Run autonomous optimization loops with atomic changes and numeric verification.
npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-loop-jangtrinh
Please help me install this Agent Skill. Skill: ck:loop Source: https://github.com/jangtrinh/demo-Lam/tree/main/.claude/skills/ck-loop Command: npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-loop-jangtrinh
ck:loop helps you improve a measurable metric through repeated, controlled experiments with automatic commit/rollback so you can move toward better results without manually tracking every try.
Run ck:loop with a Goal, a file Scope glob, and a Verify command that prints exactly one numeric value so it can iteratively improve that metric.
High-intent search queries and answers about installing and using this skill.
Autonomous iterative optimization works by applying one atomic change per iteration, using a Verify command that outputs a single number to decide whether to keep or discard the modification. It commits before verification and reverts regressions while preserving experiment history for pattern learning.
Automatically reverting regressions is handled by guard-and-revert logic that commits before verification and rolls back any changes failing the Guard check. This ensures unrelated correctness is preserved while moving toward better measurable results.
Metric-driven verification requires a Verify command that prints exactly one numeric value to confirm whether an atomic change improved the target metric. Iteration outcomes, including deltas and keep/discard status, are appended to a TSV file for trend analysis.
A clean git working tree is a strict precondition enforced before starting autonomous optimization loops. This ensures the system can safely commit experiments, track iteration deltas, and revert regressions without conflicting with uncommitted local changes.
Tracking progress during autonomous performance tuning is handled by appending iteration outcomes to a loop-results.tsv file. This log records the delta, keep or discard status, and explanations for each attempt, enabling trend analysis and pattern learning from the experiment history.
hotriluanRun iterative experiments to optimize measurable metrics in git repositories.
EdgeHuntRun iterative optimization loops against measurable metrics in git-tracked projects.
oabdelmaksoudExecute build, type check, lint, test, security scan, and diff review.
Sanchez-78Automate troubleshooting and performance improvement for underperforming crypto trading bots.
