This policy establishes guidelines for the responsible and productive use of AI tools across our Accord Project codebases. We believe AI can accelerate development, improve quality, and lower barriers to contribution — but only when used with transparency, accountability, and respect for the community.
AI-assisted contributions are welcome and encouraged. However, transparency is non-negotiable.
- Disclose AI usage: Contributors should indicate when AI tools were used to generate or substantially modify code, documentation, or other artifacts — pull requests descptions must have a completed AI checklist.
- Own the output: The contributor remains fully responsible for any AI-generated content. Submitting AI output does not transfer accountability — you must review, understand, and stand behind every line.
- Verify correctness: AI-generated code must meet the same quality bar as human-written code. Do not raise pull requests for content you have not validated against the project's standards, tests, and conventions.
- Respect licensing and IP: Ensure AI-generated content does not introduce copyrighted material, proprietary code, or license-incompatible snippets. When in doubt, err on the side of caution or ask the community.
AI makes it easy to generate large volumes of code quickly. That does not make it appropriate to submit large volumes of unvetted work.
- Quality over quantity: A well-crafted, focused PR is worth far more than a sprawling AI-generated changeset. Review and refine AI output before submitting.
- Follow contribution guidelines: AI-generated contributions must adhere to the same templates, commit conventions, and review processes as any other contribution.
- Avoid low-value noise: Do not use AI to mass-generate trivial PRs (cosmetic rewrites, bulk reformatting, speculative refactors) that create review burden without meaningful impact.
- Respond to feedback: If a maintainer requests changes on an AI-assisted PR, engage thoughtfully — do not simply re-prompt the AI and push a new revision without understanding the feedback.
AI should amplify intentional effort, not replace it with guesswork.
- Start with issues: Prioritise work that addresses open issues, roadmap items, or documented feature requests. AI is most valuable when solving real, agreed-upon problems.
- Discuss before building: For non-trivial changes, open an issue or discussion first. Getting alignment on approach before generating code saves everyone's time.
- Align with project architecture: Use AI to work within the project's existing patterns and conventions, not to impose unfamiliar abstractions or frameworks.
- Avoid speculative contributions: Do not use AI to generate features, refactors, or "improvements" that no one has asked for. Unsolicited large-scale changes are unlikely to be merged and consume review capacity.
This policy is a living document and will evolve as AI tooling and community practices mature. Feedback and suggestions are welcome via issues or discussions.