Optimize bounded system parameters against scalar objectives using ask/tell trials.
npx skills add https://github.com/Optim-Agent/optim-agent --skill optim-agent
Please help me install this Agent Skill. Skill: optim-agent Source: https://github.com/Optim-Agent/optim-agent/tree/main Command: npx skills add https://github.com/Optim-Agent/optim-agent --skill optim-agent
This Skill helps you optimize configurable system parameters against a measurable scalar objective when evaluations are expensive, noisy, or too complex to tune by hand. It is designed for situations where reading the project code can improve the quality of trial proposals and reduce wasted search budget.
Ask the agent to read the project, identify the tunable parameters and objective, then propose and record one valid trial using optim-agent's ask/tell workflow.
High-intent search queries and answers about installing and using this skill.
Hyperparameter tuning for expensive model training is handled by reading project code to understand parameter interactions, proposing bounded trials, and recording results through an ask/tell workflow. This repository-aware trial selection reduces wasted search budget and ensures auditable outcomes.
Inference latency and cost trade-offs can be optimized using coding agents that read the target repository, respect declared parameter spaces, and validate proposals. The Skill applies bounded search and safe fallback behavior to handle black-box objectives and recover from failed trials.
The ask/tell workflow for black-box optimization drives studies through explicit trial creation, evaluation, and result recording. It requires declaring bounded parameter spaces, validating proposals against them, and maintaining a trial history so optimization remains auditable and resumable.
Quantitative strategy optimization with repository-aware trial selection starts by having the agent read project files to identify tunable parameters and measurable scalar objectives. You then declare bounded parameter spaces and use the ask/tell workflow to propose, validate, and record trials.
Reinforcement learning controller tuning is a supported use case. The Skill optimizes bounded system parameters against measurable scalar objectives, using coding agents to read project code, validate proposals against declared spaces, and record results safely.
When a trial proposal is invalid or fails, the Skill applies safe fallback behavior. It validates proposals against the declared bounded parameter space and records outcomes through the ask/tell study history, ensuring the optimization process recovers without wasting search budget.
hoavdcAnalyze performance gaps and map them to measurable learning objectives.
Lord1EgyptLoad, run, and fine-tune pre-trained transformer models for NLP, vision, and audio tasks.
djd1mGenerate a Claude Code toolkit from SPARC or idea2prd documentation.
lucifertrjMinimize continuous black-box objectives using Particle Swarm Optimization.
