What problem does it solve?
Running Claude-style skills (SKILL.md files with helper scripts) inside GitHub Agentic Workflows requires an MCP server that can discover, expose, and execute them. This Skill explains how to configure the Skillz MCP server with Docker so agents can call skills as tools.
Core Features & Use Cases
- Docker-based MCP configuration: Add Skillz as an MCP server in workflow frontmatter using the intellectronica/skillz container with a mounted skills directory.
- Flexible skill layouts: Supports flat Claude Code-compatible directories, nested folders, and zip-packaged skills, each containing a SKILL.md with YAML frontmatter.
- Environment and debugging options: Pass environment variables (such as API keys from secrets) and enable verbose logging for troubleshooting.
- Use Case: You have a folder of custom skills like summarize-docs and translate. Mount that folder into the Skillz container in your workflow so the AI agent can invoke those skills when triaging issues or processing documents.
Quick Start
Add the skillz MCP server to your workflow frontmatter with the intellectronica/skillz container, mount your skills directory at /skillz, and ask the agent to use the available skills to complete the task.