valet
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes the Homebrew installation script from the official GitHub repository, which is a well-known and trusted service. Evidence: /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)\" in SKILL.md.
- [EXTERNAL_DOWNLOADS]: Instals the valet-cli tool from the author's official Homebrew tap. Evidence: brew install valetdotdev/tap/valet-cli in SKILL.md.
- [COMMAND_EXECUTION]: Employs the bash tool to execute various system and platform management commands, including resource creation, deployment, and testing via valet exec. It also accesses local session logs in ~/.claude/projects/ to enable the session capture feature.
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection due to its ingestion of untrusted external data. \n- Ingestion points: Processes data from external URLs via WebFetch and receives JSON payloads from webhook channels. \n- Boundary markers: Instructs the agent to parse payloads inline and recommends a "Webhook Scope Rule" to restrict actions to specific identifiers provided in the data. \n- Capability inventory: Can manage platform resources and execute commands with injected secrets for verification. \n- Sanitization: Relies on structural delimiters and explicit scoping instructions rather than programmatic input sanitization.
Audit Metadata