The Agent Skills Directory

[Safe Data Parsing]: The scripts/validate_catalog.py utility utilizes a custom YAML loader (UniqueKeyLoader) that inherits from yaml.SafeLoader. This ensures that data catalog frontmatter is parsed without the risk of executing arbitrary code or complex YAML tags, effectively mitigating potential deserialization concerns.
[File Access Controls]: Path traversal protections are implemented in the _resolve_input_path function. The script explicitly checks for and rejects path segments such as '..' and ensures that all file operations are restricted to permitted directory roots, preventing unauthorized file system access.
[Input Sanitization and Validation]: The skill uses the jsonschema library to validate catalog metadata against a strict schema (assets/ds-catalog-v1.schema.json). It also includes a _assert_json_compatible check that prevents the inclusion of non-standard or non-finite numeric values that could cause issues in downstream systems.
[Credential Management Policy]: The workflow contains explicit constraints that prevent the recording of embedded credentials in the catalog files. It directs the agent to use credential-free paths or connection references, adhering to security best practices for secret management.

data-catalog