Contributing with AI Agents
AI coding assistants and autonomous agents are welcome to contribute to Hypernate – but AI-assisted contributions are held to exactly the same standards as any other. This page explains how we communicate our conventions to agents and the policy that applies when you use them.
Machine-Readable Guidance: AGENTS.md
The repository root contains an AGENTS.md file – the open, vendor-neutral convention that most coding agents (Claude, Codex, Cursor, Windsurf, and others) read automatically.
It is a terse, imperative summary of the build, test, formatting, commit, and review rules described throughout this section.
If you operate an agent, point it at the repository and it will pick up AGENTS.md on its own.
If you maintain your own agent configuration, keep it in sync with AGENTS.md rather than duplicating the rules.
Policy for AI-Assisted Contributions
- You are responsible for what you submit. Whether code is written by you or generated by a tool, signing off your commits (
git commit -s, our DCO requirement) certifies that you stand behind it. - Disclose substantial AI generation. If a pull request is largely AI-generated, say so in the description so reviewers have the right context.
- The design gate applies to agents too. Do not let an agent open a pull request for any change – trivial fixes included – before the corresponding issue is labeled
design/approved. An automated check enforces this (see Issue Triage and the Design Workflow). - Review before you submit. Do not open pull requests with unreviewed or untested output.
Run
./gradlew buildand read the diff first. - Low-effort automated submissions will be closed. Bulk or unverified AI pull requests and issue reports waste maintainer time and are not welcome.
Why This Matters
Hypernate receives a growing share of automated, LLM-assisted contributions. Clear, machine-readable guidance raises the quality of well-behaved agents, while the design-approval gate and these expectations keep the contribution process healthy for everyone.