ontology-builder-assistant

Warn

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions in SKILL.md require the agent to execute a local Python script located at 'scripts/owl_to_graphrag_schema.py'. This script is missing from the provided files, making its behavior and safety unverifiable during analysis.- [EXTERNAL_DOWNLOADS]: The skill mentions external dependencies 'rdflib' and 'neo4j-graphrag' as required for the execution of the conversion script. While these are common libraries, their usage is tied to the unverified local script execution.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its handling of external data.
  • Ingestion points: SKILL.md (Section 1) extracts 'representative data' from CSV, JSON, XML, YAML, and unstructured documents.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present.
  • Capability inventory: The skill performs file writes and executes a Python script ('scripts/owl_to_graphrag_schema.py') on the processed content.
  • Sanitization: No validation or escaping of the user-provided 'representative data' is specified.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 14, 2026, 12:40 AM
Security Audit — agent-trust-hub — ontology-builder-assistant