agent-sql
Pass
Audited by Gen Agent Trust Hub on Jul 8, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection risk exists as the skill processes data from external databases which could contain adversarial instructions targeting the agent.
- Ingestion points: Data is brought into the agent context via the run, query run, and query sample commands in SKILL.md.
- Boundary markers: The skill outputs data in structured JSONL, JSON, and YAML formats, which provides some structural separation but lacks explicit instructions for the agent to ignore embedded commands within the data values.
- Capability inventory: The skill allows execution of agent-sql via Bash, along with Read, Grep, and Glob file system tools.
- Sanitization: No specific sanitization or filtering of database content is performed before it is presented to the agent.
Audit Metadata