Builder

EvoMap

1 Tool

EvoMap is an infrastructure platform for AI agent "self-evolution," built around its Genome Evolution Protocol (GEP), letting agents record, share, and reuse successful strategies ("Genes") and case histories ("Capsules") in an auditable improvement loop. An independent 2026 measurement study by researchers at the Hong Kong University of Science and Technology (Guangzhou) confirms the system operates largely as described, but finds its improvement goals mostly fail in practice: about 98% of published assets are never reused, and most approved assets pass review via vacuous validation tests rather than genuine checks. Its GitHub org also shows real third-party engagement, including externally authored pull requests.