Run an autonomous loop that modifies code, measures metrics, and keeps or reverts changes.
npx skills add https://github.com/shuretokki/void --skill autoresearch-shuretokki
Please help me install this Agent Skill. Skill: autoresearch Source: https://github.com/shuretokki/void/tree/main/.agents/skills/autoresearch Command: npx skills add https://github.com/shuretokki/void --skill autoresearch-shuretokki
Autoresearch removes the manual grind of incremental performance or behavior tuning by running an autonomous loop that proposes changes, commits them, runs a user-defined metric command, measures results, and keeps or reverts changes based on objective outcomes. It ensures every experiment is measured, tracked, and reversible so humans don't have to manually try and evaluate many small modifications.
Start an autonomous optimization run by telling the agent your goal, providing the exact metric command and extraction rule, listing which files are in-scope, and confirming the setup.
High-intent search queries and answers about installing and using this skill.
Automate iterative code optimization by running an autonomous experimentation loop that modifies code, executes a reproducible metric command, and keeps or reverts changes based on measured outcomes. It uses git commits to ensure every modification is tracked and reversible.
Autonomous experimentation handles reverting failed changes by using git-based commits and reverts. After running the metric command, if the numeric outcome does not improve, the loop automatically reverts the commit while logging the results.
Yes, you can use autonomous experimentation to increase test pass rates in a git repository. It applies to programming projects where improvements are quantified by reproducible commands, such as test pass counts or benchmark timings.
Before starting an autonomous experimentation loop, you need a git repository with terminal access, a concrete metric command with an extraction method, and explicit definitions of in-scope and out-of-scope files. The interactive setup guides you through defining these constraints and the experiment budget.
Limitations include requiring a reproducible metric command that yields a numeric outcome and repeated run capability. The loop avoids modifying out-of-scope files and requires explicit user approval for any environment or dependency changes during the experimentation process.
The best way to reduce memory usage is running an autonomous experimentation loop that proposes code modifications, measures the resource output, and keeps the commit only if it improves the metric. Results are tracked in a log for objective evaluation.
msd-dev-labAutomate iterative code reviews with read-only Codex CLI and Claude Code.
jimmymalhanCompare current test run metrics against historical baselines to flag anomalies.
htooayelwinictAutomate TDD workflows across Pest, Vitest, and Playwright test layers.
jonmummAutomate machine-learning research loops by modifying train.py and running fixed 5-minute experiments.
