brainstorming
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell scripts (
start-server.sh,stop-server.sh) to launch and manage a local Node.js background process (server.cjs). This server facilitates the 'Visual Companion' feature by serving HTML mockups to a local port. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes untrusted data from the user's local environment.
- Ingestion points: The skill reads project files, documentation, and recent commits (
SKILL.md) and ingests browser interaction events recorded to a local state file (visual-companion.md). - Boundary markers: The instructions do not specify explicit delimiters or boundary markers for the external project data being ingested.
- Capability inventory: The skill can write design specifications to the filesystem (
SKILL.md) and execute local scripts via the bash tool to manage its web server. - Sanitization: While browser events are handled as structured JSON, there is no explicit sanitization for the content read from project files before it is processed by the agent.
Audit Metadata