The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external files that are part of the project repository, such as specification files and implementation patches, which could be used as a vector for indirect prompt injection.\n- Ingestion points: Reads a context JSON file, markdown specifications (spec_file), and implementation patches from the project and code roots.\n- Boundary markers: The instructions direct the agent to focus on specific markdown headers and tags, such as <frozen-after-approval> and ## Review Triage Log.\n- Capability inventory: The skill has the ability to read and write files within the project and code directories. It is explicitly instructed not to perform network operations, implement code, or run tests.\n- Sanitization: There is no evidence of explicit sanitization or filtering of the content read from the external artifacts before it is interpreted by the agent.

bmad-loop-resolve