block/warn/off scales with the level
/smoke-gate — strictness of the "look at the real output" gate scales with the level
/due-diligence, /spike-first — mandatory at L2, optional at L0

Done

Tell the user the level is recorded, the rails are calibrated, and they can change it anytime by editing the mrbaeksang:level= marker — re-running this skill is only needed to reconfigure.

同仓库更多 Skills

同仓库

onboard-dependency

Mrbaeksang/skills

Onboard any new external dependency before using it — research the official docs, vendor them as the single source of truth, then generate a blocking guard. Use when adding or first using any library, framework, SDK, API, or CLI tool; when editing package.json / requirements.txt / pyproject.toml / go.mod / Cargo.toml; or when the user mentions adopting a new tool. This is a rail, not advice — the trigger is unconditional, never "when uncertain".

2026-06-210

no-dodge

Mrbaeksang/skills

The rigor gate that refuses LLM avoidance. Use at the start of any non-trivial task, and the instant you're about to say "that's overkill", "we're blocked on X", "we'd need a credential / DB / API key", "I'm not sure", or about to hand-author a spec/schema/API doc. Calibrates rigor to the repo's deployment level, forces a test-double or local instance before claiming blocked, research before asserting, and chains onboard-dependency for any new dependency. One gate; it chains the rest.

2026-06-210

name	setup-mrbaeksang-skills
description	Configure the mrbaeksang rails in a repo — record its deployment level and confirm the enforcing hooks are active. Run this once after installing the plugin. Use when the user runs /setup-mrbaeksang-skills or asks to set up / configure the rails.
disable-model-invocation	true

Setup mrbaeksang skills

The rails calibrate everything off one number: the repo's deployment level. This skill records it and confirms the enforcing hooks are live. Run it once per repo.

Steps

Explain, then ask the deployment level. It's the master dial — the user may not know the term. Levels:
- L0 — local / prototype. Just making it work. Rails off; lightest.
- L1 — staging / internal. Real but limited users. Research full, smoke-gate auto, hooks warn.
- L2 — production. External/paying users, real responsibility. Full enforcement; hooks block.
Recommend L2 unless the user says it's a throwaway. Never assume — always ask.
Write docs/agents/deployment-level.md with two machine-readable markers, each on its own line:
```


```
current = what hooks enforce now (block/warn/off); target = what you design toward (seams). Then the human-readable dial table. The markers are the single source of truth hooks grep.
Seed .claude/usage-rules.tsv from the plugin's rules/starter-usage-rules.tsv (common deps' official-usage rules). The generic usage-guard hook reads this file; /onboard-dependency appends a line per new dependency. The hook code never changes — only this data grows.

Write the ## Engineering rails section into CLAUDE.md (or AGENTS.md if that's the repo's file) so a fresh agent reads the conventions before hooks ever fire:

## Engineering rails (mrbaeksang)
- Deployment level: docs/agents/deployment-level.md (current/target). Build for target, run at current.
- External dependency? Onboard first (/onboard-dependency): research official usage grounded in the
  installed source → vendor docs/official/<dep>/CONSIDERATIONS.md → append usage rules. Don't use a
  library you haven't onboarded.
- No dodge: never ship "overkill / blocked-on-credential / not sure" — check level / use a test
  double or local instance / research first.
- Mechanism lives in code: ERD = the migration, API = generated OpenAPI. Don't hand-author
  erd.md/api.md. Decisions → docs/adr/, terms → CONTEXT.md.
- Hooks (.claude, level-calibrated): onboard-dependency-guard, doc-discipline-guard, usage-guard.

Confirm rails are active. /reload-plugins should report the mrbaeksang-rails hooks. Smoke-test: add an un-onboarded dependency to a manifest (blocked at L2 / warned at L1), and ensure docs/official/INDEX.md exists for vendored sources.

Which skills now read this

/onboard-dependency — gates new dependencies; rigor + block/warn/off scales with the level
/smoke-gate — strictness of the "look at the real output" gate scales with the level
/due-diligence, /spike-first — mandatory at L2, optional at L0

Done

Tell the user the level is recorded, the rails are calibrated, and they can change it anytime by editing the mrbaeksang:level= marker — re-running this skill is only needed to reconfigure.