postgres
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a Python script ('scripts/query.py') to interface with PostgreSQL databases for tasks such as listing tables, viewing schemas, and performing data analysis. This is the primary intended functionality of the skill.
- [EXTERNAL_DOWNLOADS]: The skill's setup instructions specify the installation of 'psycopg2-binary' via pip, which is a widely recognized and official library for PostgreSQL database connectivity in Python.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it translates user-provided natural language intent into SQL query strings for execution. While the documentation describes safety features, the ingestion of untrusted data into query parameters is a known attack vector.
- Ingestion points: User-provided questions and query requests entered into the agent context (SKILL.md).
- Boundary markers: The instructions do not define specific delimiters or escaping mechanisms to separate untrusted user input from the query generation logic.
- Capability inventory: The skill can execute shell commands ('python3 scripts/query.py') and read configuration files containing database credentials.
- Sanitization: The skill claims to use logic-based query validation (restricting statements to SELECT, SHOW, EXPLAIN, and WITH) and the PostgreSQL 'readonly=True' session mode to mitigate risks.
Audit Metadata