The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to analyze arbitrary content from the local codebase to generate technical specifications.
Ingestion points: references/step-01-understand.md (Quick Orient Scan) and references/step-02-investigate.md (Read and Analyze Code).
Boundary markers: No explicit delimiters or isolation instructions are used when processing codebase content.
Capability inventory: File system read/write access and local task invocation.
Sanitization: No filtering or escaping of ingested codebase data is performed.
[COMMAND_EXECUTION]: Step 4 of the workflow (references/step-04-review.md) involves the invocation of a local adversarial review task located at {project-root}/_bmad/core/tasks/review-adversarial-general.xml. This is a structured use of internal task-chaining capabilities within the agent framework.

quick-spec