원클릭으로 Manus에서 모든 스킬 실행

spec-write-skill

스타60

포크8

업데이트2026년 7월 7일 08:11

公开 workflow：编写、改写、迁移或按 audit findings 修复 spec-first source skill（skills/<name>/）时使用。不要用于一次性回答、解释/总结/翻译、只审计、文档导出、第三方安装、普通 spec-* workflow 执行，或手改 generated runtime mirrors。

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

sunrain520

sunrain520/spec-first

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

Spec Write Skill

spec-write-skill 接收明确、值得复用的 skill authoring 目标，并把它落到 skills/<name>/ source patch。普通实现、调试、评审或文档导出走对应 spec-* workflow 或直接回答。

Purpose

把值得复用的 skill authoring 请求转成 source-first patch：触发/边界清楚，资源按需渐进披露，验证与 generated runtime mirror 状态可交接。

Contract Summary

When To Use

新建/改写/迁移/修复 skills/<skill-name>/ source skill 的触发、边界、I/O、资源、治理或验证。

When Not To Use

一次性回答、解释/总结/翻译、只审计、文档导出、第三方安装、普通 review、普通实现/调试/评审 workflow 执行、generated mirror 修补。

Inputs

用户目标、目标 skill、相邻 skill、repo 契约、skills-governance.json、audit findings、必要外部 skill 文本。

Outputs

do-not-create-skill、near-neighbor route、source patch、治理/测试/docs 更新、验证状态和 residual risks。

Artifacts

skills/<skill-name>/SKILL.md、必要 resources/evals、治理 JSON、runtime catalog、tests、CHANGELOG；generated mirrors 只由 spec-first init 投影。

Failure Modes

复用价值/输出/近邻排除不清，entry surface 不清，source/runtime 混淆，audit finding 被自动改写。

Workflow

先资格判断和意图澄清，再定 mode/tier/entry surface，写 patch，更新治理/测试/docs，跑匹配风险的 gate。

Downstream Consumers

spec-work、spec-skill-audit、runtime catalog、skill 维护者和最终用户。

Scenario Capability

Follows docs/contracts/workflows/scenario-capability-matrix.md (default). Overrides: none

Hard Boundaries

Source of truth 是 skills/、src/cli/contracts/dual-host-governance/skills-governance.json、相关 docs/tests；不要手改 .claude/、.codex/ 或 .agents/skills/。
Durable source mutation 需要明确授权或 active source-edit workflow；否则先输出 patch plan / preview。
脚本或测试只证明确定性事实；LLM 判断触发语义、边界和质量风险。
外部 skill 必须重写 entry surface、source/runtime 边界、验证和治理口径。
不照搬 yao-meta-skill 的完整 SkillOps 平台；只借鉴资格判断、意图澄清、分级 gate、资源边界和反模式 eval。

Quality Tiers

scaffold 只做 structure/source boundary；production 加 trigger/boundary eval；library 加 packaging/readiness evidence；governed 加 owner/review cadence/rollback boundary。细节见 Authoring Method 与 Delivery Gates。

Workflow

资格判断：读 Authoring Method，确认 repeated use、reusable output contract、near-neighbor exclusions 和 non-goals；一次性/解释/文档导出/未来 outline 输出 do-not-create-skill 或 near-neighbor route。
目标模糊时只问会改变 package 设计的 2-3 个问题；收敛 recurring job、real inputs、required outputs、exclusions、至少 1 个 should-trigger 示例、至少 1 个 near-neighbor 示例、quality tier 和 first eval target。
明确模式、质量层级、目标 repo、目标 skill 名称和 entry surface：workflow_command、standalone_skill 或 internal_only。
读取相邻 skill、治理记录和项目契约；新建/改写前读 Skill Quality Vocabulary。借鉴按 external benchmark -> user source -> local fit，只提炼 pattern，不复制 wording；必须先通过 local fit，不复制外部 invocation 假设。
先列真实 branch，再写触发描述：描述是 trigger contract，不是摘要；每个 branch 只保留一个触发，包含正向意图、负向边界和近邻；先测试 route，再扩展目录。
设计信息层级：先按 branch 分配资源，所有 branch 共用步骤和边界留在 SKILL.md，条件细节下沉 references/ 并写清 context pointer 的读取条件；逐字资源放置映射（哪类内容放 SKILL.md/references/scripts/assets/evals）以 Skill Quality Vocabulary 的 Information Hierarchy 为概念 SSOT。空目录、装饰性 reports、弱 pointer、未引用资源不进 baseline。
为写入、shell、runtime、delegate 或 handoff 步骤写可检查 completion criterion；同时检查 clarity(done/not done) 与 demand(需要多少 legwork)。读-only 轻 skill 可保持更轻。
对新增或改写 prose 做 sentence-level no-op pruning：逐句判断是否改变触发、读取、写入、判断、验证或 handoff；不改变行为的句子优先删除，不用润色保留。
更新 source-owned consumers：新增 skill 必须更新 skills-governance.json；用户可见或 catalog 变化要更新 runtime catalog、必要 docs/tests 和 CHANGELOG.md。
按 Delivery Gates 跑与 tier 匹配的最窄验证；目标 package 提供官方 quick_validate.py 时加跑。可分发或复杂 skill 尽量跑 package smoke、fresh-source eval 或 forward-testing，并记录未执行原因。
输出变更摘要、验证结果、generated runtime mirror 状态、residual risks 和必要下一步。

References

STOP Reference Trigger Map:

Before steps 1-4 or sentence-level pruning in step 8, read Authoring Method to decide qualification, intent dialogue, reference scan, authoring discipline, Evidence Matrix readiness, and anti-pattern coverage; if unread, do not create or rewrite source yet.
Before new/rewrite work, trigger description changes, information hierarchy design, or branch/resource placement in steps 4-8, read Skill Quality Vocabulary; it is the conceptual SSOT for resource placement, trigger/boundary wording, steering, and pruning. If unread, keep must-have rules in SKILL.md instead of hiding them in references.
Before selecting quality tier, adding resources/evals/scripts, running validation, or closing out in steps 6 and 10-11, read Delivery Gates; if unread, use scaffold-level checks only and record not_checked_with_reason.

이 저장소의 다른 Skills

같은 저장소

spec-mcp-setup

sunrain520/spec-first

Install, configure, verify, and refresh required harness runtime readiness facts for spec-first workflows on Claude Code, Codex, Kiro, Qoder, or Cursor.

2026-07-0760

spec-rule-miner

sunrain520/spec-first

Use this standalone skill when the user asks to mine a repo's existing coding conventions for future AI coding, generate or refresh project rules with AGENTS.md/CLAUDE.md pointers, create Cursor or Qoder rule files from actual code evidence, or make AI-generated code follow a specific project's habits. Do not use for team standards governance, normal code review/debug/refactor work, linter/formatter configuration, generic best practices, unsupported tool rule files such as .cursorrules or .kiro/steering rules, or generated runtime mirror edits.

2026-07-0660

agent-native-architecture

sunrain520/spec-first

Build applications where agents are first-class citizens. Use this skill when designing autonomous agents, creating MCP tools, implementing self-modifying systems, or building apps where features are outcomes achieved by agents operating in a loop.

2026-07-0660

feature-video

sunrain520/spec-first

Capture a visual demo reel (GIF, terminal recording, screenshots) for PR descriptions. Use when shipping UI changes, CLI features, or any work with observable behavior that benefits from visual proof. Also use when asked to add a demo, record a GIF, screenshot a feature, show what changed visually, create a demo reel, capture evidence, add proof to a PR, or create a before/after comparison.

2026-07-0660

proof

sunrain520/spec-first

Create, share, view, comment on, edit, and run human-in-the-loop review loops over markdown documents via Proof, the collaborative markdown editor at proofeditor.ai ("Proof editor"). Use when the user wants to render or view a local markdown file in Proof, share markdown to get a URL, iterate collaboratively on a Proof doc, comment on or suggest edits in Proof, HITL a spec/plan/draft for human review, sync a Proof doc back to local, or work from a proofeditor.ai URL. Trigger on phrases like "view this in proof", "share to proof", "iterate with proof", or "HITL this doc", and on spec-brainstorm / spec-ideate / spec-plan handoffs for human review. Also match clear requests for a rendered/shared markdown review surface even if the user does not name Proof. Do not trigger on "proof" meaning evidence, math/logic proof, burden of proof, proof-of-concept, or bare "proofread this" requests where inline text review is expected.

2026-07-0660

spec-code-review

sunrain520/spec-first

Review code diffs, PRs, or branch implementation changes using tiered persona reviewers, confidence-gated findings, and merge/dedup synthesis. Do not use for requirements/plan/task-pack document review, implementation execution, planning unresolved work, or commit/push/PR creation.

2026-07-0660

name	spec-write-skill
description	公开 workflow：编写、改写、迁移或按 audit findings 修复 spec-first source skill（skills/<name>/）时使用。不要用于一次性回答、解释/总结/翻译、只审计、文档导出、第三方安装、普通 spec-* workflow 执行，或手改 generated runtime mirrors。

Spec Write Skill

Purpose

把值得复用的 skill authoring 请求转成 source-first patch：触发/边界清楚，资源按需渐进披露，验证与 generated runtime mirror 状态可交接。

Contract Summary

When To Use

新建/改写/迁移/修复 skills/<skill-name>/ source skill 的触发、边界、I/O、资源、治理或验证。

When Not To Use

一次性回答、解释/总结/翻译、只审计、文档导出、第三方安装、普通 review、普通实现/调试/评审 workflow 执行、generated mirror 修补。

