skills.rest
NVIDIA-NeMo
NVIDIA-NeMoOFFICIAL
Explore all skills by this author →
WebsiteGitHub

skills-optimization

Run containerized evaluation suites and apply automated code improvements to agent skills.

58|16|Updated May 14, 2026
Search Tags:#automation#debugging#optimization#agent#evaluation#nemo
One-click install
npx skills add https://github.com/NVIDIA-NeMo/nemo-platform --skill skills-optimization
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skills-optimization
Source: https://github.com/NVIDIA-NeMo/nemo-platform/tree/main/plugins/nemo-agents/src/nemo_agents_plugin/skills/skills-optimization
Command: npx skills add https://github.com/NVIDIA-NeMo/nemo-platform --skill skills-optimization

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill addresses the challenge of maintaining and improving agent performance by automating the evaluation, analysis, and iterative refinement of agentic skills.

Core Features & Use Cases

  • Automated Evaluation: Run containerized eval suites (Harbor or NAT) to benchmark agent performance.
  • Failure Analysis: Cluster regressions and generate hypotheses to identify why specific evaluations fail.
  • Skill Optimization Loop: Execute an end-to-end cycle of evaluation, analysis, and automated code refinement to improve agent behavior.
  • Use Case: Use this to automatically optimize an agent's skill set by running a suite of tests, analyzing the results, and applying verified improvements to the codebase.

Quick Start

Run the skills optimization loop on the current agent using the configuration defined in the agent-improver.yml file.

Frequently Asked Questions about skills-optimization

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate agent skill evaluation and iterative code refinement?▼

Automate agent skill evaluation by running containerized eval suites to benchmark performance, clustering regressions to analyze failures, and applying verified code improvements iteratively to refine the agent's behavior.

How does failure clustering work for agentic workflow regressions?▼

Failure clustering analyzes evaluation regressions by grouping similar failures together and generating hypotheses to identify why specific agentic evaluations fail, enabling targeted code improvements.

Do I need Docker and Claude CLI to run automated skill optimization?▼

Yes, autonomous code edits and verification require the Claude CLI and Docker to execute containerized evaluation suites and perform the end-to-end skill optimization loop.

Can I use Harbor or NAT for containerized agent benchmarking?▼

Yes, you can run containerized evaluation suites using either Harbor or NAT task execution to benchmark agent performance and identify regressions for automated improvement.

What is the best way to optimize an agent's skill set using evaluation suites?▼

The best way is executing an end-to-end optimization loop that benchmarks performance, analyzes failure clusters, and applies autonomous code edits to iteratively improve the agent skill set.

Related Skills

Explore More Software Engineering Skills→
TheWayWithinTheWayWithin
13

saas-onboarding

Design and automate multi-step SaaS user onboarding flows with activation milestones.

Community
Intermediate
tan-yong-shengtan-yong-sheng
1

gstack-canary

Monitor live apps for console errors, performance regressions, and page failures.

Community
Advanced
onchainengineeronchainengineer

tbench

Automate Unix agent benchmarking with Terminal-Bench 2.0 on Docker or Daytona.

Community
Advanced
OneWave-AIOneWave-AI
245

code-review-pro

Review code for security vulnerabilities, performance bottlenecks, and maintainability issues.

Official
Basic