What problem does it solve?
Evolver turns scattered prompt tweaks and runtime failures into auditable, protocol-bound evolution steps so teams can systematically repair and improve AI agents without ad hoc, unsafe edits. It reduces manual triage by analyzing logs, selecting reusable Genes/Capsules, and emitting a GEP prompt that guides the next corrective or enhancement action. The project prioritizes Bahasa Indonesia for evolution explanations and audit logs when presenting human-facing narratives.
Core Features & Use Cases
- GEP Prompt Generation: Scan memory and logs, select Genes or Capsules, and emit a strict Genome Evolution Protocol prompt for deterministic evolution steps.
- Safe Solidify / Validation: Run gated validation commands, canary tests, and git-based rollback to protect the codebase during solidification.
- Local Hub Proxy & Worker Modes: Operate fully offline or connect to an EvoMap Hub via a local proxy for skill sharing, worker pool, and heartbeat.
- Use Case: Harden a flaky agent loop by detecting recurring error signals, selecting a repair Gene, and producing an auditable EvolutionEvent for review and controlled application.
Quick Start
Run node index.js to analyze memory, select a Gene, and produce a GEP prompt for human review or downstream execution.