tangent
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of
subprocess.runintangent.pyandreadout.py. These calls are used to invoke the local companion script and the Superwall CLI tool to query the platform's backend and retrieve project metadata. - [EXTERNAL_DOWNLOADS]: Instructions in
SKILL.mdguide the agent or user to install thesuperwallCLI vianpm install -g superwall. This is a vendor-owned package required for the skill's functionality. The scripts also fetch paywall screenshots fromapi.superwall.comfor report generation. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically campaign names, audience descriptions, and paywall names from the Superwall dashboard. While the Python script uses
html.escapeto sanitize data for the generated HTML reports, the agent's task of digesting this data for the user presents a standard indirect injection surface (severity: LOW).
Audit Metadata