proxy-manager
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates interaction with the
webshareCLI tool and a local Python helper script to manage proxy provisioning, IP authorization, and account statistics. This usage is restricted to legitimate administrative tasks related to the user's proxy account. - [INDIRECT_PROMPT_INJECTION]: The skill processes output from the
webshareCLI, which represents a potential data ingestion surface. 1. Ingestion points: Outputs from CLI commands such asproxies list,plans list, andstatsare read into the agent context. 2. Boundary markers: The skill instructions explicitly direct the agent to confirm with the user before performing any mutations (e.g.,proxies refresh,config set). 3. Capability inventory: The skill usesBashto execute the vendor CLI, run a local Python script, and open browser URLs. 4. Sanitization: The skill recommends the use of the--jsonflag to ensure data is parsed as structured content rather than free-form text.
Audit Metadata