triaging-live-sql-activity

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection because it processes and displays live SQL activity from CockroachDB cluster views. Instructions embedded within user-authored SQL queries could potentially influence the agent when it triages long-running work.\n
  • Ingestion points: Results from diagnostic queries such as SHOW CLUSTER STATEMENTS and SHOW CLUSTER SESSIONS referenced in SKILL.md and references/sql-queries.md.\n
  • Boundary markers: There are no explicit delimiters or instructions to the agent to treat the retrieved SQL query text as untrusted data.\n
  • Capability inventory: The skill provides the agent with the ability to perform administrative database operations, specifically CANCEL QUERY and CANCEL SESSION.\n
  • Sanitization: The skill employs a substring() filter in its SQL templates to truncate query text to 200 characters, which limits the available payload size for any potential injection.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 05:05 PM