The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external markdown files to drive task execution via the build-task skill. This creates a surface for indirect prompt injection where a malicious blueprint could influence agent actions.
Ingestion points: The skill reads the file specified by the user-provided plan_path in SKILL.md.
Boundary markers: No specific delimiters or instructions are defined to separate the blueprint data from the agent's core instructions.
Capability inventory: The skill possesses the ability to invoke external skills (build-task and ship-it) and modify local files based on the input data.
Sanitization: The skill lacks validation or sanitization for task subjects or metadata extracted from the markdown table.

run-blueprint