Token导航 LogoToken导航

Agent Skill · 研究检索

complete-task(完成任务)

complete-task 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。

complete-task 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,权限为调用外部服务;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 67,439。

总安装 245周安装 10GitHub Stars 公开资料未说明下载 78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:complete-task(完成任务)
来源仓库:https://github.com/fancy-bread/sdlc-workflow-skills
仓库路径:skills/complete-task
安装命令:
npx skills add https://github.com/fancy-bread/sdlc-workflow-skills --skill complete-task
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/fancy-bread/sdlc-workflow-skills --skill complete-task

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。

complete-task 提交变更、推送远程仓库并创建 Pull Request。自动转换 issue 状态至 "Code Review",遵循 {type}/{TASK_KEY} 分支命名。包含 spec 摘要提取和蓝图解析,确保实现符合架构设计。

需确认当前分支为 feature/* 格式且 spec 文件位于 specs/ 目录。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/complete-task

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/fancy-bread/sdlc-workflow-skills --skill compl…;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

    它需要调用外部服务,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

complete-task 怎么安装?

在终端执行 npx skills add https://github.com/fancy-bread/sdlc-workflow-skills --skill complete-task。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

complete-task 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。

complete-task 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

complete-task 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。

complete-task 用的人多吗?

公开数据是累计安装 245 次,近一周安装 10 次,下载 78 次。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。

complete-task 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

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。

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。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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
complete-task

完成任务 · 本页

245—调用外部服务未明确未通过

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

Complete Task

Overview

Commit changes, push to remote, create pull request, and transition issue to "Code Review" status.

Definitions

  • {TASK_KEY}: Story/Issue ID from the issue tracker (e.g., FB-6, PROJ-123, KAN-42)
  • Branch Name Format: Use short format {type}/{TASK_KEY} (e.g., feat/FB-6, fix/PROJ-123)

- Short format is recommended: feat/FB-6 (not feat/FB-6-file-watching-workspace-commands) - Important: Be consistent within a project - use the same format for all branches

  • Spec Summary: Content from spec file located at specs/{FEATURE_DOMAIN}/spec.md

- Contains Blueprint (Context, Architecture, Anti-Patterns) and Contract (DoD, Guardrails, Scenarios) - Used in PR body to document feature contracts

  • Plan Summary: Content from plan file located at .plans/{TASK_KEY}-*.plan.md

- Plan File Selection: If multiple files match the pattern .plans/{TASK_KEY}-*.plan.md: - Use the most recently modified file (check file modification time) - If modification time cannot be determined, use the first file found alphabetically - Report which file was selected: "Using plan file: {filename}" - Contains implementation details, requirements, and acceptance criteria

  • Completed Checklist: Markdown checklist posted as a comment to the issue showing what work was completed
  • Current Branch: Should match the expected format {type}/{TASK_KEY} (e.g., feat/FB-6, fix/PROJ-123)

Prerequisites

Before proceeding, verify:

  1. MCP Status Validation: Perform MCP server status checks (see mcp-status.md for detailed steps)

- Test each configured MCP server connection (Atlassian, GitHub, etc.) - Verify all required integrations are authorized and operational - If any MCP server fails validation, STOP and report the failure. Do not proceed. - MCP Tool Usage Standards: MCP tool usage should follow best practices (check schema files, validate parameters, handle errors gracefully). These standards are documented in AGENTS.md §3 Operational Boundaries if AGENTS.md exists, but apply universally regardless.

  1. Branch Verification:

- Verify current branch matches expected format: {type}/{TASK_KEY} - Confirm branch exists on remote (if not, this will be created on push) - If branch doesn't match expected format, STOP and report the mismatch.

  1. Test Verification:

- Run all tests locally and verify they pass - If any tests fail, STOP and fix them before proceeding. Do not commit failing tests.

  1. Documentation Files:

- Check if spec exists at specs/{FEATURE_DOMAIN}/spec.md - Check if plan exists at .plans/{TASK_KEY}-*.plan.md - Plan File Selection: If multiple files match the pattern .plans/{TASK_KEY}-*.plan.md: - Use the most recently modified file (check file modification time) - If modification time cannot be determined, use the first file found alphabetically - Report which file was selected: "Using plan file: {filename}" - If neither spec nor plan exists, WARN but proceed: PR will not include detailed context

  1. Verify Spec Updated (if applicable):

- If code changes affected API contracts, data models, or quality targets: - Check if spec was updated in the staged changes - If spec exists but wasn't updated: WARN user about Same-Commit Rule violation - Strongly recommend including spec update in this commit

