Token导航 LogoToken导航

Agent Skill · 研究检索

feature-spec-reviewing(功能规格审查)

feature-spec-reviewing 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

feature-spec-reviewing 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 71,596。

总安装 194周安装 8GitHub Stars 3下载 63核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:feature-spec-reviewing(功能规格审查)
来源仓库:https://github.com/meriley/claude-code-skills
仓库路径:skills/feature-spec-reviewing
安装命令:
npx skills add https://github.com/meriley/claude-code-skills --skill feature-spec-reviewing
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/meriley/claude-code-skills --skill feature-spec-reviewing

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

feature-spec-reviewing 系统化审查功能规格,确保其完整、可测试且适合实施。适用于 Codex、Claude、Cursor、Gemini CLI,适合在工程工作开始前检查规格完整性时使用。重点识别边缘情况和错误处理的遗漏,防止后续开发返工。

安装前需确认权限范围和维护状态,注意可能触发文件读写和文档分析操作。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/feature-spec-reviewing

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/meriley/claude-code-skills --skill feature-spe…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

feature-spec-reviewing 怎么安装?

在终端执行 npx skills add https://github.com/meriley/claude-code-skills --skill feature-spec-reviewin…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

feature-spec-reviewing 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

feature-spec-reviewing 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

feature-spec-reviewing 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

feature-spec-reviewing 用的人多吗?

公开数据是累计安装 194 次,近一周安装 8 次,下载 63 次,GitHub 3 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

feature-spec-reviewing 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
feature-spec-reviewing

功能规格审查 · 本页

1943需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Feature Spec Reviewing

Purpose

Systematically review feature specifications to ensure they are complete, testable, and ready for implementation. Catches gaps in edge cases and error handling before engineering work begins.

When NOT to Use This Skill

  • Creating feature specs (use feature-spec-writing instead)
  • Reviewing PRDs (use prd-reviewing instead)
  • Reviewing technical specs (use technical-spec-reviewing instead)
  • Code review (different domain)
  • No spec exists yet (nothing to review)

Review Workflow

Step 1: Document Analysis

Read the entire feature spec and assess structure:

Initial Assessment:
- [ ] Feature definition is clear one-sentence
- [ ] Primary user story present
- [ ] Acceptance criteria exist
- [ ] Error states documented
- [ ] Definition of Done checklist exists

Red Flags:

  • Feature definition is vague or multi-sentence
  • Multiple unrelated user stories bundled
  • Only happy path documented
  • No error handling defined

Step 2: Feature Definition Review

Evaluate the feature definition:

Feature Definition Checklist:
1. Is it a single sentence?
2. Does it follow [Verb] [what] for [whom] to [outcome]?
3. Is the scope clear and bounded?
4. Is it too broad (actually multiple features)?
5. Does it align with parent PRD (if applicable)?

Severity Guide:

IssueSeverity
No feature definitionBLOCKER
Multi-feature bundlingBLOCKER
Vague definitionCRITICAL
Misaligned with PRDMAJOR
Minor clarity issuesMINOR

Step 3: User Story Quality

Evaluate user story against standard format:

User Story Checklist:
- [ ] Follows "As a... I want... So that..." format
- [ ] User type is specific (not just "user")
- [ ] Goal is concrete and achievable
- [ ] Benefit explains real value
- [ ] Story is right-sized (not an epic)
- [ ] Variants cover key user paths (if applicable)

Common Issues:

IssueExampleSeverity
Missing format"Add quick view feature"CRITICAL
Generic user"As a user" instead of "As a shopper"MAJOR
No benefitMissing "so that" clauseMAJOR
Epic-sizedEntire workflow in one storyCRITICAL

Step 4: Acceptance Criteria Review

Evaluate acceptance criteria for completeness and testability:

Criteria Coverage Checklist:
- [ ] Happy path (primary success scenario)
- [ ] Input validation (valid and invalid)
- [ ] Edge cases (boundaries, empty, max)
- [ ] Error handling (network, server, timeout)
- [ ] Performance expectations
- [ ] Security requirements (if applicable)
- [ ] Accessibility (if UI feature)

Testability Test: For each criterion, ask: "Can QA write a test case from this?"

❌ Untestable: "Quick view should be fast"
✅ Testable: "Quick view modal appears within 300ms of hover"

❌ Untestable: "Handle errors gracefully"
✅ Testable: "On network failure, show 'Connection lost' and retry button"

❌ Untestable: "Work on mobile"
✅ Testable: "On screens < 768px, bottom sheet replaces modal"

Step 5: Edge Case Coverage

Verify edge cases are comprehensive:

Edge Case Categories:
□ Empty state (no data, first use)
□ Minimum values (0, 1, empty string)
□ Maximum values (limits, overflow)
□ Null/undefined handling
□ Concurrent access (race conditions)
□ Timeout scenarios
□ Partial failure (some items succeed, some fail)
□ Network conditions (slow, offline, reconnect)

Coverage Scoring:

CoverageRatingAction
7-8 categoriesStrongApprove
5-6 categoriesAdequateMinor improvements
3-4 categoriesWeakCritical - add cases
< 3 categoriesMissingBlocker - major gaps

Step 6: Error States Review

Evaluate error states matrix:

Error States Checklist:
- [ ] All API calls have failure handling
- [ ] User-facing messages are helpful
- [ ] System actions are appropriate
- [ ] Recovery paths are clear
- [ ] Logging/alerting defined

Error Categories to Verify:

CategoryExample Errors
InputInvalid format, missing required, too long
NetworkOffline, timeout, partial response
Server500, 503, rate limit
Business LogicConflict, not found, unauthorized
ExternalThird-party API down, invalid response

Step 7: Technical Feasibility

Assess implementation readiness:

Feasibility Assessment:
- [ ] Dependencies identified and available
- [ ] Non-functional requirements realistic
- [ ] Data requirements clear
- [ ] No obvious technical blockers
- [ ] Performance targets achievable

Step 8: Parent PRD Alignment

If feature has parent PRD:

Alignment Check:
- [ ] Feature implements specific PRD user story
- [ ] Acceptance criteria align with PRD scope
- [ ] Out-of-scope items match PRD exclusions
- [ ] Success metrics support PRD goals

Step 9: Generate Review Report

Compile findings using the template below.

