EQUIPA
Active DevelopmentMulti-agent AI orchestration — your AI development team
About
EQUIPA is a multi-agent AI orchestration platform. It coordinates Claude-powered developer, tester, and reviewer agents in iterative dev-test loops to ship production-quality code. Named for the European Portuguese word for "team".
The orchestrator manages task creation, agent dispatch, progress tracking, error recovery, and cumulative learning. Each task gets an isolated git worktree, and agents retry with feedback until tests pass or budget is exhausted.
EQUIPA scored 54% on FeatureBench (3.4x the best published agent) and is competitive with top agents on SWE-bench Verified. Features include episodic memory with Q-values, autoresearch retry loops, ForgeSmith self-improvement, and a plugin system for extensions.
Features
- Multi-agent dev-test loops with Claude Opus and Sonnet
- Autoresearch retry loop inspired by Andrej Karpathy
- Cumulative knowledge with episodic memory and Q-values
- ForgeSmith self-improvement (rule-based + prompt evolution)
- Plugin system for private extensions via Python entry points