Steps

  1. Prepare commit

- Check for linting errors and fix them - If linting errors cannot be fixed automatically, STOP and report the issue. Ask user for guidance. - Run all tests locally to ensure they pass - If tests fail, STOP and fix them before committing. - Stage all changes - Create conventional commit message - Format: {type}: {description} ({TASK_KEY}) - Types: feat, fix, refactor, docs, test, chore - Description should be clear and concise - Example: feat: add user authentication (PROJ-123)

  1. Run Constitutional Review Gate

- Check if AGENTS.md exists: - Use glob_file_search or read_file to check for AGENTS.md in repo root - If AGENTS.md exists: - Read AGENTS.md Operational Boundaries (Tier 1: ALWAYS, Tier 2: ASK, Tier 3: NEVER) - Proceed with full Constitutional Review - If AGENTS.md does NOT exist: - Log: "⚠️ AGENTS.md not found - Constitutional Review skipped" - Store status: "Constitutional Review: SKIPPED (AGENTS.md not found)" - Proceed to next step (commit and push) without Constitutional Review - If AGENTS.md exists, continue with review: - If spec exists, read specs/{FEATURE_DOMAIN}/spec.md for functional context - Get code diff: - Use run_terminal_cmd: git diff main...HEAD to get all changes in PR - Include file paths, additions, deletions - Invoke Critic Agent using Cursor's built-in review: - Use Cursor's native review capabilities (no external API keys required) - The agent executing /complete-task acts as the Critic Agent, using Cursor's built-in review features - No external dependencies: All review happens within Cursor IDE using native capabilities - Prompt structure for the review: You are a Critic Agent performing Constitutional Review. **Constitution** (from AGENTS.md Operational Boundaries, if AGENTS.md exists): [If AGENTS.md exists: Paste full text of Tier 1 (ALWAYS), Tier 2 (ASK), Tier 3 (NEVER) sections] [If AGENTS.md does not exist: Note "AGENTS.md not found - Constitutional Review skipped. Proceeding with commit/PR without Constitutional Review."] **Spec** (functional context): [If spec exists: Paste Blueprint and Contract sections] [If no spec: Note "No spec found - validate against Constitution only"] **Code Changes** (git diff main...HEAD): [Paste full diff output] **Task**: Validate code against Constitution. For each violation found: 1. **Category**: CRITICAL (Tier 3 NEVER) | WARNING (Tier 2 ASK) | INFO (Tier 1 ALWAYS) 2. **Description**: What constitutional principle was violated? 3. **Impact**: Why this matters (performance, security, maintainability, scalability) 4. **Remediation**: Ordered steps to fix the violation 5. **Location**: File:Line reference where violation occurs Output structured markdown report using this format: ## Constitutional Review **Status**: PASSED | FAILED | WARNING ### Tier 3 Violations (CRITICAL - NEVER) [List each violation OR state "None"] ### Tier 2 Violations (WARNING - ASK) [List each violation OR state "None"] ### Tier 1 Violations (INFO - ALWAYS) [List each violation OR state "None"] ### Summary - Total violations: X - Critical: Y (Tier 3) - Warnings: Z (Tier 2) - Info: W (Tier 1) - Parse Critic output: - Extract violations by severity (CRITICAL, WARNING, INFO) - Count violations in each tier - Store full report for PR body - Gate Decision (only if AGENTS.md exists and review was performed): - If CRITICAL violations found (Tier 3: NEVER): - STOP immediately - Display full violation report with remediation steps - Report: "❌ Constitutional Review FAILED. PR creation blocked due to Tier 3 violations." - Instruct user: "Fix violations listed above and retry /complete-task" - Do NOT proceed to commit/push/PR steps - If only WARNING/INFO violations (Tier 2, Tier 1): - Log: "⚠️ Constitutional Review passed with warnings/suggestions." - Store report for inclusion in PR body - Proceed to next step (commit and push) - If no violations: - Log: "✅ Constitutional Review: PASSED" - Store "PASSED" status for PR body - Proceed to next step - If AGENTS.md does not exist: - Skip Constitutional Review - Log: "⚠️ AGENTS.md not found - Constitutional Review skipped" - Store "SKIPPED" status for PR body - Proceed directly to commit and push (no review gate)

  1. Commit and push changes

- Commit staged changes with the conventional commit message - Push to remote branch - If push fails (e.g., authentication, network, conflicts), STOP and report the error. - If branch doesn't exist on remote, it will be created on first push.

