Iteratively adjust code or AI models using metrics or LLM-as-judge evaluations.
npx skills add https://github.com/ylxai/hafiportrait-saas --skill optimize-ylxai
Please help me install this Agent Skill. Skill: Optimize Source: https://github.com/ylxai/hafiportrait-saas/tree/main/.kiro/skills/Optimize Command: npx skills add https://github.com/ylxai/hafiportrait-saas --skill optimize-ylxai
The Optimize skill unit helps in automatically optimizing various targets like code, AI models, and tasks. It allows users to either work with quantitative metrics (e.g., performance score, bundle size) or qualitative assessments based on binary evals using an LLM-as-judge system.
To run the Optimize skill for optimizing performance, type:
/optimize --metric "lighthouse_score" --higher-is-better
--measure "npx lighthouse http://localhost:3000 --output=json"
--extract "jq '.categories.performance.score * 100' lighthouse.json"
--files "src//*.tsx,src//*.css"
--budget 120
High-intent search queries and answers about installing and using this skill.
You can automate web performance tuning by running the optimization loop in metric mode, configuring it to measure Lighthouse scores and iteratively adjust source files to improve the results.
LLM-as-judge evaluation is a qualitative approach where an AI model assesses text outputs against custom criteria and test inputs, using binary judgments to iteratively improve prompts and skills.
You set up autonomous machine learning optimization by defining a target metric and measurement command, then allowing the system to iteratively modify model parameters and evaluate outcomes to maximize performance.
Yes, you can optimize code autonomously using either metric mode for quantitative measurements like bundle size or eval mode for qualitative feedback using LLM-as-judge binary evaluations.
The best way to iteratively improve AI agent performance is through an autonomous optimization loop that adjusts parameters, measures outcomes via custom eval criteria, and learns from successive improvements.
A primary limitation is the dependency on the defined measurement budget and the accuracy of your extraction commands; the autonomous optimization loop requires precise metric extraction to function correctly.
xlelibertyRun autonomous experiment loops to optimize target files using git and evaluator metrics.
nidhi-subrahBenchmark and optimize Codex-flow v3 performance with Flash Attention and HNSW indexing.
keshavraj936Run Lighthouse audits and analyze Core Web Vitals for web performance.
inbharataiAnalyzes code for bugs, security vulnerabilities, performance bottlenecks, and style violations.
