skill-builder
暂无简介
- License 许可
- MIT
- 所属仓库
- zxpmail/ReqForge
- 最近更新
- 2026-10-07
- Stars
- ★ 18
- Forks
- 5
02 资源文件 8
- 参考文档commands/skill-builder.md472 B
- 参考文档references/anti-rationalization.md875 B
- 参考文档references/dimension-checklist.md1.5 KB
- 参考文档references/first-principles.md682 B
- 参考文档references/output-style.md929 B
- 参考文档references/workflow.md3.3 KB
- 参考文档skill.json933 B
- 参考文档templates/skill-template.md6.2 KB
03 SKILL.md 正文
<!-- forge: skill-builder v1.0 -->
name: skill-builder description: Used when the user wants to create a new Skill, or when an EVOLUTION.md proposal auto-generates a new Skill. Creates a structurally consistent new Skill following the framework's modular conventions. version: 1.0.0 updated: 2026-05-26 requires: []
<!-- begin: task --> [Task] Create a new Skill that conforms to framework conventions based on the user's described needs or an EVOLUTION.md fourth-layer proposal. Ensure the new Skill shares the same structure, unified style, and plug-and-play modularity as existing Skills.
<!-- end: task --> <!-- begin: not-for --> [Not For] - Modifying existing Skills -> edit the SKILL.md directly instead - Evolving rules from feedback patterns -> use /evolution-engine instead - Recording feedback about a Skill -> use /feedback-writer instead
<!-- end: not-for --> <!-- begin: dependency-check --> [Dependency Check] Required: None (this Skill does not depend on external files)
Optional:
- Related records in ../../feedback/ -> if from an EVOLUTION.md proposal, read the original feedback to understand the need's background
<!-- end: dependency-check -->
<!-- begin: first-principles -->
[First Principles]
→ references/first-principles.md
核心:Template First / Reference Existing / Minimum Necessary / Web-First。
<!-- end: first-principles -->
<!-- begin: output-style -->
[Output Style]
→ references/output-style.md
Tone: Architect explaining a blueprint — structured, precise, prescriptive.
<!-- end: output-style -->
<!-- begin: file-structure -->
[File Structure]
skill-builder/ ├── SKILL.md # Main Skill definition (this file) ├── templates/ │ └── skill-template.md # Skeleton template for new Skills └── references/ ├── first-principles.md ├── output-style.md ├── workflow.md ├── dimension-checklist.md └── anti-rationalization.md
<!-- end: file-structure --> <!-- begin: dimension-checklist --> [Dimension Checklist] See references/dimension-checklist.md for the full dimension checklist.
Must-have dimensions:
- **Decidable Triggers**: description specifies when to use AND when not
- **Workflow Executability**: each step specifies concrete action
- **Gotchas from Practice**: ≥3 specific failure points with guidance
- **Boundary Clarity**: [Not For] section with explicit exclusions
- **Dependency Check**: required deps have failure guidance
<!-- end: dimension-checklist -->
<!-- begin: anti-rationalization-checklist -->
[Anti-Rationalization Checklist]
→ references/anti-rationalization.md
遇 skipping template / skipping reference / TBD placeholders 时读取。
<!-- end: anti-rationalization-checklist -->
<!-- begin: gotchas -->
[Gotchas]
Skipping template: "I know the structure well enough" — read the template anyway. Every time you skip, you'll miss something: a section heading, a required field, or the consistent format.
Not cross-referencing existing Skills: Writing in a different style from the rest of the codebase. Always read 1-2 existing Skills before creating a new one. Consistency matters for maintainability.
Empty sections: "To be filled later" is technical debt. If a section isn't needed, don't include it. If it IS needed, fill it now. Empty sections in a Skill cause confusion when the Skill is invoked.
Missing Gotchas: You're building a Skill that WILL accumulate failure points. If you don't leave a [Gotchas] section, where will those lessons go? Nowhere — they'll be repeated.
Skipping Skill eval: Shipping without triggers.json / cases.json — you won't know if description misfires or outputs regress. Run pnpm skill-eval after every meaningful SKILL.md change.
<!-- end: gotchas --> <!-- begin: quality-rubric --> [Quality Rubric] 16-item, 32-point scoring system. Ship threshold: ≥ 24 with no critical item scoring 0.
