observability-setup

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill demonstrates an indirect prompt injection surface by consuming external configuration data from 'plans/manifests/observability-manifest.yaml' to drive its orchestration logic.
  • Ingestion points: The orchestrator reads artifact definitions, SQL queries, and metadata from the 'observability-manifest.yaml' file during the mandatory Phase 0.
  • Boundary markers: There are no explicit delimiters or boundary instructions provided to the agent to distinguish between its internal instructions and the potentially untrusted content of the manifest file.
  • Capability inventory: The skill possesses significant capabilities, including the ability to create monitors, retrieve schema information, and deploy dashboards/alerts using the Databricks SDK ('w.data_quality.create_monitor', 'w.schemas.get').
  • Sanitization: The implementation uses 'yaml.safe_load()' to prevent YAML-level exploitation, but it lacks secondary validation or sanitization of the logical content (e.g., table names or SQL queries) before they are passed to SDK methods.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 02:33 AM