开发只读github未标认证来源可访问许可证需确认审计通过

planning-github-task（规划 GitHub task）

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

GitHub Stars

公开资料未说明

下载量

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

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：planning-github-task（规划 GitHub task）
来源仓库：https://github.com/b-mendoza/agent-skills
仓库路径：skills/planning-github-task
安装命令：
npx skills add https://github.com/b-mendoza/agent-skills --skill planning-github-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/b-mendoza/agent-skills --skill planning-github-task

简介

用于围绕 GitHub 仓库、Issue、PR 和代码协作流程提供辅助能力。

适合查询项目状态、整理变更、创建协作事项并生成可执行步骤。
使用时需区分只读查询与写入操作，避免越权修改。
涉及私有仓库或敏感操作时，应确认 token 权限和用户授权。
建议先预览变更内容再执行，防止意外覆盖或数据丢失。

SKILL.md

Planning GitHub Task

Plan exactly how to execute one task from the task plan for a GitHub issue. This orchestrator does three things: think (interpret summaries and plan state), decide (choose the next planning dispatch or re-plan path), and dispatch (hand execution-heavy work to co-located subagents). The planning artifacts live on disk; the orchestrator keeps only concise summaries in context.

Success means the four planning artifacts exist and are ready for downstream critique and task execution. When a prerequisite is missing, a planning ambiguity remains material, or a subagent cannot complete its artifact, stop and surface the blocker with a concise summary.

Inputs

Input	Required	Example
`ISSUE_SLUG`	Yes	`acme-app-42`
`TASK_NUMBER`	Yes	`3`
`RE_PLAN`	No	`true`
`DECISIONS_FILE`	No	`docs/acme-app-42-task-3-decisions.md`

Read ./references/data-contracts.md when you need the upstream prerequisites or downstream artifact expectations.

Use RE_PLAN and DECISIONS_FILE only for critique-driven reruns. Read ./references/pipeline.md when deciding which stages to rerun.

Pipeline Stages

execution-prepper validates the task and writes the execution brief.
execution-planner inspects the codebase and writes the execution plan.
test-strategist writes the behavior-driven test specification.
refactoring-advisor writes the refactoring recommendation.
The skill returns only a concise planning summary and the artifact paths needed for downstream critique and task execution.

Subagent Registry

Subagent	Path	Purpose
`execution-prepper`	`./subagents/execution-prepper.md`	Validate task readiness and assemble the execution brief
`execution-planner`	`./subagents/execution-planner.md`	Inspect the codebase and write the implementation plan
`test-strategist`	`./subagents/test-strategist.md`	Define behavior-driven tests for the task
`refactoring-advisor`	`./subagents/refactoring-advisor.md`	Recommend only the refactoring needed for this task

Read a subagent definition only when you are about to dispatch that subagent.

Output Contract

This skill writes only workflow-planning artifacts that capture planning state for this task:

Artifact	Produced by	Consumed by
`docs/<ISSUE_SLUG>-task-<TASK_NUMBER>-brief.md`	`execution-prepper`	All downstream planning subagents, downstream critique, task execution
`docs/<ISSUE_SLUG>-task-<TASK_NUMBER>-execution-plan.md`	`execution-planner`	Downstream critique, task execution
`docs/<ISSUE_SLUG>-task-<TASK_NUMBER>-test-spec.md`	`test-strategist`	Downstream critique, task execution
`docs/<ISSUE_SLUG>-task-<TASK_NUMBER>-refactoring-plan.md`	`refactoring-advisor`	Downstream critique, task execution

On a successful run, all four files exist. On a re-plan, overwrite only the files owned by the subagents that are re-run. These workflow-state documents stay on disk for the life of the workflow and are never committed to git.

Dispatch Rules

Read ./references/data-contracts.md when checking prerequisites or artifact expectations.
Read ./references/pipeline.md when running the standard planning pipeline or a critique-driven re-plan.
At each boundary, validate that the current required input exists before dispatch: the task plan and task section for execution-prepper, then the prior-stage artifacts for later subagents. Confirm that the expected output artifact was written before advancing.
Pass only the explicit handoffs required for the current stage: task identity inputs for execution-prepper, plus RE_PLAN and DECISIONS_FILE on critique-driven reruns; then prior-stage artifact paths and DECISIONS_FILE when applicable for later stages.
Keep only verdicts, file paths, and next-step-relevant notes from each subagent.
Advance sequentially. Rerun only the subagents invalidated by critique, plus their downstream dependents.
Surface blockers and pause for resolution when ambiguity remains.

