github-integration

DNYoussef

更新日 Today

78 閲覧

メタaimcp

について

このGitHub連携スキルは、PRレビュー、マルチリポジトリ操作、MCPツール統合によるプロジェクト管理など、GitHubに特化したタスクを調整します。構造化されたルーティングに安全ガードレールと標準化された手順を備え、一貫したGitHub自動化を実現します。組み込みのコンプライアンスとツール連携を活用し、複数のGitHubワークフローを自動化または調整する必要がある場合にご利用ください。

クイックインストール

Claude Code

推奨

プラグインコマンド推奨

/plugin add https://github.com/DNYoussef/context-cascade

Git クローン代替

git clone https://github.com/DNYoussef/context-cascade.git ~/.claude/skills/github-integration

このコマンドをClaude Codeにコピー＆ペーストしてスキルをインストールします

ドキュメント

L1 Improvement

Added a centralized SOP in Prompt Architect style with Skill Forge guardrails for all GitHub subskills.
Documented routing, MCP tool expectations, and confidence ceilings.
Introduced structure-first documentation and memory tagging.

STANDARD OPERATING PROCEDURE

Purpose

Route and coordinate GitHub tasks across subskills (PR review, multi-repo, project management, releases, workflow automation) with consistent SOPs and MCP integrations.

Trigger Conditions

Positive: GitHub PR reviews, multi-repo coordination, project board automation, release orchestration, or workflow automation requests.
Negative: non-GitHub SCM or local-only tasks; route to platform-specific skills.

Guardrails

Structure-first docs: SKILL, README, MCP guide kept current.
Explicit routing to subskills; do not mix flows without stating boundaries.
Enforce least-privilege credentials; never log secrets.
Confidence ceilings required on analyses and automation changes.
Memory tagging for runs and auditability.

Execution Phases

Intent & Routing – Identify which subskill applies; confirm repository scope, permissions, and risk level.
Setup – Ensure MCP servers (Claude Flow, Flow Nexus if used) are configured; validate tokens; set WHO/WHY/PROJECT/WHEN tags.
Plan – Map actions, safety checks, and rollback; align with subskill SOP.
Execute – Run subskill workflows (review/multi-repo/project/release/actions) with logging and dry-runs where possible.
Validate – Verify results (tests, checks, approvals) and ensure no secrets leaked.
Deliver – Summarize actions, outputs, risks, and confidence line; archive in memory.

Output Format

Routed subskill(s), repo scope, MCP servers used, and actions taken.
Results/metrics, risks, and follow-ups.
Confidence: X.XX (ceiling: TYPE Y.YY) and memory namespace.

Validation Checklist

Correct subskill chosen; permissions confirmed.
MCP servers configured; secrets protected.
Actions logged with rollback/cleanup notes.
Memory tagged; confidence ceiling declared.

Integration

Subskills: PR review, multi-repo, project management, release management, workflow automation folders under this skill.
MCP: see MCP-INTEGRATION-GUIDE.md for commands; tag sessions with WHO/WHY/PROJECT/WHEN.
Memory MCP: skills/tooling/github-integration/{project}/{timestamp} for runs.

Confidence: 0.70 (ceiling: inference 0.70) – SOP aligns GitHub integrations with Prompt Architect and Skill Forge guardrails.

GitHub リポジトリ

DNYoussef/context-cascade

パス: skills/tooling/github-integration