The Agent Skills Directory

[SAFE]: The skill is designed for administrative configuration of the Sumsub platform. It handles sensitive credentials (SUMSUB_SECRET_KEY) using standard environment variables and explicitly instructs the agent not to echo secrets back to the user.
[SAFE]: A sandbox-only policy is enforced in the authentication script (sumsub_curl.sh), which checks for the sbx: prefix on tokens to prevent accidental production impact during development or testing.
[SAFE]: The skill implements a modular architecture, delegating authentication and permission checks to dedicated helper scripts and external skill references, ensuring a clear separation of concerns.
[SAFE]: All shell scripts were reviewed for command injection or remote execution vulnerabilities; none were found. The scripts are focused wrappers for curl operations targeting the official Sumsub API domain (api.sumsub.com).
[SAFE]: The translation of user intent into SumScript conditionEl expressions is governed by a detailed schema reference (kyt-rule-schema.md), which helps ensure that generated rules are syntactically correct and use only supported fields.

sumsub-create-kyt-rules