vs-item-onboarding

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data (JSON, CSV, JSONL, and MySQL tables) to perform schema inference and data ingestion.
  • Ingestion points: Untrusted data enters via user-provided local file paths and database connection configurations during the export and inference stages.
  • Boundary markers: The skill uses specific markers (<!-- vs-schema-confirm: BEGIN --> and <!-- vs-schema-confirm: END -->) to separate backend-inferred schema data from other prose, enabling clear user validation.
  • Capability inventory: The workflow has the capability to execute shell commands using the vendor's vs CLI and standard utilities like curl and jq for file processing and ingestion.
  • Sanitization: Risk is mitigated by mandatory human-in-the-loop checkpoints (Step 7: Schema Confirmation and Step 8: Behavior Type Confirmation) where the agent must wait for explicit user approval before proceeding with resource creation or data writing.
  • [SAFE]: The skill adheres to secure credential management by instructing users to set environment variables for database access instead of inputting secrets directly into the chat environment.
  • [SAFE]: Data uploads are conducted via presigned URLs provided by the platform's API, which restricts data transfers to authorized storage buckets without requiring the agent to manage long-lived credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 06:27 AM
Security Audit — agent-trust-hub — vs-item-onboarding