majiayu000/agent-patterns

majiayu000 のプロフィール

記録された最新のソース活動: 2026年10月2日 08:41

SkillsMP カタログ更新: 2026年10月17日 00:09

収集済み skills: 7

GitHub スター: 1
GitHub フォーク: 0

リポジトリの README

README の元ファイル：README.md

README の元ファイルを読む

README 全文を表示

# Agent Patterns Skill Pack Reusable workflow skills for Codex and Claude Code, covering PR review, CI diagnosis, context handoff, session-history mining, and human review gates. Use them to carry evidence, scope, failure handling, and verification through repeatable coding-agent work. Each skill keeps its workflow in `SKILL.md`, with bundled scripts and references where needed. This repository is a skill pack; it does not run an agent service. ## Quick start ```sh git clone https://github.com/majiayu000/agent-patterns.git cd agent-patterns python3 skills/agent-patterns/scripts/pattern_tool.py list python3 skills/agent-patterns/scripts/pattern_tool.py search "pr review" ``` Read the selected skill's `SKILL.md` before using it. To install a skill, copy its whole folder into your host's configured skill directory, preserving its `scripts/` and `references/` resources. Invoke it through that host's skill interface; the listing helper itself does not execute the workflow. Start with [PR review](skills/pr-review-risk-plan/SKILL.md), [CI failure diagnosis](skills/ci-failure-diagnosis/SKILL.md), or [context handoff](skills/context-handoff-pack/SKILL.md). See the [meta skill](skills/agent-patterns/SKILL.md) for the pattern contract and [Spellbook mapping](docs/spellbook-mapping.md) for porting notes. ## Choose a skill for the problem | Your task | Start with | Bring this evidence | | --- | --- | --- | | Review a PR before changing or merging it | [PR Review Risk Plan](skills/pr-review-risk-plan/SKILL.md) | PR URL or local diff, base/head, intended behavior, and relevant checks | | Explain a red CI run before proposing a fix | [CI Failure Diagnosis](skills/ci-failure-diagnosis/SKILL.md) | Run URL, failing job/command, exact commit, and available logs | | Resume after compaction or hand work to another agent | [Context Handoff Pack](skills/context-handoff-pack/SKILL.md) | Repo path, current goal, changed files, decisions, verification, and one next action | | Require a human decision before changes land | [Review Gate](skills/review-gate/SKILL.md) | Proposed diff, risks, verification, and the exact action requiring approval | After installing the whole skill folder, select it through your host's skill interface and give it a concrete request, for example: > Use CI Failure Diagnosis for this run: `<run URL>`. The repository is > `<repo path>` and the failing head is `<commit SHA>`. Read the failing job > evidence, identify the smallest defensible cause, and report which checks you > completed before proposing a fix. For a handoff, use Context Handoff Pack with the exact repository path and goal. Its local-history probe can help find session evidence; historical logs remain hints until current Git, runtime, and test state are verified. You can use it with live workspace evidence when local history is unavailable. `pattern_tool.py search` finds manifest metadata. It does not invoke an agent, execute the selected workflow, grant approval, or prove a fix works. This is **majiayu000/agent-patterns**, a workflow skill pack; each linked `SKILL.md` defines its own evidence and failure contract. For a reproducible script or packaging problem, open a [repository issue](https://github.com/majiayu000/agent-patterns/issues) with the skill, command, version or commit, and redacted error. See [LICENSE](LICENSE) for reuse terms. ## Skills - `skills/agent-patterns`: meta skill for creating, auditing, validating, and packaging executable workflow patterns. - `skills/pr-review-risk-plan`: concrete PR review workflow skill. - `skills/ci-failure-diagnosis`: concrete CI root-cause workflow skill. - `skills/context-handoff-pack`: concrete long-session handoff workflow skill with a Codex local-history probe. - `skills/agent-session-pattern-miner`: concrete workflow-mining skill that turns local Codex/Claude session history into new-skill and enhancement candidates. - `skills/review-gate`: explicit human approval gate before agent-generated changes land. - `skills/skill-lifeguard`: reliability contract for evolving skills from observed failures. The scripts under `skills/agent-patterns/scripts/` are bundled skill resources. `pattern_tool.py` validates and searches lightweight manifests in each concrete skill's `references/pattern-manifest.json`. Each runnable session-history skill vendors its small utility module inside its own skill folder so copying the folder remains standalone. Scripts are not the main product surface. `SKILL.md` is the source of truth for workflow behavior. Manifests are search/index metadata only. ## Checks ```sh QUICK_VALIDATE="${SKILL_CREATOR_QUICK_VALIDATE:-${CODEX_HOME:-$HOME/.codex}/skills/.system/skill-creator/scripts/quick_validate.py}" python3 "$QUICK_VALIDATE" skills/agent-patterns python3 "$QUICK_VALIDATE" skills/pr-review-risk-plan python3 "$QUICK_VALIDATE" skills/ci-failure-diagnosis python3 "$QUICK_VALIDATE" skills/context-handoff-pack python3 "$QUICK_VALIDATE" skills/agent-session-pattern-miner python3 "$QUICK_VALIDATE" skills/review-gate python3 "$QUICK_VALIDATE" skills/skill-lifeguard python3 skills/agent-patterns/scripts/pattern_tool.py validate SKILL_CREATOR_QUICK_VALIDATE="$QUICK_VALIDATE" python3 -m unittest discover -s tests -v ```

