Run autonomous hill-climbing loops to optimize scalar objectives via scored candidate changes.
npx skills add https://github.com/thistleknot/skills --skill autoresearch-thistleknot
Please help me install this Agent Skill. Skill: autoresearch Source: https://github.com/thistleknot/skills/tree/main/autoresearch Command: npx skills add https://github.com/thistleknot/skills --skill autoresearch-thistleknot
Autoresearch eliminates slow, manual iteration when you need better results against a measurable metric by running an autonomous propose→measure→keep loop.
Use Case: You want higher unit-test pass rate or lower evaluation loss by repeatedly patching prompts, configurations, or code while logging every accepted change.
Ask an AI agent to run autoresearch on your metric by defining an objective and a scorer, then iteratively proposing small changes, measuring score deltas, and committing each accepted improvement.
High-intent search queries and answers about installing and using this skill.
Automate prompt optimization by defining a scalar objective and a scorer to evaluate candidates. An autonomous hill-climbing loop proposes small changes, measures score deltas, and commits only accepted improvements to iteratively raise your evaluation metrics.
Tune hyperparameters by running an autonomous iterative loop that proposes small configuration deltas and measures score changes. The system keeps only improvements, applying checkpointing and rollback to the best accepted state to ensure measured progress.
Yes, autonomous search applies to code patch optimization by iteratively proposing small code changes and measuring the score delta. It accepts only patches that improve your defined scalar objective, logging every accepted change while halting via budget or patience rules.
Hill climbing for metric optimization works by generating incremental candidate proposals from the current state using LLMs or structured patterns. It measures score deltas, keeps only accepted improvements, and stops when budget, patience, or convergence criteria are met.
Yes, the autonomous optimization loop requires deterministic or averaged scoring to reliably measure score deltas. It uses explicit budget and patience stopping criteria along with checkpointing and rollback to maintain the best accepted state during search.
Autonomous search loops should only accept improvements when properly configured with a scalar objective and measured score deltas. Ensure your scorer is deterministic or averaged, and verify that checkpointing and rollback rules are correctly keeping the best accepted state.
justinkang221Verify and optimize the second autocorrelation inequality score C with Dinkelbach and L-BFGS.
estebanrfpIteratively modify, execute, and evaluate targets against a defined metric.
lioartoilOptimize headless PR review prompts via iterative hill-climbing loops on GitHub.
xotongAutomate metric-driven iterative optimization with parallel experiments and gating rules.
