collect-evidence

To aggregate GitHub pull request metadata and commits into a structured bundle, use an automated skill to fetch PR titles, bodies, changelists, and comments. It assembles this data in-memory while preserving raw content for downstream processing without writing to disk.

Yes, you can enrich GitHub PR evidence with context from Linear, Slack, and Notion. The aggregation process retrieves optional sources like Linear issues, Slack threads, and Notion pages when available, integrating them directly into the in-memory evidence bundle.

The best way to automate collecting pull request data for batch processing workflows is using a dedicated aggregation skill. It applies to batch-refine workflows across PRs, fetching metadata and changed files to form an in-memory bundle and exposing a sufficiency verdict for orchestration.

Collecting PR evidence does not require writing temporary files to disk. The aggregation process preserves raw content and operates entirely in-memory, forming an evidence bundle that can be consumed directly by downstream stages without touching the file system.

To determine if collected pull request evidence is sufficient for automated review, the aggregation process exposes a sufficiency verdict. This verdict evaluates the assembled in-memory bundle of PR metadata, commits, and comments, guiding downstream orchestration on whether to proceed.

Yes, you can use Greptile comments to enrich the pull request evidence bundle. The collection process fetches PR-related evidence and optionally integrates Greptile comments alongside Linear issues, Slack threads, memento notes, and Notion pages when available.