test-connector

Pass

Audited by Gen Agent Trust Hub on Oct 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes several shell commands to manage the development lifecycle, including environment creation with uv, state management via the Fivetran CLI (fivetran reset), and execution of a local runner script (run_connector.py).
  • [EXTERNAL_DOWNLOADS]: Installs the fivetran_connector_sdk and py-spy packages from a public registry at runtime. It also installs dependencies defined in a user-provided requirements.txt file.
  • [DYNAMIC_EXECUTION]: Uses python -c to generate and execute local Python scripts at runtime for querying the warehouse.db DuckDB file and reporting record counts to the user.
  • [INDIRECT_PROMPT_INJECTION]: This skill presents an attack surface by ingesting external data that could influence agent behavior.
  • Ingestion points: Reads sdk-reference.md, requirements.txt, and sync results from the local DuckDB database into the agent's context.
  • Boundary markers: Missing. No delimiters or explicit warnings are used to isolate untrusted data from instructions.
  • Capability inventory: Subprocess execution (uv, python, fivetran CLI) and file system access.
  • Sanitization: None observed for data interpolated from records or local configuration files.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 27, 2026, 04:57 AM