agent-creator

Configure Claude Code subagents with restricted tool access by defining least-privilege permissions, system prompts, and safety properties within standardized YAML frontmatter blueprints. The Skill validates these metadata blocks to ensure predictable, constrained autonomous behavior.

Orchestrate multiple Claude Code subagents in parallel workflows by defining sequential, parallel, or hierarchical execution patterns. This Skill generates structured blueprints that standardize multi-agent orchestration, enabling teams to rapidly compose autonomous agents with predictable behavior.

YAML frontmatter defines subagent behavior by extracting structured metadata including name, tagline, category, complexity, keywords, dependencies, components, and safety properties. The Skill locates and validates SKILL.md files to ensure strict formatting and non-ambiguous field values.

You can use this subagent orchestration approach for specialized code analysis tasks, as well as requirements gathering, security reviews, and automated testing. The centralized templates allow rapid configuration of focused Claude instances with explicit tool restrictions.

You need existing SKILL.md files at skill roots for the Skill to read YAML frontmatter and verify identity with the root directory. It locates and validates these files to extract structured blueprints, ensuring each valid skill generates one YAML document with all required fields populated.