brainstorming

Pass

Audited by Gen Agent Trust Hub on Jun 30, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill's workflow involves exploring the project context by reading files, documentation, and commit history. This ingestion of untrusted data from the local workspace represents a surface for indirect prompt injection. * Ingestion points: SKILL.md (Project context discovery), scripts/server.cjs (User event processing). * Boundary markers: No specific delimiters are defined to isolate project data from instructions. * Capability inventory: File system write access, execution of planning skills, and local server management. * Sanitization: No explicit sanitization or validation of project-sourced data is described.
  • [EXTERNAL_DOWNLOADS]: Fetches a branding image from the vendor's domain primeradiant.com to render the visual companion's interface.
  • [COMMAND_EXECUTION]: Deploys a local Node.js server and uses shell scripts for session management, lifecycle control, and automated platform-specific browser launching.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 30, 2026, 05:32 PM
Security Audit — agent-trust-hub — brainstorming