agent-auth-cli
Pass
Audited by Gen Agent Trust Hub on May 4, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill ingests untrusted data from external Agent Auth providers, such as discovery documents and capability schemas, which can lead to indirect prompt injection. 1. Ingestion points: External provider configurations fetched via
auth-agent discoverand JSON schemas retrieved viaauth-agent describe. 2. Boundary markers: The instructions lack explicit delimiters or instructions for the agent to ignore potentially malicious content within these schemas. 3. Capability inventory: Theauth-agent executecommand allows the agent to perform actions based on the schemas provided by the external source. 4. Sanitization: No mechanisms are specified to sanitize or validate the external provider data before it is used to construct CLI commands. - [COMMAND_EXECUTION]: All primary functions of the skill are performed through the
auth-agentCLI, which handles connection lifecycles and manages local storage for keys and configuration in the~/.agent-auth/directory. - [EXTERNAL_DOWNLOADS]: The skill directs the agent to use the
@auth/agent-cliNode.js package for its core functionality.
Audit Metadata