  1. Create pull request (optional)

- Note: PR creation is optional. You may skip this step if you prefer to create the PR manually or if it's not needed. - If creating PR: CI/CD is a PR gate. Local tests passing is the prerequisite to PR creation. CI/CD will run automatically after PR is created. - If skipping PR: Proceed directly to Step 6 (Update issue) after pushing.

  1. Create pull request (if proceeding with PR creation)

- After pushing, get the latest commit SHA: - Use run_terminal_cmd: git rev-parse HEAD to get current commit SHA - Or use mcp_github_list_commits with sha = branch name to get latest commit - Read documentation for PR body: - First, check for Spec at specs/{FEATURE_DOMAIN}/spec.md: - If exists, extract: Context (from Blueprint), Definition of Done (from Contract), key scenarios - Include in PR body as "Feature Spec" section - Then, check for Plan at .plans/{TASK_KEY}-*.plan.md: - If exists, extract: Story, Context, Scope, Acceptance Criteria, Implementation Steps summary - Include in PR body as "Implementation Plan" section - If neither exists: Note in PR body that no detailed documentation was found - Check if PR already exists: - Use mcp_github_get_pull_request with method="get" to check for existing PRs - If PR exists, update it instead of creating new one - Create completed checklist comment for the issue (include PR link if PR was created): ## Completed Checklist - [x] Plan reviewed and implemented - [x] Code changes completed - [x] Unit tests written and passing - [x] Constitutional Review passed - [x] Spec updated (if behavior changed) - [x] Documentation updated - [x] Linting errors fixed - [x] All tests passing locally - [x] Changes committed and pushed - [x] PR created (if applicable) Pull Request: {PR_URL} (only if PR was created) - Create PR with: - Title: {type}: {description} ({TASK_KEY}) (matching commit message) - Body should include: - Constitutional Review report (from Step 2): - If PASSED: Include "✅ Constitutional Review: PASSED" - If WARNING/INFO: Include full Constitutional Review report with violations - If SKIPPED: Include "⚠️ Constitutional Review: SKIPPED (AGENTS.md not found)" - Constitutional Review report should appear first in PR body - Feature Spec summary (if spec exists) with Blueprint context and Contract DoD - Implementation Plan summary (if plan exists) with extracted sections - Note: "CI/CD checks will run automatically on this PR" - Link to the issue - PR body template: ## Summary [Brief description of changes] ## Constitutional Review [Insert Constitutional Review report from Step 2] - If PASSED: "✅ Constitutional Review: PASSED - No violations found" - If WARNING/INFO: Full report with Tier 2 and Tier 1 violations - If SKIPPED: "⚠️ Constitutional Review: SKIPPED (AGENTS.md not found) - Proceeded without Constitutional Review" ## Feature Spec [If spec exists, include Context and Definition of Done] ## Implementation Plan [If plan exists, include key implementation steps] ## Verification - ✅ Tests: All passing locally - ✅ Linting: No errors - ✅ Constitutional Review: [PASSED | WARNING] ## Related Issue Closes {TASK_KEY} - Set base branch (typically main or develop) - Common failure scenarios: - Branch doesn't exist on remote: Ensure branch was pushed successfully - PR already exists: Check using mcp_github_get_pull_request before creating - Permission errors: Verify GitHub MCP authentication - Invalid base branch: Verify base branch exists - If PR creation fails, STOP and report the specific error with context. - Link PR to the issue (using issue tracker's linking mechanism) - After PR creation, monitor CI/CD status: - Use mcp_github_get_pull_request with method="get_status" to check CI/CD status - CI/CD will run automatically as a gate on the PR - Note status in PR body or comments as needed

  1. Update issue

- If PR was created: - Add PR link as a comment to the issue - Format: Pull Request: {PR_URL} - If PR was not created: - Add comment indicating changes are committed and pushed, ready for review - Format: Changes committed and pushed to branch {type}/{TASK_KEY}. Ready for review. - Transition issue to "Code Review" status: - First, get available transitions using mcp_atlassian_getTransitionsForJiraIssue - Find transition to "Code Review" status - Transition using mcp_atlassian_transitionJiraIssue - Verify issue status is now "Code Review" - If transition fails (permissions, workflow rules), STOP and report the error.

