Run iterative experiments applying one hypothesis-driven change per iteration.
npx skills add https://github.com/kok1eee/o-m-cc --skill experiment-kok1eee
Please help me install this Agent Skill. Skill: experiment Source: https://github.com/kok1eee/o-m-cc/tree/main/skills/experiment Command: npx skills add https://github.com/kok1eee/o-m-cc --skill experiment-kok1eee
Experiment provides a structured, repeatable loop to test single-code changes, measure their impact, and keep or revert them based on numeric evidence, eliminating ad-hoc trial-and-error and ambiguous outcomes.
Start an experiment to reduce API latency by recording a baseline with a measurement command, iterating one change at a time, and updating progress.md with results.
High-intent search queries and answers about installing and using this skill.
To reduce API latency or memory usage, run iterative experiments by applying one hypothesis-driven change per iteration, measuring its numeric impact, and committing or reverting the change based on the results. Progress is tracked in a progress.md file.
The best way to test single refactoring changes is to use an isolated sub-agent workflow that implements one change at a time, runs a measurement command, and keeps or reverts the code based on numeric evidence recorded in progress.md.
Yes, you can integrate Monitor for asynchronous measurement of long-running benchmarks or test suites, allowing the experiment workflow to continue planning the next iteration while the current performance measurement runs in the background.
Yes, performance optimization using this iterative experiment workflow requires version control to commit or revert changes and the ability to spawn fresh agent contexts per iteration to apply isolated, hypothesis-driven modifications.
You should use an iterative experiment workflow for exploratory bug-fixing when the root cause is uncertain, or for UI/UX A/B-style tweaks, as it requires applying a single change, measuring the numeric impact, and reverting if the metric does not improve.
You track baseline metrics, current best results, iterations, and failed hypotheses during code optimization by using a progress.md file as the single source of truth, ensuring all experimental changes are validated against numeric evidence.
Ronin15Compare SDL3 HammerEngine metrics against baselines to flag performance regressions.
eglenn-devGenerate git commit messages from status, diff, and log.
tiankong0101-byteMonitor long-running automated processes and recover them from stalls or deadlocks.
OpenCloudOSProfile Linux runtime events in memory for real-time system performance analysis.
