data-contracts
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill recommends installing
datacontract-cliandacryl-datahubvia pip. These are the official and standard tools required for the data contract and data cataloging tasks described in the skill. - [COMMAND_EXECUTION]: The skill provides various CLI commands for linting, testing, and publishing data contracts using
datacontract-cli,git, andcurl. These are standard operational commands for the intended workflow. - [CREDENTIALS_UNSAFE]: The skill demonstrates excellent security posture by explicitly warning against hardcoding credentials in configuration files and providing examples of how to safely use environment variables (e.g.,
DATACONTRACT_POSTGRES_PASSWORD,AWS_SECRET_ACCESS_KEY) for authentication. - [DATA_EXFILTRATION]: The skill includes instructions for publishing contract metadata to external endpoints such as DataHub and OpenMetadata using
curland thepublishcommand. These are documented as intended integration points for centralized data governance and use placeholder domains (e.g.,example.com). - [INDIRECT_PROMPT_INJECTION]: The skill involves processing external data sources and configuration files (YAML, SQL, data from S3/Kafka).
- Ingestion points: Reads
datacontract.yamlfiles and samples data from various database and streaming servers. - Boundary markers: None explicitly defined in the documentation, but standard CLI tool boundaries apply.
- Capability inventory: Shell execution (
datacontract,curl), network requests (publish), and file system writes (export). - Sanitization: Not explicitly described, as the skill provides usage instructions rather than the underlying tool implementation. The analysis finds this surface to be standard for the task and not indicative of malicious intent.
Audit Metadata