Tools

MCP Tools (Atlassian)

  • mcp_atlassian_atlassianUserInfo - Verify Atlassian MCP connection
  • Obtaining CloudId for Atlassian Tools:

- Method 1 (Recommended): Use mcp_atlassian_getAccessibleAtlassianResources - Returns list of accessible resources with cloudId values - Use the first result or match by site name - Only call if cloudId is not already known or has expired - Method 2: Extract from Atlassian URLs - Jira URL format: https://{site}.atlassian.net/... - CloudId can be extracted from the URL or obtained via API - Error Handling: If cloudId cannot be determined, STOP and report: "Unable to determine Atlassian cloudId. Please verify MCP configuration."

  • mcp_atlassian_getJiraIssue - Fetch story details by {TASK_KEY}

- Parameters: cloudId, issueIdOrKey = {TASK_KEY}

  • mcp_atlassian_getTransitionsForJiraIssue - Get available status transitions

- Parameters: cloudId, issueIdOrKey = {TASK_KEY}

  • mcp_atlassian_transitionJiraIssue - Transition issue to "Code Review" status

- Parameters: cloudId, issueIdOrKey = {TASK_KEY}, transition = {id: "transition-id"}

  • mcp_atlassian_addCommentToJiraIssue - Add completed checklist and PR link comments

- Parameters: cloudId, issueIdOrKey = {TASK_KEY}, commentBody = markdown content

MCP Tools (GitHub)

  • A lightweight read-only GitHub MCP tool to verify connection (see Cursor Settings → Tools & MCP for exact names)
  • mcp_github_list_branches - List branches (verify current branch exists)

- Parameters: owner, repo

  • mcp_github_list_commits - Get latest commit SHA from branch

- Parameters: owner, repo, sha = branch name

  • mcp_github_get_commit - Get commit details

- Parameters: owner, repo, sha = commit SHA, include_diff = false

  • mcp_github_get_pull_request - Get PR details (check if PR already exists)

- Parameters: owner, repo, pullNumber, method = "get"

  • mcp_github_create_pull_request - Create new pull request

- Parameters: owner, repo, title, head = {type}/{TASK_KEY}, base = main (or default), body = PR description

  • mcp_github_get_pull_request - Get PR status and CI/CD checks

- Parameters: owner, repo, pullNumber, method = "get_status"

Filesystem Tools

  • glob_file_search - Find specs matching specs/{FEATURE_DOMAIN}/spec.md or plans matching .plans/{TASK_KEY}-*.plan.md

- Pattern for plans: **/.plans/{TASK_KEY}-*.plan.md - Pattern for specs: **/specs/*/spec.md

  • read_file - Read spec/plan file content for PR body

- Spec location: specs/{FEATURE_DOMAIN}/spec.md - Plan location: .plans/{TASK_KEY}-{description}.plan.md

  • read_lints - Check for linting errors

- Parameters: paths = array of file paths or directories

Terminal Tools

  • run_terminal_cmd - Execute git and test commands

- git status - Check current branch and uncommitted changes - git branch --show-current - Get current branch name - git rev-parse HEAD - Get current commit SHA - git diff main...HEAD - Get code changes for Constitutional Review - git add. or git add <files> - Stage changes - git commit -m "{type}: {description} ({TASK_KEY})" - Commit with conventional format - git push origin {type}/{TASK_KEY} - Push branch to remote - npm test or pytest or project-specific test command - Run tests locally - npm run lint or project-specific lint command - Check linting

Completion Checklist

  • MCP status validation performed
  • Current branch verified (matches {type}/{TASK_KEY} format)
  • All tests passing locally
  • Documentation files checked (spec and/or plan)
  • Spec updated if behavior changed (Same-Commit Rule)
  • Linting errors fixed
  • Changes staged
  • Commit message follows convention
  • Changes committed
  • Constitutional Review Gate executed
  • Constitutional Review passed (or warnings only)
  • Pushed to remote
  • PR created (optional - may be skipped)
  • If PR created: CI/CD checks running (as PR gate)
  • Completed checklist added to issue
  • If PR created: PR created with spec/plan/review summary
  • If PR created: PR linked to issue
  • If PR created: PR link added to issue comment
  • Issue updated with status (PR link or commit/push confirmation)
  • Issue transitioned to "Code Review"

Guidance

Role

Act as a software engineer responsible for completing development work on a task and preparing it for code review. You are methodical, thorough, and follow established development workflows and standards.

