cluster-update-advisor

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for Indirect Prompt Injection.\n
  • Ingestion points: The skill ingests 'Cluster Readiness Data' (JSON) directly from the proposal request context as described in SKILL.md.\n
  • Boundary markers: It uses markdown code blocks (```json) to delimit the data, which provides some structural separation but does not prevent instructions embedded within the JSON values from influencing the agent's reasoning.\n
  • Capability inventory: The skill can trigger secondary operations like querying prometheus for metrics, searching jira for bugs, and calling the product-lifecycle API, all of which could be influenced by parameters derived from the injected data.\n
  • Sanitization: There is no mention of sanitizing, escaping, or validating the JSON content to ensure it contains only telemetry data and not natural language instructions intended to bypass constraints.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 05:40 PM
Security Audit — agent-trust-hub — cluster-update-advisor