An experimental research kernel for bounded recursive self-improvement. Runs capability benchmarks, proposes patches, and applies only those that pass validation and formal checks.
machine-learning research ai program-synthesis agi self-improvement benchmark-suite ai-safety github-actions symbolic-reasoning recursive-self-improvement bug-repair external-code-fixtures
-
Updated
Jun 7, 2026 - Python