Instruction

Execute the complete-task workflow to finalize development work on a specified task. This includes:

  1. Performing prerequisite validation checks
  2. Preparing and committing changes following conventions
  3. Pushing changes to remote
  4. Optionally creating a pull request with proper documentation (may be skipped)
  5. Updating the issue tracker with completion status

Context

  • The task is tracked in an issue management system (Jira, Azure DevOps, etc.)
  • Specs may exist at specs/{FEATURE_DOMAIN}/spec.md with permanent feature contracts
  • Plans may exist at .plans/{TASK_KEY}-*.plan.md with transient implementation details
  • AGENTS.md defines 3-tier Operational Boundaries (Constitution)
  • Constitutional Review validates code against Constitution before PR creation using Cursor's built-in review capabilities (no external API keys required)
  • Implements ASDLC Review Gate between Quality Gates and Acceptance Gates
  • Development work has been completed on a branch following the {type}/{TASK_KEY} format
  • MCP integrations provide access to issue trackers and version control
  • Automated CI/CD pipelines run checks on pushed branches
  • The codebase follows conventional commit standards and branch naming conventions
  • Same-Commit Rule: Spec updates must be committed with code changes that affect behavior/contracts
  • ASDLC patterns: Constitutional Review, The Spec, The PBI, Context Gates
  • ASDLC pillars: Quality Control (Review Gate), Standardized Parts (Constitution)

Examples

ASDLC: Constitutional Review — Validates changes against AGENTS.md before PR creation.

Example Branch Names:

  • feat/PROJ-123 (story/feature)
  • fix/FB-6 (bug fix)
  • chore/KAN-42 (maintenance task)
  • refactor/PROJ-321 (refactoring)

Example Commit Messages:

  • feat: add user authentication (PROJ-123)
  • fix: resolve login timeout error (FB-6)
  • refactor: simplify auth service (PROJ-321)
  • test: add unit tests for auth module (PROJ-123)
  • docs: update API documentation (PROJ-123)

Example PR Title:

  • feat: add user authentication (PROJ-123)

Example PR Body:

## Summary
Implements user authentication feature as specified in PROJ-123.

## Constitutional Review
✅ **Constitutional Review: PASSED** - No violations found

All code changes comply with AGENTS.md Operational Boundaries:
- Tier 1 (ALWAYS): Command structure standards followed
- Tier 2 (ASK): No high-risk operations without approval
- Tier 3 (NEVER): No security violations or anti-patterns detected

## Feature Spec
[If spec exists at specs/user-authentication/spec.md]

### Context
User authentication is required to secure access to the application. OAuth2 provides a standard, secure authentication method.

### Definition of Done
- [ ] User can log in using OAuth2
- [ ] Session persists for 24 hours
- [ ] Failed attempts are logged
- [ ] Tests pass and coverage maintained

### Key Scenarios
- **Given** user has valid OAuth2 credentials
- **When** user attempts login
- **Then** session is created and user is authenticated

## Implementation Plan
[If plan exists at .plans/PROJ-123-*.plan.md]

### Implementation Steps
1. Set up OAuth2 library and configuration
2. Implement AuthService for OAuth2 flow
3. Create SessionManager for session handling
4. Add AuthMiddleware for request validation
5. Create login/logout API endpoints

## Verification Status
- ✅ Build: Passing
- ✅ Tests: All passing (95% coverage)
- ✅ Linting: No errors
- ✅ Constitutional Review: PASSED
- ⏳ Security scan: Pending

## Related Issue
Closes PROJ-123

Example Completed Checklist Comment (with PR):

## Completed Checklist

- [x] Plan reviewed and implemented
- [x] Code changes completed
- [x] Unit tests written and passing
- [x] Constitutional Review passed
- [x] Spec updated (if behavior changed)
- [x] Documentation updated
- [x] Linting errors fixed
- [x] All tests passing locally
- [x] Changes committed and pushed
- [x] PR created
- [x] Automated checks passing

Pull Request: https://github.com/owner/repo/pull/42

Example Completed Checklist Comment (without PR):

## Completed Checklist

- [x] Plan reviewed and implemented
- [x] Code changes completed
- [x] Unit tests written and passing
- [x] Documentation updated
- [x] Linting errors fixed
- [x] All tests passing locally
- [x] Changes committed and pushed to branch feat/FB-6

Ready for review.

