sumsub-create-poa-preset

Pass

Audited by Gen Agent Trust Hub on Oct 3, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local shell and Python scripts to build payloads and perform authenticated API calls. The operations are limited to data processing and HTTP requests via curl.\n- [DATA_EXFILTRATION]: Network communication is directed to api.sumsub.com. The skill requires SUMSUB_APP_TOKEN and SUMSUB_SECRET_KEY environment variables. It enforces sandbox-only usage by checking for the sbx: prefix and explicitly forbids echoing secrets, following secure credential handling practices.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined configurations for POA settings.\n
  • Ingestion points: User-provided specifications are ingested via stdin into build_poa_preset.py.\n
  • Boundary markers: The skill documentation defines a strict compact spec format (YAML/JSON) for data ingestion.\n
  • Capability inventory: Uses scripts/post_poa_preset.sh and scripts/patch_poa_preset.sh to perform network operations and scripts/build_poa_preset.py for local processing.\n
  • Sanitization: The Python builder script implements strict validation of enum values (e.g., PoaSubType, IdDocType) and ISO country code formats, reducing the risk of processing malicious instructions as configuration data.\n- [EXTERNAL_DOWNLOADS]: Mentions pyyaml as a dependency for YAML parsing, which is a trusted and well-known library.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 3, 2026, 02:03 PM