ei-build-custom-synthetic-data-blocks
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill scaffolds a tool designed to take user-supplied prompts as input to drive generative AI models. This creates a data ingestion surface where untrusted input influences the samples and metadata generated and subsequently uploaded to the project.
- [COMMAND_EXECUTION]: The provided documentation includes instructions for running local development commands, such as using the
edge-impulse-blocksCLI for initialization and deployment, anddocker runfor testing containerized block execution. - [EXTERNAL_DOWNLOADS]: The
Dockerfiletemplate includes instructions to download and install Python packages from public registries. The skill also references official example repositories on GitHub within theedgeimpulseorganization. - [CREDENTIALS_UNSAFE]: The skill describes the management of sensitive credentials like project API keys and provider tokens. It adheres to security best practices by instructing users to utilize environment variables and the platform's
secretparameter type, while explicitly warning against logging these sensitive values.
Audit Metadata