latchbio-integration
Pass
Audited by Gen Agent Trust Hub on May 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the
latchpackage from the Python Package Index (PyPI). This is the official SDK for the Latch platform described in the documentation. - [COMMAND_EXECUTION]: The documentation describes using CLI commands such as
latch registerandlatch executeto build Docker containers and manage workflows. These commands are essential to the primary function of the platform and are presented in a legitimate context. - [PROMPT_INJECTION]: The skill processes external bioinformatics data files (e.g., FASTQ, BAM, VCF) and metadata records from the Latch Registry. This creates a surface for indirect prompt injection where malicious content in these data sources could attempt to influence agent behavior. \n
- Ingestion points: The skill reads external data via
LatchFile,LatchDir, andTable.getmethods as described inreferences/data-management.md. \n - Boundary markers: No specific delimiters or instructions to ignore embedded commands within data files are provided. \n
- Capability inventory: The agent has the ability to build and deploy containers (
latch register) and execute workflows via the CLI. \n - Sanitization: There is no mention of sanitizing or validating the contents of input files before they are processed by the agent.
Audit Metadata