FIND-20260323-004 · 2026-03-23 · Innovation Veille
EurekaClaw — Multi-agent AI research assistant (UCLA AGI Lab)
framework
MEDIUM
Multi-agent AI research assistant that autonomously conducts academic research — literature crawling (arXiv + Semantic Scholar), hypothesis generation, 7-stage Lean4 theorem proving, LaTeX paper writing, browser UI with live agent tracking, continual learning that distills proof strategies into reusable skills. Claude-native (Sonnet 4.6). 240 stars, Apache 2.0.
Source
https://github.com/eurekaclaw/eurekaclaw
ODS Impact
Patterns to extract: (1) continual learning / skill distillation applicable to agent self-improvement loop, (2) arXiv + Semantic Scholar crawler complements HF Papers for veille, (3) live agent tracking UI pattern for ods-dashboard, (4) 7-stage pipeline pattern for complex multi-step agent workflows.
Security Review
License: Apache 2.0 | Maintenance: EARLY (240 stars, 13 commits) | Risk: LOW — local-first, no external deps beyond APIs | Recommendation: Study patterns, do not depend on as dependency
Tags
multi-agent
research
continual-learning
lean4
agent-ui