The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes GitHub Project items containing content from issues and pull requests, creating a surface for indirect prompt injection where malicious instructions could be embedded in data.
Ingestion points: Project items, issues, and pull request content ingested via list_project_items and projects_get.
Boundary markers: Explicit instructions are provided to treat issue and PR contents as untrusted data and ignore embedded instructions.
Capability inventory: The skill can modify project structure and items via the projects_write toolset.
Sanitization: Relies on the agent following the provided safety instructions to delimit and ignore untrusted content.- [NO_CODE]: The skill consists solely of Markdown instructions and metadata with no executable code provided.

project-planning