okf-enrich
Pass
Audited by Gen Agent Trust Hub on Aug 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run CLI tools such as
okf-vizand various database-specific connectors (e.g.,okf-sqlite,okf-postgresql,okf-bigquery). These tools are used for calculating documentation coverage and syncing enriched metadata back to origin data sources. These are recognized as legitimate tools within the vendor's ecosystem. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted data from OKF bundle files (including column names, data profiles, and sample rows) to generate human-readable descriptions.
- Ingestion points: The agent reads data from concept markdown files, specifically targeting the
# Columns,## Data Profile, and## Samplesections. - Boundary markers: The instructions do not prescribe the use of specific delimiters or negative constraints to prevent the agent from following instructions potentially embedded in the ingested data.
- Capability inventory: The agent is instructed to write enriched documentation, tags, and verification metadata back into the frontmatter of local concept files.
- Sanitization: No explicit sanitization, filtering, or validation of the ingested bundle data is required by the instructions.
Audit Metadata