agent-improver

Create and audit Claude agent definitions with frontmatter validation.

7|3|Updated Mar 17, 2026

Search Tags:#audit #skills #agent #orchestration #claude-code #prompt-engineering #agent-management

Ready to run

npx skills add https://github.com/brainbrewlabs/brainbrew-devkit --skill agent-improver

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-improver
Source: https://github.com/brainbrewlabs/brainbrew-devkit/tree/main/plugin/skills/agent-improver
Command: npx skills add https://github.com/brainbrewlabs/brainbrew-devkit --skill agent-improver

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Streamlines creating, auditing, and improving Claude Code agents by providing templates, frontmatter validation, and guidance so teams avoid misconfigured or unsafe agents and speed up pipeline integration.

Core Features & Use Cases

Agent authoring: Clear template and examples for writing .claude/agents/{name}.md with required frontmatter and system prompt structure.
Audit and improvement: Checklists and frontmatter/schema references to validate name, description, tools, model selection, and skills preloading.
Chain and team integration: Guidance for adding agents to .claude/chains/, choosing between Agent teams, parallel subagents, and configuring routes, hooks, and memory scopes.

Quick Start

Create an agent named code-reviewer with a concise description, scoped tools, and required skills, then add it to the active chain flow.

Frequently Asked Questions about agent-improver

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

FAQPage Schema

How do I create a new Claude Code agent with the correct frontmatter and prompt structure?▼

To create a Claude Code agent, you write a `.claude/agents/{name}.md` file containing required frontmatter fields and a system prompt, using provided templates to satisfy frontmatter validation and ensure safe configuration.

What is the best way to integrate Claude agents into chain workflows?▼

Integrating Claude agents into chain workflows requires adding agent definitions to `.claude/chains/` and configuring routes, hooks, and memory scopes, choosing between agent teams or parallel subagents based on orchestration needs.

How do I configure agent teams and routes for Claude Code projects?▼

Configuring agent teams and routes involves defining orchestration logic in `.claude/chains/`, setting up memory scopes, and applying guidance for parallel subagents or sequential routes to structure agent interactions safely.

Can I validate agent frontmatter and model selection without running untrusted code?▼

Yes, you can validate agent frontmatter, model selection, and skills preloading through schema references and checklists without executing untrusted external code, ensuring safe agent definitions before deployment.