EvoMap/evolver

What it solves

Evolver addresses the problem of ad hoc, untraceable prompt tweaks in AI agent development. Instead of manually adjusting prompts based on trial and error, it provides a system to turn agent experiences and error patterns into auditable, reusable evolution assets called "Genes" and "Capsules."

How it works

Evolver operates as a prompt generator that scans a project's memory directory for runtime logs and error signals. It then selects the most appropriate Gene or Capsule from its asset store and emits a protocol-bound GEP prompt. This prompt guides the agent's next evolution step without directly editing source code. It can be run as a single command, in a review mode for human confirmation, or as a background daemon loop.

Who it’s for

It is designed for teams maintaining agent prompts and logs at scale, users who require deterministic and auditable evolution traces, and environments where protocol-bound changes are necessary.

Highlights

• GEP Protocol: Uses a standardized protocol for evolution with reusable assets.
• Auto-Log Analysis: Automatically scans memory and history files for errors and patterns to guide self-repair.
• Host Runtime Integration: Integrates with platforms like Cursor, Claude Code, and OpenClaw via setup hooks.
• Strategy Presets: Offers configurable strategies (balanced, innovate, harden, repair-only) to control the balance between new features and stability.
• Skill Store: Allows users to download and share reusable skills via the EvoMap network.