The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The agent processes external data (queries and execution plans) which serves as an ingestion point for potential malicious instructions.\n- Ingestion points: Slow query analysis and execution plan analysis performed via Read and Bash tools as described in SKILL.md.\n- Boundary markers: The skill lacks explicit markers or delimiters to separate untrusted query data from system instructions in the prompt logic.\n- Capability inventory: The agent is granted Bash, Write, and Edit permissions, which provide a significant attack surface if an indirect injection is successful.\n- Sanitization: There is no evidence of sanitization, escaping, or filtering for the data being analyzed.

dba-tuner