script-brief
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of arbitrary user-provided story ideas, character settings, and novel fragments in
SKILL.md. This content is directly structured into a creative brief intended as instructions for a downstream agent (drama-script). The absence of explicit boundary markers or instruction isolation mechanisms creates a surface where malicious instructions embedded in the user's story ideas could potentially override the intended behavior of the downstream creative process. - Ingestion points: Processes user-provided story ideas and fragments in the input section of
SKILL.md. - Boundary markers: Absent; the system does not use delimiters (e.g., XML tags or specific tokens) to isolate user-provided data from the instructional template.
- Capability inventory: The skill itself does not invoke sensitive tools, but its output is a primary input for the
drama-scriptcapability. - Sanitization: Absent; the instructions focus on thematic extraction and expansion rather than security-focused filtering or escaping of user input.
Audit Metadata