Constraints

Rules (Must Follow):

  1. Operational Standards Compliance: This command follows operational standards (documented in AGENTS.md if present, but apply universally):

- MCP Tool Usage: Check schema files, validate parameters, handle errors gracefully - Safety Limits: Never commit secrets, API keys, or sensitive data. Never commit changes automatically without user review. - AGENTS.md Optional: If AGENTS.md exists, Constitutional Review is performed. If missing, review is skipped and command proceeds. - See AGENTS.md §3 Operational Boundaries (if present) for detailed standards

  1. Same-Commit Rule: If code changes API contracts, data models, or quality targets → update spec in same commit:

- Verify spec was updated in staged changes - If spec exists but wasn't updated: WARN user and recommend including spec update - Spec and code changes must be committed together

  1. Constitutional Review Gate: Before PR creation, validate code against AGENTS.md:

- Use Cursor's built-in review capabilities (no external API keys required) - Run Critic Agent validation against 3-tier Operational Boundaries (Tier 1: ALWAYS, Tier 2: ASK, Tier 3: NEVER) - BLOCK PR creation on Tier 3 (NEVER) violations - display violations and STOP - WARN on Tier 2 (ASK) and Tier 1 (ALWAYS) violations but allow PR creation - Include Constitutional Review report in PR body - If CRITICAL violations found, instruct user on remediation before retrying

  1. Prerequisites Must Pass: Do not proceed if MCP validation, branch verification, or test verification fails. STOP and report the issue.
  2. Conventional Commits: All commits must follow the format: {type}: {description} ({TASK_KEY})

- Valid types: feat, fix, refactor, docs, test, chore - Description should be clear and concise - Task key must be included in parentheses

  1. Branch Naming: Current branch must match short format {type}/{TASK_KEY} (e.g., feat/FB-6, fix/PROJ-123)

- Use short format: feat/FB-6 (not descriptive format) - Important: Be consistent - use short format for all branches

  1. Test Requirements: All tests must pass locally before committing. Do not commit failing tests.
  2. Linting: All linting errors must be fixed before committing. If errors cannot be fixed automatically, STOP and ask for guidance.
  3. PR Creation is Optional: PR creation is optional. You may skip PR creation if you prefer to create it manually or if it's not needed. If creating PR:

- CI/CD runs automatically after PR creation as a gate. Local tests passing is the prerequisite to PR creation. Monitor CI/CD status after PR is created. - PR body should include: - Constitutional Review report (from Step 2) - Feature Spec summary (if spec exists at specs/{FEATURE_DOMAIN}/spec.md) - Implementation Plan summary (if plan exists at .plans/{TASK_KEY}-*.plan.md) - If neither spec nor plan exists, note this in PR body

  1. Documentation File Handling:
  • Check for both spec and plan files
  • If neither exists, WARN but proceed (PR will lack detailed context)
  • Prefer spec content over plan content when both exist
  1. Error Handling: If any step fails (push, PR creation, issue transition), STOP and report the specific error. Do not proceed with remaining steps.

Existing Standards (Reference):

  • MCP status validation: See mcp-status.md for detailed MCP server connection checks
  • Spec guidance: See specs/README.md for Same-Commit Rule and when to update specs
  • Constitutional Review: See AGENTS.md Operational Boundaries for 3-tier system
  • Branch naming: Type prefix format ({type}/{TASK_KEY}) as established in start-task.md
  • Commit message format: {type}: {description} ({TASK_KEY}) (consistent across commands)
  • Documentation locations: Specs at specs/{FEATURE_DOMAIN}/spec.md, Plans at .plans/{TASK_KEY}-*.plan.md
  • Issue workflow: Tasks transition through "To Do" → "In Progress" → "Code Review" → "Done"
  • ASDLC patterns: The Spec (permanent state), The PBI (transient delta), Living Specs (Same-Commit Rule), Constitutional Review (Review Gate), Context Gates (gate system)

Output

  1. Committed Changes: All changes committed with conventional commit format (including spec updates if behavior changed)
  2. Constitutional Review Report: Validation report showing PASSED, WARNING, or FAILED status with violations
  3. Pushed Branch: Branch pushed to remote repository (if Constitutional Review passed or warnings only)
  4. Pull Request (optional): If PR creation was chosen and Constitutional Review passed, PR created with Constitutional Review report, spec summary (if exists), plan summary (if exists), verification status, and issue link. CI/CD checks will run automatically.
  5. Updated Issue: Issue updated with completed checklist, status information (PR link if created, or commit/push confirmation if PR skipped), and transitioned to "Code Review" status

All outputs should be verified and any failures should be reported immediately with specific error details. CRITICAL Constitutional Review violations block PR creation.

平台分布

Codex

36.45%
按下载量换算28

Claude

27%
按下载量换算21

Cursor

18.25%
按下载量换算14

Gemini CLI

9.94%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

未通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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