Inputs

用户目标、目标 skill、相邻 skill、repo 契约、skills-governance.json、audit findings、必要外部 skill 文本。

Outputs

do-not-create-skill、near-neighbor route、source patch、治理/测试/docs 更新、验证状态和 residual risks。

Artifacts

skills/<skill-name>/SKILL.md、必要 resources/evals、治理 JSON、runtime catalog、tests、CHANGELOG；generated mirrors 只由 spec-first init 投影。

Failure Modes

复用价值/输出/近邻排除不清，entry surface 不清，source/runtime 混淆，audit finding 被自动改写。

Workflow

先资格判断和意图澄清，再定 mode/tier/entry surface，写 patch，更新治理/测试/docs，跑匹配风险的 gate。

Downstream Consumers

spec-work、spec-skill-audit、runtime catalog、skill 维护者和最终用户。

Scenario Capability

Follows docs/contracts/workflows/scenario-capability-matrix.md (default). Overrides: none

Hard Boundaries

Source of truth 是 skills/、src/cli/contracts/dual-host-governance/skills-governance.json、相关 docs/tests；不要手改 .claude/、.codex/ 或 .agents/skills/。
Durable source mutation 需要明确授权或 active source-edit workflow；否则先输出 patch plan / preview。
脚本或测试只证明确定性事实；LLM 判断触发语义、边界和质量风险。
外部 skill 必须重写 entry surface、source/runtime 边界、验证和治理口径。
不照搬 yao-meta-skill 的完整 SkillOps 平台；只借鉴资格判断、意图澄清、分级 gate、资源边界和反模式 eval。

Quality Tiers

Workflow

资格判断：读 Authoring Method，确认 repeated use、reusable output contract、near-neighbor exclusions 和 non-goals；一次性/解释/文档导出/未来 outline 输出 do-not-create-skill 或 near-neighbor route。
目标模糊时只问会改变 package 设计的 2-3 个问题；收敛 recurring job、real inputs、required outputs、exclusions、至少 1 个 should-trigger 示例、至少 1 个 near-neighbor 示例、quality tier 和 first eval target。
明确模式、质量层级、目标 repo、目标 skill 名称和 entry surface：workflow_command、standalone_skill 或 internal_only。
读取相邻 skill、治理记录和项目契约；新建/改写前读 Skill Quality Vocabulary。借鉴按 external benchmark -> user source -> local fit，只提炼 pattern，不复制 wording；必须先通过 local fit，不复制外部 invocation 假设。
先列真实 branch，再写触发描述：描述是 trigger contract，不是摘要；每个 branch 只保留一个触发，包含正向意图、负向边界和近邻；先测试 route，再扩展目录。
设计信息层级：先按 branch 分配资源，所有 branch 共用步骤和边界留在 SKILL.md，条件细节下沉 references/ 并写清 context pointer 的读取条件；逐字资源放置映射（哪类内容放 SKILL.md/references/scripts/assets/evals）以 Skill Quality Vocabulary 的 Information Hierarchy 为概念 SSOT。空目录、装饰性 reports、弱 pointer、未引用资源不进 baseline。
为写入、shell、runtime、delegate 或 handoff 步骤写可检查 completion criterion；同时检查 clarity(done/not done) 与 demand(需要多少 legwork)。读-only 轻 skill 可保持更轻。
对新增或改写 prose 做 sentence-level no-op pruning：逐句判断是否改变触发、读取、写入、判断、验证或 handoff；不改变行为的句子优先删除，不用润色保留。
更新 source-owned consumers：新增 skill 必须更新 skills-governance.json；用户可见或 catalog 变化要更新 runtime catalog、必要 docs/tests 和 CHANGELOG.md。
按 Delivery Gates 跑与 tier 匹配的最窄验证；目标 package 提供官方 quick_validate.py 时加跑。可分发或复杂 skill 尽量跑 package smoke、fresh-source eval 或 forward-testing，并记录未执行原因。
输出变更摘要、验证结果、generated runtime mirror 状态、residual risks 和必要下一步。

References

STOP Reference Trigger Map:

Before steps 1-4 or sentence-level pruning in step 8, read Authoring Method to decide qualification, intent dialogue, reference scan, authoring discipline, Evidence Matrix readiness, and anti-pattern coverage; if unread, do not create or rewrite source yet.
Before new/rewrite work, trigger description changes, information hierarchy design, or branch/resource placement in steps 4-8, read Skill Quality Vocabulary; it is the conceptual SSOT for resource placement, trigger/boundary wording, steering, and pruning. If unread, keep must-have rules in SKILL.md instead of hiding them in references.
Before selecting quality tier, adding resources/evals/scripts, running validation, or closing out in steps 6 and 10-11, read Delivery Gates; if unread, use scaffold-level checks only and record not_checked_with_reason.