building-multi-connector-agent

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a development guide that provides boilerplate code for integrating Airbyte connectors into an AI agent. It contains no malicious logic, obfuscation, or unauthorized access attempts.- [CREDENTIALS_SAFE]: The skill correctly demonstrates using environment variables (os.getenv) and .env files for managing sensitive credentials like AIRBYTE_CLIENT_SECRET, which is a standard security best practice.- [EXTERNAL_DOWNLOADS]: The skill installs the official airbyte-agent-sdk via uv pip, which is expected and originates from the skill's author (airbytehq).- [PROMPT_INJECTION]: The skill describes an architecture for an agent that processes untrusted user input and interacts with external tools (Jira, Slack), which is an indirect prompt injection surface.
  • Ingestion points: User input is passed to agent.run() in the run loop.
  • Boundary markers: None are defined in the example system prompt.
  • Capability inventory: The agent has tools to execute actions on Jira and Slack via the Airbyte SDK.
  • Sanitization: None is performed in the provided template code.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 10:49 AM