The Agent Skills Directory

[SAFE]: The skill implements a structured methodology for decision-making. Analysis of the instructions and provided documentation shows no signs of malicious patterns such as prompt injection, data exfiltration, or unauthorized command execution. The skill operates by orchestrating model perspectives and generating text-based Product Requirements Documents (PRDs). All external references point to legitimate documentation or existing repositories for methodology credit.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes arbitrary user decisions and product ideas. However, the risk is mitigated by design:
Ingestion points: User-supplied decisions or product ideas ingested in SKILL.md (Step 1).
Boundary markers: The 'Frame' step (Step 1) requires the agent to restate the input into a strict six-line format before analysis begins, creating a structural boundary.
Capability inventory: The skill is limited to text synthesis, calling other model instances for independent review, and referencing a secondary small-briefs skill for project breakdown.
Sanitization: The framing process acts as a normalization layer that prevents embedded instructions in user data from directly influencing the advisor lenses.

council