The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves reading software design documents, specifications, and proposals from the docs/sdd/ directory. These files represent untrusted data that could contain malicious instructions designed to subvert the task planning process if the files are sourced from external contributors or malicious repositories.
Ingestion points: The skill uses the Read() tool to ingest markdown artifacts from docs/sdd/{change-name}/ and mem_search() to retrieve related observations from the agent's memory.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potentially adversarial content embedded within the design documents.
Capability inventory: The skill has permissions to read files, write new task plans to the local disk, and save data to the agent's memory (Engram MCP).
Sanitization: The wiring logic does not specify any sanitization, filtering, or validation steps for the content ingested from the design artifacts.

sdd-tasks