sumsub-create-cross-check-preset

Pass

Audited by Gen Agent Trust Hub on Sep 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses several Bash scripts (get_cross_check_preset.sh, patch_cross_check_preset.sh, post_cross_check_preset.sh) to perform network operations and calculate cryptographic signatures using openssl. These scripts are used for legitimate communication with the Sumsub API and are confined to the intended functionality.- [DATA_EXFILTRATION]: The skill performs network requests to api.sumsub.com via curl. It handles sensitive credentials (SUMSUB_APP_TOKEN, SUMSUB_SECRET_KEY) provided via environment variables. This behavior is expected for a vendor-authored tool interacting with its own official API. The scripts include security checks to ensure that only sandbox tokens (prefixed with sbx:) are accepted by default, reducing the risk of accidental production data exposure.- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided configuration data (JSON/YAML) to construct API payloads.
  • Ingestion points: User input provided via stdin to the build_cross_check_preset.py script.
  • Boundary markers: The instructions explicitly require the agent to confirm the user's intent and warn them about overriding default settings before proceeding.
  • Capability inventory: The skill possesses network access (curl) and command execution capabilities (Bash, Python).
  • Sanitization: The build_cross_check_preset.py script performs thorough validation of the input schema, including data types, string lengths, and enum value checks, which effectively mitigates the risk of schema confusion or malicious payload construction.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 4, 2026, 09:08 PM
Security Audit — agent-trust-hub — sumsub-create-cross-check-preset