cf-access-wall

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes an external JSON specification file to define security policies, which represents an indirect injection surface.
  • Ingestion points: scripts/setup-access.mjs reads a configuration file from a user-provided file path to determine Cloudflare Access settings.
  • Boundary markers: The script does not utilize explicit boundary markers or isolation logic for the data extracted from the JSON specification.
  • Capability inventory: The skill can modify Cloudflare Access policies, add users to identity allow-lists, and manage service tokens via API calls.
  • Sanitization: No explicit validation of the JSON content against the provided schema is performed within the script before it interacts with the Cloudflare API.
  • [COMMAND_EXECUTION]: The skill provides Node.js scripts that automate sensitive administrative tasks through browser automation and API interactions.
  • scripts/gh-oauth-app.mjs uses browser automation to manage GitHub settings, creating applications and extracting secrets to a local vault.
  • scripts/setup-access.mjs interacts with the Cloudflare management API to provision Zero Trust resources based on the provided spec.
  • These actions are intended for security infrastructure management and include identity verification steps to prevent misconfiguration.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 02:12 AM