drive-dogfood
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONNO_CODE
Full Analysis
- [COMMAND_EXECUTION]: The skill executes 'server-side probes' including database queries, file reads, and API calls to verify application state. It also performs setup tasks and may start the application if it is offline. These capabilities are intended for testing but allow significant access to system state.
- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it parses and executes actions directly from user-provided HTML guide files. 1. Ingestion points: External dogfood guide files (HTML). 2. Boundary markers: The skill does not define delimiters or provide instructions to the agent to ignore potentially malicious embedded commands within the guides. 3. Capability inventory: Browser automation (Chromium/Playwright), local file writing for ledgers, and backend command execution for probes and setup. 4. Sanitization: No sanitization or validation of the guide file content is performed before execution.
- [NO_CODE]: The skill consists only of instructions in SKILL.md and does not include any accompanying scripts, binaries, or configuration files.
Audit Metadata