skill-support

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Provides instructions to download and execute an installation script directly from the vendor's GitHub repository (skillrecordings/support) via a shell pipe.\n- [COMMAND_EXECUTION]: Enables the agent to execute a wide variety of commands via the skill binary, including system diagnostics (skill doctor), database connectivity checks (skill db-status), and event inspection (skill inngest event).\n- [DATA_EXFILTRATION]: Documents the handling of sensitive configuration and credentials, such as auth tokens and encrypted environment files stored in the user's home directory (~/.config/skill/) and the project root (.env.encrypted).\n- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external customer support sources, creating a surface for indirect prompt injection.\n
  • Ingestion points: Data is retrieved via skill front conversation, skill front message, and skill linear issue commands.\n
  • Boundary markers: No explicit delimiters or instructions to ignore instructions in the data are present in the command templates.\n
  • Capability inventory: The CLI possesses capabilities for network access (API requests to Front, Linear, and Vercel) and file system writes (skill dataset build).\n
  • Sanitization: No sanitization of external message content is documented in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 10:32 PM