# AI Tool Use Policy
Gest welcomes contributions made with AI tools, including agentic development workflows. However, we require a
**human in the loop** for all contributions.
## Policy
- **You are the author.** As a contributor, you are fully accountable for your contributions regardless of what tools
you used to create them.
- **Review your work.** You must carefully read and review all tool-generated code or text before submitting it for
review.
- **Be transparent.** Note when your contribution contains substantial amounts of tool-generated content.
- **Be prepared.** You should be able to answer questions about your work during review.
- **Stay in the loop.** Agents may assist with development, but a human must review and approve all actions that affect
our shared spaces -- pull requests, issues, comments, and code.
## Agents Welcome, With Oversight
Unlike many projects, we do not prohibit agentic development. AI agents can be powerful collaborators. What we prohibit
is **unsupervised** agent activity in our spaces. Specifically:
- Agents **must not** take action in our repositories (opening PRs, commenting on issues, posting reviews) without
explicit human review and approval of that specific action.
- Automated review tools that publish comments without human review are not allowed.
- A human must be the one to click "submit" -- not a cron job, not a webhook, not a bot.
## Rationale
Processing proposals, issues, PRs, and comments takes maintainer time and energy. Unreviewed tool output shifts effort
from the contributor to the reviewer, making it an extractive contribution. The human-in-the-loop requirement ensures
that contributors take ownership of their work and that maintainer time is respected.
## Handling Violations
If a maintainer determines that a contribution was submitted without meaningful human oversight, they may request
changes, close the PR, or lock the conversation at their discretion.