pythia-tune
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill handles SQL statements sourced from code or user queries, creating a potential surface for indirect injection. (1) Ingestion points: SQL strings from
pythia src,pythia grep, and user input inSKILL.mdandreference/tuning-checklist.md. (2) Boundary markers: No delimiters or 'ignore' instructions are present. (3) Capability inventory: Usespythia explain,pythia sql(execution), andpythia deploy. (4) Sanitization: Not specified in instructions. - [COMMAND_EXECUTION]: The skill instructs the agent to execute 'pythia' CLI commands and shell utilities like
time. These are necessary for query tuning and performance measurement. - [SAFE]: The skill implements operational best practices, including checking for database licenses (AWR/ASH) and respecting schema ownership for administrative tasks like gathering statistics.
Audit Metadata