Hermes_Learning_Loop

Autonomous learning loop that analyzes tool execution traces to derive, verify, and deploy optimized Ethoswarm skills (SKILL.md and system_skill definitions), safely refines existing playbooks using a Two-Failure Rule, and recommends MCP server expansions while strictly enforcing Babbage Standard isolation and secret-safety constraints.