Collect GitHub evidence on your computer.

Download and run locally

The collector uses Python's standard library. It reads GitHub metadata and writes files on your computer. The site does not receive your credentials or evidence. Review the script before running it.

python3 collect_github.py OWNER/REPOSITORY --start 2026-09-01 --end 2026-09-30 --output ./evidence

Choose the repository and period authorized by your team. Public repositories need no token. For private repositories, add --allow-private --use-token and supply an authorized token in the local GITHUB_TOKEN environment variable. Use the least access needed; never paste a token into this site or a worksheet.

Review the output

Each run creates a new timestamped folder with packet.json and an importable worksheet.json. Repository settings are point-in-time snapshots; workflow metadata covers the requested dates. Permission failures and capped samples remain visible. Neither metadata nor a successful fetch establishes compliance.

Import the worksheet into the evidence planner and review the packet with the evidence reviewer skill. Unknown owners, reviewers, and deadlines stay blank.

Repeat collection

Add --repeat-seconds 86400 to refresh daily while the local process is running. Stop with Ctrl-C. Every refresh creates a separate snapshot; existing packets are retained. This does not install a background service. Period arguments stay fixed until you change them.

GitHub API limits and permissions apply. See rate limits and workflow API documentation. No workflow logs or artifacts are downloaded.