grafana
Fail
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The Python code block for the Grafana API contains hardcoded credentials (
auth=('admin', 'admin')) targeting a local host. While likely intended as placeholders, using default administrative credentials in automation code poses a risk of unauthorized access if used in production environments. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process dashboard definitions, panel metadata, and metric queries (PromQL). This creates a vulnerability surface where malicious instructions could be embedded in external data (e.g., panel titles, legend formats, or query results) and processed by the agent.
- Ingestion points: Dashboard JSON structures, template variables, and PromQL expressions defined in
SKILL.md. - Boundary markers: None provided to distinguish between system instructions and data-embedded content.
- Capability inventory: The skill can interact with the Grafana API to update dashboards via the
grafana-apiPython client as shown inSKILL.md. - Sanitization: There are no instructions for sanitizing or validating external dashboard configurations or query strings before processing.
Recommendations
- AI detected serious security threats
Audit Metadata