Improve your AI agent through benchmark trials.
Optimizes AI agents by repeatedly evaluating benchmarks and evolving their prompts, skills, and memory.
When to use it
Use when optimizing agent state against benchmarks or building automated self-improvement loops.
Give it an agent optimization request; it builds a working evolution workflow and produces an optimized agent workspace.
The evolution workflow runs Python code importing agent_evolve; no Python version is specified.
The Bedrock authentication path requires configured AWS credentials; Anthropic authentication is presented as an alternative provider path.
This skill
Claude
Sends benchmark observations to Claude
a-evolve
Installs a-evolve