Documentation v1.0
Docs GitHub Integration AI PR Summaries

AI PR Summaries

AI PR Summaries analyze a pull request's diff against the story's acceptance criteria and tasks, giving you an instant understanding of what the PR covers and what might be missing.

Requirements

  • Pro or Teams subscription
  • A linked GitHub repository with at least one PR connected to a story

Generating a summary

  1. Navigate to a story with a linked pull request
  2. In the Pull Requests section of the sidebar, click Generate AI Summary
  3. Wait a few seconds while Vibe analyzes the diff
  4. The summary appears inline below the PR

What's in a summary

Each AI PR Summary includes:

  • Changes Summary — Brief overview of what the PR does
  • Acceptance Criteria Coverage — Maps each criterion to Covered / Partially Covered / Not Covered
  • Task Coverage — Maps each task to whether it appears implemented
  • Potential Issues — Flags missing edge cases or concerns
  • Confidence Score — High / Medium / Low confidence that the PR fully addresses the story

Regenerating summaries

Summaries are cached on the PR record. If the PR is updated with new commits, click Regenerate to get a fresh analysis.

API and MCP access

AI PR Summaries are also available programmatically:

  • REST API: POST /api/v1/stories/:story_id/pr_summary with optional pr_id parameter
  • MCP Tool: summarize_pr with story_id and optional pr_id

Both default to the latest open PR if no pr_id is specified.

Does this count against AI usage?

Yes, generating a PR summary uses one AI generation from your monthly allowance.

Last updated July 18, 2026