remember
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the
signetCLI tool to store data in a local daemon. Assignetis a resource belonging to the vendor (Signet-AI), this is considered standard functionality. Evidence includes the command:signet remember "<content>" -w <agent-name>. - [SAFE]: The skill performs health checks on a local daemon via
curl -s http://localhost:3850/health. This involves only local communication and is not associated with data exfiltration. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by capturing and storing user-provided content without sanitization. 1. Ingestion points: Untrusted data enters the agent context through the
<content>argument inSKILL.md. 2. Boundary markers: None are present to delimit user content or instruct the agent to ignore embedded instructions. 3. Capability inventory: The skill has the capability to execute shell commands via thesignetCLI across its implementation. 4. Sanitization: No sanitization or validation of the input content is performed before storage.
Audit Metadata