okf-bigquery

Pass

Audited by Gen Agent Trust Hub on Sep 3, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads and builds the okf-bigquery binary from the author's GitHub repository. It also depends on several external libraries, including official Google Cloud BigQuery and API packages, as well as a core OKF library from the same vendor.
  • [COMMAND_EXECUTION]: The skill operates as a command-line tool that interacts with Google Cloud BigQuery. It executes queries to retrieve table metadata, compute data profiles, and sample row data for documentation purposes. It also performs metadata updates when synchronizing OKF bundles back to BigQuery.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted data.
  • Ingestion points: The runIngest function in main.go reads OKF Markdown files from the local file system that represent table and field documentation.
  • Boundary markers: The parser looks for specific Markdown headers (e.g., # Columns) and table structures to isolate descriptive text from metadata.
  • Capability inventory: The skill has the capability to write the extracted descriptions back to a live BigQuery dataset metadata using the t.Update API.
  • Sanitization: The tool performs basic string trimming and structure validation on the ingested content before applying it to the database metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 3, 2026, 06:11 AM
Security Audit — agent-trust-hub — okf-bigquery