claude-hud

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of established and widely-used Node.js packages (blessed, blessed-contrib, chalk, ws) from the official npm registry to support its terminal dashboard and communication features.- [INDIRECT_PROMPT_INJECTION]: The skill processes and displays external data, including tool inputs/outputs and token usage metrics, creating a potential surface for indirect prompt injection. * Ingestion points: Data enters the context via API response headers in context-monitor.js and WebSocket messages in server.js. * Boundary markers: Content is handled as structured JSON without explicit delimiters for embedded instructions. * Capability inventory: The skill facilitates terminal UI rendering and maintains a local WebSocket server. * Sanitization: The implementation uses string truncation for displayed tool data and provides explicit guidelines to avoid logging sensitive code, mitigating the risk of data exposure.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:07 AM
Security Audit — agent-trust-hub — claude-hud