brainstorming
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input during brainstorming sessions and interpolates it into a structured YAML template without sanitization. The generated documentation is intended to be consumed by downstream agents such as project managers and architects, creating an attack surface where malicious instructions embedded in 'ideas' could influence subsequent agent behavior.
- Ingestion points: User input collected during Step 1 (Session Setup) and Step 3 (Execute Techniques Interactively) in SKILL.md.
- Boundary markers: Absent; the template in resources/brainstorming-output-tmpl.yaml does not use delimiters or explicit instructions to prevent downstream agents from interpreting user-provided content as commands.
- Capability inventory: The skill triggers file generation through create-doc and integrates with strategic skills like pm and architect.
- Sanitization: No evidence of input validation, escaping, or filtering for user-provided strings before they are rendered into the final output document.
Audit Metadata