superwall-editor
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local bash script,
scripts/sw-editor.sh, to interact with the Superwall API. It uses standard utilities likecurlfor network requests andjqfor JSON processing, all of which are scoped to the intended functionality of paywall management. - [EXTERNAL_DOWNLOADS]: The CLI communicates with
superwall-mcp.superwall.com. This is the official relay domain for the Superwall editor and is considered a legitimate vendor resource for the 'superwall' author. - [CREDENTIALS_UNSAFE]: The skill manages
SUPERWALL_API_KEYthrough environment variables or local.envfiles. It implements local state protection by applyingchmod 600to thestate.jsonfile, ensuring that short-lived session tokens are only accessible to the current user, which is a standard security practice. - [INDIRECT_PROMPT_INJECTION]: The skill dynamically discovers available tools from a running browser session at runtime. While this introduces a dependency on the integrity of the remote browser session's tool definitions, the surface is inherent to the live-editing workflow and is protected by the authenticated relay system.
Audit Metadata