promql-generator

Pass

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill incorporates untrusted user data—such as metric names, label keys, and thresholds—into the logic of generated PromQL queries, creating a surface for indirect prompt injection.\n
  • Ingestion points: Requirements and parameters gathered from user prompts and via the AskUserQuestion tool during the monitoring goal and parameter discovery phases (Stages 1-3).\n
  • Boundary markers: The workflow mandates a 'Query Plan' stage (Stage 4) where the agent must explain the query logic in plain English and obtain user confirmation before any code is generated.\n
  • Capability inventory: The skill generates PromQL queries for alerts, recording rules, and dashboards; it does not possess the capability to execute these queries directly on a server.\n
  • Sanitization: The skill includes an automated validation step (Stage 6) that invokes the devops-skills:promql-validator skill to check the generated output for syntax errors and architectural anti-patterns.\n- [NO_CODE]: The skill does not include any executable scripts, binaries, or active code components. It consists entirely of instructional markdown, YAML configuration examples, and PromQL patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 4, 2026, 08:54 AM