Thanks for helping improve opencode-agentspawn.
We’re optimizing for agent reliability, collaboration, and safety. High-value contributions include:
- Better workspace patterns (doc hygiene, boot ordering, reuse-first/verify-first)
- Better example prompts (domain-agnostic; no real credentials)
- OpenCode config snippets (primary vs subagent workflows)
- Optional GitHub Actions for secret scanning (must be safe-by-default)
- Improvements to docs clarity
- Anything that requires sharing real secrets or account identifiers
- Provider-specific hacks or anything ToS-adjacent
- Keep templates minimal and copy/pasteable
- Prefer short, actionable docs over theory
- Avoid hardcoding personal paths, emails, IDs
- Fork / use a branch
- Make changes
- Open a PR with:
- what problem it solves
- why it’s safe
- examples (before/after)