letterly-update-subscription

Warn

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script subscription_manager.py uses Playwright to perform automated UI interactions on financial platforms. It specifically targets Stripe Checkout and Stripe Billing portals to programmatically click buttons like 'Start trial' and 'Cancel plan'. Automated interaction with payment gateways carries a risk of unintended financial transactions or subscription states if the UI changes or the automation fails.
  • [DATA_EXFILTRATION]: The skill accesses the shared browser profile via the get_shared_context_path() utility. This directory contains highly sensitive information, including active login sessions (cookies), local storage, and potentially saved credentials. Accessing this profile allows the skill to operate as the authenticated user across multiple web services, which is a high-privilege capability.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect injection because it navigates to external websites and relies on parsing their content to decide which actions to take. Maliciously crafted content on these sites could potentially mislead the element selection logic.
  • Ingestion points: Web page content from web.letterly.app and stripe.com (found in subscription_manager.py).
  • Boundary markers: None detected.
  • Capability inventory: Full browser automation via Playwright, including navigation and clicking (found in subscription_manager.py).
  • Sanitization: None; the script relies on specific string matches (text=, data-testid) to find elements, which can be manipulated by site content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 28, 2026, 10:16 PM
Security Audit — agent-trust-hub — letterly-update-subscription