truto-unified-mappings
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides setup instructions to download the Truto CLI from vendor-controlled infrastructure at
cli.truto.one. It also automatically downloads a local ONNX model (all-MiniLM-L6-v2) to power local embeddings for documentation search. - [REMOTE_CODE_EXECUTION]: The documentation includes instructions for installing the CLI by piping remote scripts to a shell (e.g.,
curl -fsSL https://cli.truto.one/install.sh | bash). This is part of the vendor's standard installation process for the toolset provided by the skill author. - [COMMAND_EXECUTION]: The skill relies on the execution of various CLI commands to handle authentication, profile setup, and the build/validate/apply lifecycle for mapping files.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external API documentation provided via URLs or web searches to ground the mapping generation. This represents a potential surface for indirect prompt injection. The skill implements an evidence chain of security controls, including automated validation of generated JSONata against unified schemas, cross-integration leak detection, and restricted read-only proxy sampling (
GETrequests only). - [DYNAMIC_EXECUTION]: The core functionality involves an agentic loop that dynamically generates and evaluates JSONata mapping logic to transform proxy responses into unified model schemas. This behavior is the primary intended use case for the skill.
Audit Metadata