Plugin Claude Code
Autonomous research loops for any domain. 10 commands: core loop, plan, debug, fix, predict, security, scenario, reason, ship.
30 2mo ago A
tokens not measured
original MIT
Autonomous research loop skill for LLM agents — inspired by Karpathy's autoresearch. Define a goal, let the agent iterate.
Plugin Claude Code
Autonomous research loops for any domain. 10 commands: core loop, plan, debug, fix, predict, security, scenario, reason, ship.
Plugin Claude Code
Bundles 13 skills · 1,507 tokens together
Autonomous research loops with 10 commands. Generalizes Karpathy's autoresearch loop to any domain with mechanical evaluation, overnight persistence, and zero dependencies.