data-context-extractor

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill implements a meta-generator pattern that is susceptible to indirect prompt injection (Category 8).
  • Ingestion points: Database schema metadata (table names, column names) retrieved via discovery tools and analyst responses to business logic questions in Phase 2.
  • Boundary markers: The provided templates (references/skill-template.md and references/domain-template.md) do not include delimiters or protective instructions to prevent the agent from obeying malicious commands embedded in the ingested schema data.
  • Capability inventory: The skill creates a directory structure, writes markdown and SQL files, and executes a Python packaging script.
  • Sanitization: There is no evidence of escaping or validation of external strings before they are interpolated into the generated skill files.
  • [COMMAND_EXECUTION]: The skill includes a utility script (scripts/package_data_skill.py) intended for local execution.
  • The script uses the zipfile and pathlib libraries to bundle the generated skill into a distributable archive.
  • It performs directory traversal and file read/write operations within the local environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 12:22 PM
Security Audit — agent-trust-hub — data-context-extractor