リポジトリの README を読み込んでいます…

このリポジトリの skills

ソフトウェア品質保証アナリスト・テスターその他コンピュータ職ソフトウェア開発者

3 件の職業カテゴリ · 86% 分類済み

skill

職業分類

説明

更新

agent-patterns

その他コンピュータ職

Create, convert, audit, and package executable AI agent workflow patterns as Codex/Claude-style skills. Use when the user wants an agent workflow library, reusable workflow schema, pattern-to-skill conversion, skill pack design, exporter/validator behavior,…

原文の言語: 英語

2026-10-02

agent-session-pattern-miner

その他コンピュータ職

Mine local Codex and Claude chat/session records into reusable agent workflow candidates. Use when the user asks to inspect agent chat history, find repeatable workflows, decide what should become a skill, compare new-skill candidates against existing skills,…

原文の言語: 英語

2026-10-02

context-handoff-pack

未分類

Create compact, evidence-backed handoff packs for long-running agent tasks. Use before context compaction, Codex/Claude session handoff, task resumption, or when the user asks for exact current status across files, constraints, tests, decisions, blockers,…

原文の言語: 英語

2026-10-02

pr-review-risk-plan

ソフトウェア品質保証アナリスト・テスター

Review pull requests by collecting diff context, ranking behavioral risks, identifying missing tests, and producing verification commands. Use for non-trivial PR review, merge-readiness checks, review comment preparation, or turning review findings into a…

原文の言語: 英語

2026-08-19

review-gate

ソフトウェア品質保証アナリスト・テスター

Force an explicit review step before any agent-generated diff is applied, committed, or landed. Collect diff, run risk analysis, produce a review pack, and require human approval of the plan and risks. Use before git commit, push, PR creation, or direct…

原文の言語: 英語

2026-08-19

skill-lifeguard

ソフトウェア品質保証アナリスト・テスター

Create, audit, and harden skills so they become self-maintaining instead of cat-and-mouse. Use when designing new skills, reviewing existing ones for drift, or turning brittle workflows into reliable ones. Forces Negative Examples, Verification Checkpoints,…

原文の言語: 英語

2026-08-19

ci-failure-diagnosis

ソフトウェア開発者

Diagnose failing CI by collecting fresh check evidence, reproducing the smallest failing command locally when possible, identifying one defensible root cause, and proposing a minimal fix plan. Use when CI is red, tests differ between local and remote, or the…

原文の言語: 英語

2026-06-30

収集済み skill 7 件中 7 件を表示しています。

AI アシスタントでインストール

このプロンプトをコピーして、使用中の AI アシスタントに送ってください。

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/majiayu000/agent-patterns.