load-pr-context（加载公关上下文）

Load PR Context

Name

github:load-pr-context - Load pull request context for the current branch

Synopsis

/load-pr-context [owner/repo] [branch]

Description

Loads the full context of an open pull request for the given branch, including description, changes, CI status, and review comments. Uses GitHub MCP tools for structured data access.

Implementation

Given owner/repo and branch arguments:

1. Use mcp__github__list_pull_requests with owner, repo, head: "OWNER:BRANCH", state: "open" to find the PR
2. If no PR is found, report that no open PR exists for this branch and stop
3. If a PR exists, gather context silently (do not narrate each step):

- Use mcp__github__get_pull_request with the PR number to read description, labels, reviewers, and metadata - Use mcp__github__get_pull_request_diff or mcp__github__list_pull_request_files to understand what has changed - Use mcp__github__list_workflow_runs with branch to check CI status - Check for review comments via mcp__github__list_review_comments_on_pull_request - Check for linked issues mentioned in the PR description

1. Present a concise summary:

- PR title, number, and status - Key points from the description - Files changed (count and notable files) - CI status (passing/failing) - Review status (approved, changes requested, pending) - Any blocking issues or failing checks

Example Interaction

User: /load-pr-context thebushidocollective/han feat/pr-context

Claude: ## PR #142: Add PR/MR context detection hooks

- **Status**: Open, 2 reviews pending
- **Description**: Adds SessionStart hooks to GitHub and GitLab plugins that detect open PRs/MRs on the current branch
- **Changes**: 8 files (+245, -12)
- **CI**: All checks passing
- **Reviews**: 1 approved, 1 pending from @reviewer

Ready to assist with this PR.

Arguments

• owner/repo (required): Repository in owner/repo format
• branch (required): Branch name to find the PR for

Tips

• This skill is typically invoked automatically when starting a session on a feature branch
• All data is fetched via MCP tools (no CLI commands needed)
• The summary is kept concise to avoid cluttering the session start