Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计异常

github-auto-implementGitHub auto 实现

Agent Skill

用于围绕 GitHub 仓库、Issue、Pull Request、分支、提交和代码协作流程提供辅助能力。它适合让 Agent 查询项目状态、整理变更、辅助创建或检查协作事项,并把仓库中的信息转成可执行的下一步。使用时需要区分只读查询和写入操作;涉及创建 PR、修改 Issue、推送分支或访问私有仓库时,应确认 token 权限、目标仓库范围和用户授权。

总安装

186

周安装

8

GitHub Stars

1

下载量

65
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:github-auto-implement(GitHub auto 实现)
来源仓库:https://github.com/costicapuntaru/agentica
仓库路径:skills/github-auto-implement
安装命令:
npx skills add https://github.com/costicapuntaru/agentica --skill github-auto-implement
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/costicapuntaru/agentica --skill github-auto-implement

简介

围绕 GitHub 仓库、Issue、PR 等协作流程提供辅助能力,支持项目状态查询与变更整理。

  • 适用于查询项目信息、梳理代码变更、协助创建或检查协作事项,并将仓库内容转化为可执行操作。
  • 通过 npx skills add 命令从指定仓库安装,需区分只读查询和写入操作。
  • 涉及 PR 创建、Issue 修改或私有仓库访问时,需确认 token 权限与仓库授权范围。
  • github-auto-implement 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Purpose

Pick the next unassigned issue from GitHub, investigate it formally, implement it using TDD, and open a PR—all in one shot.

How to Run (Daemon Mode)

This skill is designed to be run repeatedly by a daemon. To start the auto-implement loop:

node resources/scripts/github-daemon.mjs

The daemon will poll for ready issues (defined in .agenTica.js, default: autobot:ready) and execute this workflow autonomously.

Optional Configuration

Create a .agenTica.ts (or .agenTica.js) in your project root to customize how issues are picked, how base branches are resolved, and switch between agents or models dynamically.

See the agenTica.sample.ts in the repo root for a full list of over 40 lifecycle hooks used by this workflow.

Step 1: Find the Next Issue

Use gh issue list to find ready, unassigned tickets.

# Query is configurable in .agenTica.js. Default:
gh issue list --search 'is:open is:issue label:autobot:ready -label:autobot:question -label:autobot:in-progress' --json number,title,url,labels,comments,body
  • Skip: Humans-only tickets or those requiring clarification.
  • Action: Pick the first ready issue.

Step 2: Branch Initialization

  1. Checkout {{BASE_BRANCH}} and pull the latest: git fetch origin git checkout {{BASE_BRANCH}} git pull origin {{BASE_BRANCH}}
  2. Create a new branch off {{BASE_BRANCH}}: git checkout -b {{BRANCH_NAME}}
Epic & dependency context: {{BASE_BRANCH}} is resolved by the daemon before this skill runs. - For one-off tickets: {{BASE_BRANCH}} = the repo default branch (main). - For epic tasks with Parent PRD: #N: {{BASE_BRANCH}} = epic/<N>-<slug> (the epic feature branch). - For tasks with Blocked by: #N: {{BASE_BRANCH}} = the blocking issue's open PR branch (PR chaining).

Step 3: Triage & Investigation (Investigation Phase)

Instead of jumping into code, investigate the issue formally.

  • Skill: triage-issue
  • Action: Run triage-issue on the picked ticket. Identify the root cause, establish a reproduction case, and draft a TDD-based fix plan.
  • Checkpoint: Save the triage result to github-auto-implement/tmp/triage/{{ISSUE_ID}}.md.

Step 4: Implementation (TDD Phase)

Execute the fix plan using a strict Red-Green-Refactor loop.

  • Skill: tdd
  • Action: Run the tdd skill using the plan generated in the Triage phase. Ensure each requirement has a corresponding test and that all tests pass.

Step 5: Post-Task Validation (Quality Phase)

Ensure the implementation meets project standards.

  • Skill: post-task-validation (and react-doctor if applicable).
  • Action: Run full validation (linting, type-checking, and all smoke tests). Fix any regressions autonomously.

Step 6: Documentation & PR Metadata

  1. Commit: git commit -m "Fixes #{{ISSUE_ID}}: {{ISSUE_TITLE}}"
  2. Push: git push -u origin {{BRANCH_NAME}}
  3. Write PR metadata to github-auto-implement/tmp/pr-meta.json (the daemon will create the PR): {"title": "Fixes #{{ISSUE_ID}}: {{ISSUE_TITLE}} [#{{ISSUE_ID}}]", "body": "<markdown body referencing triage notes, TDD results, and including 'Fixes #{{ISSUE_ID}}'>"}

- Title: Must include [#{{ISSUE_ID}}] so the daemon can locate it for dependent tasks. - Body: Reference the triage notes and TDD results. Include Fixes #{{ISSUE_ID}} so GitHub cross-references the issue.

Step 7: Final Status

Print DONE as the very last line of output.

Guardrails

  • 🤖 Prefix: Every comment posted to GitHub MUST start with the 🤖 emoji.
  • Clarification Requested: If the issue description is ambiguous during Triage:

1. Comment: gh issue comment <number> --body "🤖 Need clarification on...". 2. Edit Labels: Swap current inProgress label for the question label (as defined in .agenTica.js). 3. Output: Print the exact string CLARIFICATION_REQUESTED as the final output. 4. Stop: Reset the branch and stop immediately.

  • No as Casting: Follow typescript-no-as-casting (Zod validation preferred).
  • Tracer Bullets: If the task is large, ensures it can be implemented in a single vertical slice.
  • Recursive Grill: If requirements are unclear during drafting, re-trigger a mini-grill.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

32.98%
按下载量换算21

Claude

32.18%
按下载量换算21

Cursor

17.78%
按下载量换算12

Gemini CLI

9.99%
按下载量换算6

安全审计

Gen Agent Trust Hub

未通过

Socket

可疑

Snyk

未通过

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills