Do you sign AI-generated PRs and issues with the model name?

When an AI agent like Claude Code opens a pull request or issue, reviewers should know straight away that AI wrote it and which model did.

A generic "Generated with Claude Code" line doesn't say which model was used. Some tools also add a link to the agent session, which may expose private context.

Use one short footer that names the model.

❌ Figure: Bad example - No model information

🙂 Figure: OK example - Multiple lines of footer that repeat themselves

✅ Figure: Good example - One line for a PR, with the model name in bold

Why name the model?

• Review effort - Reviewers can decide how closely to check the work
• Traceability - If a model causes problems, you can find the PRs it wrote
• Transparency - Clients and team members can see where AI was used

Automate it with a Claude Code skill

Add a user-level skill at ~/.claude/skills/model-attribution/SKILL.md. The skill tells Claude to add this footer every time it creates or edits a PR or issue.

Tips

• Always use the model that actually did the work. Don't hard-code an old model name
• Never include links to agent sessions or transcripts
• Keep only one footer. When you edit a PR description, update the footer instead of adding another
• Keep using Co-Authored-By trailers on commits. The footer is for PR and issue descriptions only