The Agent Skills Directory

[SAFE]: The skill provides a structured template and workflow for documenting project requirements. It operates on local codebase data and conversation context, using tools like Write and Bash for legitimate purposes such as file creation and local script execution.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests untrusted data from the codebase to synthesize a PRD. However, the risk is minimal as the output is a text-based documentation template.
Ingestion points: Codebase files (read via Read, Grep, Glob tools) and conversation context.
Boundary markers: Absent; no specific delimiters or warnings are used to prevent the agent from following instructions found in external data.
Capability inventory: Bash, Write, and Edit tools are available for file system and shell operations.
Sanitization: Absent; no sanitization or filtering is applied to the ingested content before it is processed.

to-prd