goose-integration
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill documents the installation of the Goose CLI (
block-goose-cli) and provides numerous examples of executing thegoosecommand to start sessions, run recipes, and manage configurations. - [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by design, as it functions as a bridge to a secondary agent.
- Ingestion points: User-provided task descriptions are piped into the
gooseCLI tool (documented inSKILL.md). - Boundary markers: The command examples in the documentation do not include delimiters or instructions to ignore embedded malicious prompts in the task data.
- Capability inventory: The
gooseCLI is capable of file access, persistent task scheduling, and interacting with thousands of MCP servers, providing a significant capability surface for an injected instruction to exploit. - Sanitization: No sanitization or escaping of the user-provided strings is implemented before they are passed to the external agent session.
Audit Metadata