Agent harnessAI and agent development
self-evolve
Iterate agent skills with deterministic acceptance and rollback gates
- Stars
- 0
- Forks
- 0
- Updated
- Updated Jun 26, 2026
MIT
Iterate agent skills with deterministic acceptance and rollback gates
self-evolve provides a methodology and deterministic harness for improving a skill, repository, or project over repeated cycles. The loop reflects, proposes, evaluates, judges, and accepts changes, while code—not the model—decides acceptance, rejection, rollback, and which signal source to trust. It supports program checks, anchor verification, and generative evaluation for open-ended work where ground truth is limited.
Resource types
Use cases
Platforms
Build extensible coding agents with a unified model API
Build extensible agents for long-running, multi-step work
Runtime
Capabilities
Public GitHub facts last synced Jul 10, 2026.
Build and run open agents with tools, memory, permissions, and multi-agent coordination.
Run long-lived business workflows as recoverable multi-agent execution graphs.