Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
Optim-agent assists machine learning engineers and researchers in refining hyperparameters for model training, inference, and reinforcement learning. It is designed for expensive black-box evaluations where analyzing the project structure can improve trial selection.
The tool provides a structured approach to optimizing quantitative strategies and scientific workflows, reducing the manual effort required to reach peak performance in complex system configurations.
WHO IT'S FOR
Machine Learning Engineers
optimizing model training and inference parameters
Quantitative Researchers
tuning quantitative strategies via black-box evaluations
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
In a terminal, install the stable Python package so it is importable. This option uses the stable source; do not also install the GitHub source variant.
In a terminal, install the latest Python package source from GitHub. This option uses the GitHub source; do not also install the stable package variant.
Installs: Complete optim-agent Claude Code plugin bundle, version 0.1.1 · Claude Code
Before you start
Python
File access and shell access
Authenticated Claude agent CLI on PATH
Run this command in a terminal to add the Optim-Agent marketplace and install its optim-agent plugin.
claude plugin marketplace add Optim-Agent/optim-agent && claude plugin install optim-agent@optim-agent
In a terminal, install the stable Python package so it is importable. This option uses the stable source; do not also install the GitHub source variant.
Installs: Complete optim-agent Claude Code plugin bundle, version 0.1.1 · Claude Code
Before you start
Python
File access and shell access
Authenticated Claude agent CLI on PATH
Run this command in a terminal to add the Optim-Agent marketplace and install its optim-agent plugin.
claude plugin marketplace add Optim-Agent/optim-agent && claude plugin install optim-agent@optim-agent
In a terminal, install the latest Python package source from GitHub. This option uses the GitHub source; do not also install the stable package variant.
Installs: Canonical optim-agent workflow supplied as the repository or SKILL.md · Claude Code, OpenCode/OpenClaw, another coding-agent environment
Before you start
Python
File access and shell access
At least one authenticated agent CLI on PATH: claude, codex, or opencode
Run from: agent-visible workspace
Manually place this repository or its SKILL.md in the agent-visible workspace, then ask the agent to follow the optim-agent workflow.
In a terminal, install the stable Python package so it is importable. This option uses the stable source; do not also install the GitHub source variant.
Agent-visible workspace with latest GitHub package
Installs: Canonical optim-agent workflow supplied as the repository or SKILL.md · Claude Code, OpenCode/OpenClaw, another coding-agent environment
Before you start
Python
File access and shell access
At least one authenticated agent CLI on PATH: claude, codex, or opencode
Run from: agent-visible workspace
Manually place this repository or its SKILL.md in the agent-visible workspace, then ask the agent to follow the optim-agent workflow.
In a terminal, install the latest Python package source from GitHub. This option uses the GitHub source; do not also install the stable package variant.
Measurable scalar objective and minimize/maximize direction
Trial budget
Evaluation command
Runtime or cost limit
Fixed workload or seed
Use the optim-agent skill to inspect <evaluation-entry-point> and <parameter-definition-files>. Prepare a concise pre-trial brief covering each tunable parameter's type, legal range, meaning, interactions, and operational constraints; the measurable scalar objective and whether to minimize or maximize it; the proposed trial budget; the exact evaluation command; runtime or cost limits; the fixed workload or seed; and the current/default configuration to use as the baseline. Treat the angle-bracketed paths as placeholders for my project files. Do not run any trials yet.
Claude
Codex
OpenClaw
OpenCode
DeepSeek