| # | Dimension | Pts | Critical | Scoring |
|---|-----------|-----|----------|---------|
| 1 | Decidable triggers | 2 | YES | 2 = description specifies when to use AND when not to use; 1 = when-to-use only; 0 = vague ("helps with X") |
| 2 | Principle depth | 2 | no | 2 = each principle has concrete implication; 1 = principles exist but some are trivial; 0 = fewer than 3 principles |
| 3 | Gotchas from practice | 2 | no | 2 = ≥3 specific failure points with "what to do instead"; 1 = 1-2 gotchas; 0 = none |
| 4 | Workflow executability | 2 | YES | 2 = each step specifies concrete action; 1 = steps exist but some are vague; 0 = no workflow |
| 5 | Dependency completeness | 2 | no | 2 = required deps have failure guidance, optional have degraded mode; 1 = deps listed but no guidance; 0 = no dep check |
| 6 | Output Style defined | 2 | no | 2 = tone + principles + typical expressions; 1 = partial; 0 = absent |
| 7 | Domain dimensions | 2 | no | 2 = domain checklist with must-have/recommended/optional; 1 = checklist exists but flat; 0 = none |
| 8 | Anti-rationalization | 2 | no | 2 = ≥3 rationalizations enumerated with correct response; 1 = 1-2; 0 = none |
| 9 | No placeholders | 2 | YES | 2 = zero TBD/FIXME/template markers; 0 = any found |
| 10 | Cross-reference consistency | 2 | no | 2 = Workflow refs Strategy, Strategy refs Checklist; 1 = partial refs; 0 = sections are isolated |
| 11 | Boundary clarity | 2 | YES | 2 = [Not For] section with explicit exclusion conditions; 1 = boundaries mentioned in description; 0 = none |
| 12 | Template alignment | 2 | no | 2 = follows skill-template.md structure; 1 = mostly aligned; 0 = diverges significantly |
| 13 | File size discipline | 2 | no | 2 = ≤500 lines; 1 = 501-600 lines; 0 = >600 lines |
| 14 | Naming convention | 2 | no | 2 = kebab-case dir + [Section] format; 1 = one violation; 0 = multiple violations |
| 15 | Output artifacts listed | 2 | no | 2 = explicit artifact list; 1 = implicit; 0 = none |
| 16 | Initialization wired | 2 | no | 2 = points to first Workflow step; 1 = present but vague; 0 = absent |
**Scoring**: Run `pnpm validate-skill --score core/skills/<name>` to compute.
<!-- end: quality-rubric -->
<!-- begin: output-artifacts -->
[Output Artifacts]
- skills/<skill-name>/SKILL.md — new Skill definition file (relative to framework root)
- skills/<skill-name>/templates/ — template directory for the new Skill (if any, relative to framework root)
- .forge/skills/<skill-name>/eval/ — Skill eval pack (triggers.json + cases.json); run pnpm skill-eval init <skill-name> then pnpm skill-eval <skill-name> (see skill-eval.md)
<!-- end: output-artifacts --> <!-- begin: creation-standards --> [Creation Standards] <!-- end: creation-standards --> <!-- begin: three-layer-modularity --> [Three-Layer Modularity] The framework's three-layer architecture — each layer is independent and decoupled:
**Layer 1: Atomic Capabilities (Sections)**
Each Skill consists of multiple independent Sections, each being an atomic capability module:
- [Dimension Checklist] — defines "what to check / collect"
- [Strategy] — defines "how to do it"
- Workflow — defines "in what order"
- [Dependency Check] — defines "what prerequisites are needed"
These are building blocks — the same patterns can be reused across different Skills.
Changing one Section does not affect other Sections.
**Layer 2: Skill (SKILL.md)**
A Skill = combination of multiple atomic capabilities, solving a complete domain problem.
Changing one Skill does not affect other Skills.
**Layer 3: Workflow (Main Control File)**
The main control file orchestrates the execution order and trigger conditions of multiple Skills.
Changing the workflow does not require changing Skill content.
<!-- end: three-layer-modularity -->
<!-- begin: section-categories -->
[Section Categories]
**Required** (all Skills have these):
- [Task] — one sentence describing what it does
- [Dependency Check] — check prerequisites on startup
- [First Principles] — 3-5 core principles
- [File Structure] — Skill directory structure
- [Initialization] — entry point
**Recommended** (most Skills have these):
- [Output Style] — tone + principles + typical expressions
- [DOMAIN Dimension/Checklist] — domain-specific inspection dimensions (name tailored to domain)
- [DOMAIN Strategy] — domain-specific methodology (name tailored to domain)
**On-Demand** (specific Skill types need these):
- [Information Sufficiency Check] — collection / analysis type Skills
- [Rollback Strategy] — release / deployment type Skills
- [Phase Completion Check] — development type Skills
- Multi-mode workflow — Skills with multiple execution modes
<!-- end: section-categories -->
<!-- begin: naming-conventions -->
[Naming Conventions]
- Skill name: kebab-case (e.g., skill-builder, dev-planner)
- Directory: skills/[skill-name]/ (relative to framework root, e.g., .claude/skills/, .cursor/rules/skills/, .opencode/skills/)
- Main file: SKILL.md
- Template files (if any): templates/ subdirectory
<!-- end: naming-conventions -->
<!-- begin: format-conventions -->
[Format Conventions]
- Section titles use [Title] format
- Content indented by 4 spaces
- Frontmatter only has name and description
- Written in Chinese
<!-- end: format-conventions -->
[Review Strategy] Reference [Quality Rubric] for scoring criteria and the Workflow section below for the creation sequence.
<!-- begin: workflow -->
[Workflow]
→ references/workflow.md
6 步流程:Requirements → Reference → Structure → Fill → Create → Register。
<!-- end: workflow -->
<!-- begin: initialization -->
[Initialization]
Execute [Step 1: Requirements Gathering]
<!-- end: initialization -->
04 同仓库更多 Skills
feedback-writer
Analyze and record user corrections, feedback, and behavior preferences as structured feedback files for the evolution engine.
evolution-engine
Scan accumulated feedback to identify patterns, generate evolution proposals for SKILL.md, rules, or new Skills.
db-migration-checker
暂无简介
request-dispatcher
暂无简介
reqforge-greenkeeper
暂无简介
release-builder
暂无简介