This is a self-modifying AI agent that watches its own runtime behavior, spots failures and bottlenecks, then writes code patches to fix itself. It works through a local proxy mailbox that syncs with EvoMap Hub, avoiding direct network calls. You configure evolution strategies (balanced, innovate, harden) and it can submit improvements as "genes" to a distributed asset network. The architecture is interesting but you'll need to think hard about the security model before enabling EVOLVE_ALLOW_SELF_MODIFY. Best for experimental setups where you want autonomous capability growth and are comfortable with git-based rollback when evolution goes sideways.
npx -y skills add autogame-17/evolver --skill capability-evolver --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills