odata-vocabularies-sync

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches OData vocabulary definitions from a user-provided URL during Step 1a. It specifically targets JSON files from domains like sap.github.io.
  • [COMMAND_EXECUTION]: The skill executes local shell commands via pnpm in Step 2 (update:vocabularies) and Step 4 (test -- -u). These commands run the project's internal tooling and test suites.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8). It ingests untrusted data from an external JSON URL and uses extracted fields (namespace, alias) to modify multiple TypeScript files (index.ts, loader.ts, update.ts).
  • Ingestion points: User-provided JSON URL processed in Step 1a.
  • Boundary markers: Includes logic to validate namespaces (e.g., must start with Org.OData. or com.sap.vocabularies.) and identifies reserved keywords.
  • Capability inventory: Capability to write to the local file system (modifying .ts, .md files) and execute shell scripts via pnpm across multiple steps.
  • Sanitization: Relies on basic string matching for namespaces and checks for JavaScript reserved keywords, but does not perform full escaping of all JSON-derived content before interpolation into code template strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 12:37 PM
Security Audit — agent-trust-hub — odata-vocabularies-sync