The Agent Skills Directory

Workflow Prompt Routing: The skill acts as a dispatcher, identifying user tasks and loading specific instructional files to guide the agent. This is a standard pattern for managing complex, multi-step agentic workflows.
External Resource Ingestion: The skill is designed to load prompts and guidelines from an external repository to ensure it uses the latest standards for GitHub Agentic Workflows.
Ingestion points: Fetches markdown files from the github/gh-aw repository (e.g., .github/aw/agentic-chat.md, skills/agentic-workflow-designer/SKILL.md).
Boundary markers: No explicit boundary markers or 'ignore' instructions are defined for the fetched content within the routing skill itself.
Capability inventory: The skill routes to sub-skills and prompts that handle workflow design, debugging, and optimization.
Sanitization: There is no explicit sanitization step for the fetched markdown content, which is a potential surface for indirect prompt injection if the source repository were compromised. However, this fetching behavior is central to the skill's primary purpose of providing current architectural guidance.

agentic-workflows