developing-genkit-python

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the installation script for the uv package manager from the official Astral domain (astral.sh).
  • [REMOTE_CODE_EXECUTION]: Includes instructions for installing uv using a remote script via curl | sh. The download originates from a well-known service provider.
  • [PERSISTENCE]: Describes a method for persisting the GEMINI_API_KEY environment variable by appending an export command to the user's shell profile (~/.zshrc). This is a standard configuration practice for developer tools.
  • [INDIRECT_PROMPT_INJECTION]: The skill documents the creation of AI agents that process user-supplied content and have the ability to interact with tools such as the filesystem and external APIs.
  • Ingestion points: User messages ingested through chat.send() or send_stream() in multiple reference files (e.g., agents.md, examples.md).
  • Boundary markers: Recommends using the ToolApproval middleware to define explicit boundaries for automated tool execution.
  • Capability inventory: Documentation includes examples for filesystem operations (read_file, write_file, edit_file), artifact generation, and custom business logic tools.
  • Sanitization: Provides instructions and code patterns for implementing a human-in-the-loop workflow using ToolApproval and allowed_tools lists to ensure sensitive operations are reviewed or restricted.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:48 PM
Security Audit — agent-trust-hub — developing-genkit-python