brainstorm
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it automatically reads and processes content from several local repository files (e.g., AGENTS.md, CLAUDE.md, design/gdd/game-concept.md, and reference docs) to inform its brainstorming process. If these files were to contain malicious instructions, they could influence the agent's logic or the instructions passed to sub-agents (creative-director, technical-director, etc.).\n
- Ingestion points: The workflow reads project instructions and game design documents from the local repository (referenced in SKILL.md and Phase 2 of references/claude-skill.md).\n
- Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potential embedded commands within the ingested files.\n
- Capability inventory: The skill has the ability to write files to the 'design/gdd/' directory and spawn additional agent tasks using the 'Task' tool.\n
- Sanitization: No sanitization or validation of the ingested file content is performed before it is used in prompts.
Audit Metadata