Reference Guide

This table tells you which reference to load. ## Orchestration Steps remains the source of truth for the actual pipeline order.

Situation	Reference file	Purpose
Standard planning run	`./references/pipeline.md`	Dispatch order, handoffs, and success criteria
Re-plan after critique	`./references/pipeline.md`	Targeted rerun rules and retry limit
Contract questions	`./references/data-contracts.md`	Upstream prerequisites and downstream consumers

Orchestration Steps

Read ./references/data-contracts.md and confirm the task-plan prerequisites exist for ISSUE_SLUG and TASK_NUMBER.
Read ./references/pipeline.md.
Use ./references/pipeline.md to determine the next required subagent. On a standard run, start with execution-prepper. On a critique-driven re-plan, start at the earliest invalidated stage. Pass only that subagent's required explicit inputs. For execution-prepper, pass only ISSUE_SLUG, TASK_NUMBER, and when applicable RE_PLAN and DECISIONS_FILE. For later stages, pass the artifact file paths required by ./references/pipeline.md.
Retain only its verdict, artifact path, and the notes needed for the next decision.
Continue until all required planning artifacts are written and confirmed, or a blocker is surfaced.
Report completion with the task number and title, all four artifact paths, one or two sentences on the recommended approach, the number or shape of tests specified, and the refactoring verdict.

Example

Dispatch execution-prepper with ISSUE_SLUG, TASK_NUMBER
Subagent returns: PREP: PASS Brief: docs/acme-app-42-task-2-brief.md
Dispatch execution-planner with BRIEF_FILE=docs/acme-app-42-task-2-brief.md
Subagent returns: PLAN: PASS Execution plan: docs/acme-app-42-task-2-execution-plan.md
Dispatch test-strategist with BRIEF_FILE=docs/acme-app-42-task-2-brief.md and PLAN_FILE=docs/acme-app-42-task-2-execution-plan.md
Subagent returns: TEST_SPEC: PASS Spec: docs/acme-app-42-task-2-test-spec.md
Dispatch refactoring-advisor with BRIEF_FILE=docs/acme-app-42-task-2-brief.md, PLAN_FILE=docs/acme-app-42-task-2-execution-plan.md, and TEST_SPEC_FILE=docs/acme-app-42-task-2-test-spec.md
Subagent returns: REFACTORING: PASS Refactoring plan: docs/acme-app-42-task-2-refactoring-plan.md
Tell the user: "Task 2 - Add webhook retry handling planning complete. Artifacts: brief, execution plan, test spec, and refactoring plan written. Recommended approach: add retry orchestration in the webhook service, then thread retry state through the worker path. Tests: 3 behavior groups with 6 high-priority checks. Refactoring verdict: Refactor during implementation."

The orchestrator keeps only those summaries and the artifact paths.

适合场景

用户想查找某类 Agent Skill 时

需要根据任务场景推荐可安装能力包时

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

38.55%

按下载量换算25

Claude

29.17%

按下载量换算19

Cursor

18.58%

按下载量换算12

Gemini CLI

10.3%

按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容，本身偏只读；真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。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

GitHub 仓库

github.com

/b-mendoza/agent-skills

SKILL.md 来源

github.com

/b-mendoza/agent-skills/tree/main/skills/planning-github-task

skills-sh

skills.sh

/b-mendoza/agent-skills/planning-github-task

继续浏览同类 Skills

更多开发 Skills（21,746）返回 Skills 目录

上一个 Skill

planning（规划）

planning 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息，适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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。

下一个 Skill

planning-jira-task（规划 Jira task）

用于处理 Jira 项目、任务、缺陷、Sprint、负责人和状态流转。它适合让 Agent 辅助查询工单、汇总迭代进展、创建任务或整理需求和缺陷信息。使用时要确认项目权限、字段配置和工作流规则，不同团队的 Issue 类型、状态和必填字段可能不同；涉及批量改状态、改负责人或创建工单时，应先预览变更内容再执行。