databricks-common-errors
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires access to the
Bashtool with a scope restricted todatabricks:*commands. It provides templates for managing cluster states, job runs, and updating workspace permissions via the Databricks CLI. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is designed to ingest and analyze data from potentially untrusted external sources. An attacker could craft specific error messages within cluster logs or job outputs to influence the agent's diagnostic logic.
- Ingestion points:
SKILL.md(Step 1: check error message in job run output, cluster logs, or notebook cells). - Boundary markers: None are provided to help the agent distinguish between legitimate log data and malicious instructions embedded within the logs.
- Capability inventory: The skill has the capability to modify cluster configurations and update access control lists (ACLs) using the Databricks SDK and CLI.
- Sanitization: No sanitization or validation logic is specified for the data ingested from logs and notebook cells.
Audit Metadata