okf-data-catalog
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using a local Python tool for validation and visualization tasks.
- Evidence:
python tools/okf_tool.py lint <catalog>andpython tools/okf_tool.py visualize <catalog> -o <artifact-dir>/viz.htmlare explicitly called out as quality gates inSKILL.md. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) as it ingests and processes untrusted data from external data catalogs to generate markdown documentation.
- Ingestion points: Data catalog metadata including datasets, tables, column descriptions, metrics, and join paths defined in
SKILL.md. - Boundary markers: Absent; there are no instructions to the agent to use delimiters or ignore potentially malicious instructions embedded within the imported catalog metadata (e.g., table comments or business definitions).
- Capability inventory: The skill performs file system writes to create the OKF bundle directory structure and executes local shell commands via the
okf_tool.pyscript. - Sanitization: Absent; the instructions do not require the agent to filter, escape, or validate content sourced from external documentation, query history, or dbt models.
Audit Metadata