Can I restrict the tools available to an agent?

Yes, using the 'tools' field in the agent's frontmatter, you can define an array of specific tools the agent is allowed to use, following the principle of least privilege.

What is the character limit for agent system prompts?

System prompts for agents should generally be kept under 10,000 characters, with 500 to 3,000 characters being the recommended length for optimal performance.

How are agents triggered in Claude Code?

Agents are triggered via the 'description' field in their YAML frontmatter, which must include specific XML-tagged examples showing the context, user request, and intended response.

What does 'model: inherit' do?

The 'inherit' setting ensures the subagent uses the same Claude model (e.g., Sonnet or Opus) currently active in the parent Claude Code session.

What is a Claude Code Agent?

Agents are autonomous subprocesses within Claude Code plugins designed to handle complex, multi-step tasks independently without constant user intervention.

Agent Development Architect

Name: Agent Development Architect
Author: jasonsie

byjasonsie

0•

Herramientas para Desarrolladores

Streamlines the creation of autonomous agents for Claude Code plugins through standardized structures, triggering patterns, and system prompt best practices.

Agent Development Architect provides a comprehensive framework for building specialized, autonomous subprocesses within the Claude Code ecosystem. It guides developers through the technical requirements of agent construction, including YAML frontmatter configuration, precise triggering conditions using example-based descriptions, and the design of high-performance system prompts. By following these standardized patterns, developers can create reliable subagents that handle multi-step tasks, specialized analysis, and complex code generation with distinct personas and restricted tool access.

Características Principales

01Standardized Agent File Templates

02Role-Based System Prompt Design

03Tool-Restricted Execution Scopes

04Example-Based Triggering Logic

05Model Inheritance and Customization

060 GitHub stars

Casos de Uso

01Developing automated test generators that trigger on file saves

02Building specialized security auditors with read-only tool access

03Creating autonomous code reviewers with specific style guidelines

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jasonsie/prompt-dev agent-development

For use in Claude.ai and ChatGPT

Download Skill