ei-build-custom-ai-labeling-blocks
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a developer guide for building Docker-based transformation blocks within the Edge Impulse platform.
- [CREDENTIALS_UNSAFE]: The instructions explicitly advise against printing or logging API keys and secrets, recommending the use of environment variables and platform-native secret management instead.
- [EXTERNAL_DOWNLOADS]: All external references and repository links point to the official Edge Impulse GitHub organization or well-known, trusted services such as OpenAI, Hugging Face, and Beam.cloud.
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of sample data through external models or LLMs. To mitigate risks from untrusted data, the skill implements a mandatory 'preview mode' (via the
--propose-actionsflag), which ensures that changes are staged and reviewed by a human in the Edge Impulse Studio before being applied to the dataset. - [COMMAND_EXECUTION]: The skill includes standard instructions for local testing using Docker and the Edge Impulse CLI, which are consistent with the intended developer workflow.
Audit Metadata