Severity Levels

LevelDefinitionAction
BLOCKERCannot implementMust fix before approval
CRITICALWill cause major issuesShould fix before sprint
MAJORWill cause reworkShould fix soon
MINORNice to haveCan address during implementation

Severity Examples

BLOCKER:

  • No feature definition or user story
  • Multiple features bundled as one
  • No acceptance criteria
  • Missing critical error handling

CRITICAL:

  • Acceptance criteria not testable
  • Missing edge case categories
  • No performance requirements
  • Conflicting requirements

MAJOR:

  • Vague user story benefit
  • Incomplete error states matrix
  • Missing accessibility considerations
  • Technical dependencies unclear

MINOR:

  • Minor clarity improvements
  • Additional edge cases nice to have
  • Formatting inconsistencies
  • Could add more detail

Review Report Template

# Feature Spec Review: [Feature Name]

**Reviewer:** [Name]
**Review Date:** [Date]
**Spec Version:** [Version]
**Spec Author:** [Name]

---

## Summary

**Status:** [APPROVE / APPROVE WITH CHANGES / NEEDS REVISION / MAJOR REVISION]

**Implementation Readiness:** [Ready / Ready After Fixes / Not Ready]

**Assessment:**
[2-3 sentences on overall quality and readiness]

---

## Findings by Severity

### Blockers
- [ ] [Finding with location and recommendation]

### Critical
- [ ] [Finding with location and recommendation]

### Major
- [ ] [Finding with location and recommendation]

### Minor
- [ ] [Finding with location and recommendation]

---

## Section Ratings

| Section | Rating | Notes |
|---------|--------|-------|
| Feature Definition | [Strong/Adequate/Weak/Missing] | |
| User Story | [Strong/Adequate/Weak/Missing] | |
| Acceptance Criteria | [Strong/Adequate/Weak/Missing] | |
| Edge Cases | [Strong/Adequate/Weak/Missing] | |
| Error States | [Strong/Adequate/Weak/Missing] | |
| Technical Constraints | [Strong/Adequate/Weak/Missing] | |

---

## Edge Case Coverage

□ Empty state - [Covered/Missing]
□ Minimum values - [Covered/Missing]
□ Maximum values - [Covered/Missing]
□ Null handling - [Covered/Missing]
□ Concurrent access - [Covered/Missing]
□ Timeout scenarios - [Covered/Missing]
□ Partial failure - [Covered/Missing]
□ Network conditions - [Covered/Missing]

**Coverage Score:** [X/8]

---

## Testability Assessment

**Can QA write tests?** [Yes / Partially / No]

**Problematic Criteria:**
- [Criterion that's not testable and why]

---

## Strengths
- [What the spec does well]
- [What the spec does well]

## Risks
1. [Risk and suggested mitigation]
2. [Risk and suggested mitigation]

---

## Questions for Author
1. [Clarifying question]
2. [Clarifying question]

---

## Recommendation

[Detailed recommendation with specific next steps]

Examples

Example 1: Strong Spec (Minor Feedback)

Status: APPROVE WITH CHANGES

Assessment: Well-structured feature spec with clear definition and comprehensive acceptance criteria. Minor gaps in timeout handling edge cases.

Findings:

  • MINOR: AC for network timeout missing retry limit
  • MINOR: Consider adding concurrent edit scenario

Edge Case Coverage: 7/8 (Missing: Partial failure)


Example 2: Needs Work (Critical Gaps)

Status: NEEDS REVISION

Assessment: Good user story but acceptance criteria lack edge case coverage. Error states incomplete. Should add missing scenarios before sprint planning.

Findings:

  • CRITICAL: Only happy path acceptance criteria defined
  • CRITICAL: Error states matrix incomplete (missing network, timeout)
  • MAJOR: Performance requirements not specified
  • MAJOR: Mobile responsive behavior undefined

Edge Case Coverage: 2/8


Example 3: Major Problems

Status: MAJOR REVISION REQUIRED

Assessment: Feature definition describes multiple features bundled together. Recommend splitting into separate specs.

Findings:

  • BLOCKER: "User authentication" bundles login, signup, password reset, MFA
  • BLOCKER: No acceptance criteria for individual flows
  • CRITICAL: User story is epic-sized
  • MAJOR: No error handling defined

Recommendation: Split into 4 separate feature specs.

Integration with Other Skills

Reviews Output From:

  • feature-spec-writing - Validate completed feature specs

Leads To:

  • technical-spec-writing - After spec approved
  • sparc-planning - After spec approved

Resources

  • See CHECKLIST.md for comprehensive review checklist

Related Agent

For comprehensive specification guidance that coordinates this and other spec skills, use the specification-architect agent.

平台分布

Codex

36.89%
按下载量换算23

Claude

31.92%
按下载量换算20

Cursor

16.68%
按下载量换算11

Gemini CLI

9.75%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills