improve-repo-harness
Harness: MUST invoke only when a user explicitly asks to assess or improve agent guidance, repository navigation, validation discoverability, bootstrap documentation, or CONTEXT/ADR conventions. Do not invoke autonomously.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- improve-repo-harness
- description
- Harness: MUST invoke only when a user explicitly asks to assess or improve agent guidance, repository navigation, validation discoverability, bootstrap documentation, or CONTEXT/ADR conventions. Do not invoke autonomously.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档agents/openai.yaml167 B
- 参考文档DOMAIN-CONVENTIONS.md1.5 KB
- 参考文档HARNESS-RUBRIC.md2.0 KB
- 参考文档REPO-MAP-FORMAT.md1.4 KB
03 SKILL.md 正文
Improve Repo Harness
Harness durable repository guidance so an agent can orient itself, find governing documents, and discover supported validation and bootstrap surfaces. Make small, evidence-based improvements; the repository remains the source of truth for its own structure and conventions.
Scope
This skill improves durable, repository-local guidance. It does not author a domain model or architecture decisions. Route material domain-model work to domain-modeling.
Process
1. Explore the existing harness
Read the existing agent front door, repository README and documentation index,
nearby contribution guidance, build or validation entry points, and enough
source and test structure to understand current navigation. Inspect existing
GLOSSARY.md / GLOSSARY-MAP.md or CONTEXT.md / CONTEXT-MAP.md, ADR
locations, and requirements or design documents only as evidence of established conventions.
Record the paths and concise observations that support each possible recommendation. Consult HARNESS-RUBRIC.md for the evidence to collect for all five sections.
Completion criterion: You can cite repository-local evidence for every recommendation, and have identified any section with insufficient evidence.
2. Review sections one at a time
Present the following sections in order. For the current section, show the observed evidence, the smallest recommendation, and its benefit. Ask the user to adopt, skip, or revise that recommendation, then wait for the answer before presenting the next section.
Front door
Assess whether the existing agent guidance clearly points to durable rules, documentation, and safe validation entry points. Preserve its established format and make only the smallest navigation improvement.
Navigation and repository map
Assess whether a reader can find the relevant source, tests, documents, and entry points from existing guidance. When a concise map is selected, use the optional REPO-MAP-FORMAT.md reference rather than inventing a repository-wide inventory.
Validation and bootstrap
Assess whether existing, documented validation commands and prerequisites are discoverable. Recommend a documentation change only when concrete repository evidence supports it; keep commands and prerequisites owned by their existing documentation.
Domain conventions
Assess whether established glossary/map conventions (GLOSSARY.md /
GLOSSARY-MAP.md or CONTEXT.md / CONTEXT-MAP.md) and ADR conventions are
discoverable. Use DOMAIN-CONVENTIONS.md to inspect
and route those conventions. Offer discovery or routing only; route requests
to define terms or record decisions to domain-modeling.
Optional small improvements
Assess only small supporting changes justified by the evidence, such as a missing local cross-link or an unclear ownership pointer. Defer broad rewrites and proposals without a demonstrated navigation gap.
Completion criterion: The user has recorded an adopt, skip, or revise choice for each section; skipped sections have no proposed edit.
3. Confirm the selected edits
Combine only adopted or revised recommendations into a concise draft that names the affected files and exact guidance to add or change. Recheck every claim against the evidence, identify any assumption, and ask the user for explicit confirmation before writing.
Completion criterion: The user has explicitly confirmed the displayed draft, or the process ends with no changes when nothing is selected.
4. Write and validate the confirmed harness
Apply only the confirmed local documentation edits, preserving surrounding wording and the existing source of truth for each concept. Validate that new links resolve, referenced paths exist, documented commands were already verified in repository materials, and each claim remains supported by the exploration evidence.
Completion criterion: Every confirmed edit is present, locally valid, and traceable to the user's approved draft and repository evidence.
Completion
Report the selected sections, changed files, evidence used, validation results, and any deferred gaps. State when the repository has no justified harness change rather than filling it with generic guidance.
04 同仓库更多 Skills
ui-ux-pro-max
UI/UX design intelligence for web and mobile. Local database with 79 searchable styles (88 catalog entries), 192 color palettes, 74 font pairings, 192 product types, 119 UX guidelines, 105 curated icon entries, 17 GSAP motion presets, and 25 chart types across 22 stacks (React, Next.js, Vue, Nuxt, Svelte, Astro, SwiftUI, React Native, Flutter, Tailwind, shadcn/ui, Jetpack Compose, Angular, Laravel, JavaFX, WPF, WinUI, Avalonia, Uno Platform, UWP, Three.js, and HTML/CSS). Use when designing, building, or reviewing UI: pages, components, color schemes, typography, layout, accessibility, animation, or data visualization.
figma-mcp
Implement or compare UI against Figma designs using an available Figma MCP integration. Use for Figma-backed implementation, design-token extraction, or visual-parity gap analysis; Figma MCP access is optional and must already be configured by the user.
improve-codebase-architecture
Find evidence-based, active-change architecture improvements using shared design and domain-modeling guidance. Use when a user wants to reduce architectural friction, deepen a module, or make a current change safer, more testable, and easier to navigate.
domain-modeling
Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing GLOSSARY.md or CONTEXT.md, or recording or editing an ADR.
codebase-design
Provide the shared vocabulary and discipline for designing deep modules with clear interfaces and seams. Use when a user wants to design or improve a module, find deepening opportunities, choose a seam, or make code more testable and AI-navigable.
test-coverage-review
MUST invoke to review whether added or updated tests cover code changes in a diff. Checks new logic, branches, error paths, and regression risk, producing a change-to-test matrix. Use requirements-to-tests-traceability for acceptance-criteria mapping and orchestrated-review for bundled reviews.