detecting-database-deadlocks
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to parse database error logs, which represent an untrusted data ingestion surface because log entries can include user-controllable strings such as SQL comments, table names, or failed query text.
- Ingestion points:
SKILL.md(Step 2) directs the agent to parse recent deadlock events from database logs and examineSHOW ENGINE INNODB STATUSoutput. - Boundary markers: The instructions do not provide delimiters or specific warnings to ignore instructions that might be embedded within the log data.
- Capability inventory: The skill is granted access to database CLI tools (
psql,mysql,mongosh) and file system tools (Read,Write,Grep,Glob). - Sanitization: There are no specified sanitization or escaping procedures for the external log content before it is processed by the agent.
Audit Metadata