brainstorm
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill accepts a
[topic]argument which is interpolated into the agent's context. This represents an ingestion point for untrusted data that could contain hidden instructions targeting the agent's logic or the file-writing feature. - Ingestion points: The
[topic]argument defined in theArgumentssection ofSKILL.md. - Boundary markers: None identified; user input is processed as the primary subject of the brainstorming session without explicit delimitation.
- Capability inventory: The skill includes file-writing capabilities via the
--export <path>argument. - Sanitization: No explicit sanitization, validation, or path-restriction logic is defined for the
topiccontent or the targetexportpath. - [COMMAND_EXECUTION]: The skill defines a command-like interface (
/brainstorm) with parameters that include local file system interactions. The--export <path>argument allows the agent to write the results of the brainstorming session to a user-specified file path, which could potentially be used to overwrite existing files if the underlying platform does not enforce strict path sandboxing.
Audit Metadata