A flexible, multi-interface approach for building LLM applications.
npx skills add https://github.com/vitallium/dotfiles --skill ce-optimize-vitallium
Please help me install this Agent Skill. Skill: ce-optimize Source: https://github.com/vitallium/dotfiles/tree/main/dot_agents/skills/ce-optimize Command: npx skills add https://github.com/vitallium/dotfiles --skill ce-optimize-vitallium
metric-driven optimization over software artifacts, enabling teams to define goals, run parallel experiments, measure outcomes, and converge on the best solution.
Define a spec with a measurable target and run a first serial baseline to validate the harness.
High-intent search queries and answers about installing and using this skill.
Automate metric-driven optimization by defining a spec with a measurable target, running parallel experiments, and evaluating outcomes using deterministic hard gates or LLM-as-judge scores to converge on the best solution.
Systematically improve search relevance and clustering quality by running iterative optimization loops that define baselines, apply sampling strategies, and evaluate results against structured scoring rubrics.
Yes, you can run parallel optimization experiments safely because the system maintains a durable experiment-log and strategy digest that ensure reproducibility and provide crash resilience during execution.
Evaluate prompt quality automatically by defining scoring rubrics and using LLM-as-judge scores to assess measurable outcomes, eliminating manual review while ensuring systematic experimentation.
Yes, the optimization approach supports per-experiment worktrees and dependency management, allowing both serial and parallel execution to run isolated tests without conflicts.
A strategy digest is a structured record that ensures reproducibility and transparent branch merges by maintaining a durable experiment-log of baselines, hypotheses, and evaluation results across iterative optimization loops.
Ouro-labsRun parallel experiments to optimize measurable goals with LLM-as-judge scores.
THUFanZdTurn EXPERIMENT_PLAN.md into runnable GPU experiments and collect results.
xotongAutomate metric-driven iterative optimization with parallel experiments and gating rules.
Dianshu-LiaoImplement LLM evaluation with automated metrics and LLM-as-judge patterns.
