AI Agents
mini-swe-agent
SWE-agent/mini-swe-agent
A minimalist hundred-line AI software engineering agent that reads code, runs shell commands, and fixes GitHub issues while scoring strongly on SWE-bench Verified.
Related projects
claw-code195.3kRust-built autonomous coding agent and self-maintained museum exhibit created with Gajae-Code and LazyCodex, capable of understanding codebases and performing development tasks without human intervention.
pi108.8kOffers a unified LLM interface, agent runtime loop, terminal UI, and coding-agent CLI that reads, edits, and executes code to complete software development tasks.
TradingAgents108.2kMulti-agent LLM financial trading framework coordinating specialized agents for stock analysis, decision-making, and automated trading research for quantitative developers and researchers.
autoresearch96.6kAutonomous research system where AI agents iteratively run training experiments to optimize language models overnight on single-GPU setups with minimal supervision.