The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon instructions found in project-specific files such as tasks.md, design.md, and delta.md. This represents a potential attack surface where malicious content in those files could attempt to influence the agent's behavior.
Ingestion points: Reads files from the local repository including AGENTS.md, CONVENTIONS.md, design.md, tasks.md, and delta.md.
Boundary markers: The instructions do not specify the use of delimiters or markers to isolate content read from these files.
Capability inventory: The skill can invoke other specialized skills for codebase context analysis, debugging, and documentation, and it can update decision logs and task files.
Sanitization: There is no explicit logic mentioned for sanitizing the content of the markdown files before processing the instructions within them.

implement-task