brainstorm-experiments-new
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides an interface for processing untrusted user data, which creates a potential surface for indirect prompt injection. 1. Ingestion points: The skill reads user input via the $ARGUMENTS variable and is instructed to read external files such as market research or landing page mockups mentioned in the SKILL.md context section. 2. Boundary markers: There are no explicit delimiters or specific instructions (e.g., ignore embedded commands) to separate ingested content from the system instructions. 3. Capability inventory: The skill is limited to text analysis and generation; it lacks network access, shell command execution, or privileged file system access. 4. Sanitization: No validation or sanitization logic is implemented for the processed external content.
- [NO_CODE]: No executable scripts, binaries, or complex automation were detected in the skill package, limiting the potential for malicious execution.
Audit Metadata