autoresearch-agent
Autonomous experiment loop that optimizes any file by a measurable metric. Inspired by Karpathy's autoresearch. The agent edits a target file, runs a fixed evaluation, keeps improvements (git commit), discards failures (git reset), and loops indefinitely. Use when: user wants to optimize code speed, reduce bundle/image size, improve test pass rate, optimize prompts, improve content quality (headlines, copy, CTR), or run any measurable improvement loop. Requires: a target file, an evaluation command that outputs a metric, and a git repo
Details
- Path
- skills/agent-eval/autoresearch-agent
- License
- MIT
- Bundled scripts
- 11
- Dependencies
- 3
Bundled scripts
- skills/agent-eval/autoresearch-agent/evaluators/llm_judge_content.py
- skills/agent-eval/autoresearch-agent/evaluators/benchmark_speed.py
- skills/agent-eval/autoresearch-agent/evaluators/test_pass_rate.py
- skills/agent-eval/autoresearch-agent/evaluators/build_speed.py
- skills/agent-eval/autoresearch-agent/evaluators/memory_usage.py
- skills/agent-eval/autoresearch-agent/evaluators/llm_judge_copy.py
- skills/agent-eval/autoresearch-agent/evaluators/llm_judge_prompt.py
- skills/agent-eval/autoresearch-agent/evaluators/benchmark_size.py
- skills/agent-eval/autoresearch-agent/scripts/setup_experiment.py
- skills/agent-eval/autoresearch-agent/scripts/log_results.py
- skills/agent-eval/autoresearch-agent/scripts/run_experiment.py