kibana-agent-builder
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the official
elasticCLI to perform operations against the Kibana API. These commands are mapped to specific HTTP endpoints for managing agents and tools. This is standard vendor-provided functionality for administrative tasks. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from Elasticsearch indices using search tools. While this represents a potential surface for indirect prompt injection (if indexed data contains malicious instructions), the skill includes instructions for scoping search patterns narrowly and provides a structured environment with defined parameters to mitigate these risks.
- Ingestion points: External data enters through
index_searchandesqltools that query Elasticsearch indices. - Boundary markers: The skill explicitly instructs agents to "Always use tools to retrieve data. Never answer data questions from memory," establishing a clear boundary between model knowledge and retrieved data.
- Capability inventory: The skill has capabilities to read/write agent configurations and tools via the
elasticCLI. - Sanitization: The skill uses parameterized ES|QL queries (e.g.,
?lookback_days), which prevents direct injection into the query logic itself.
Audit Metadata