The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from the Databricks workspace (e.g., job task keys and cluster metadata) which is used to modify local configuration files and provision environments.
Ingestion points: External data enters the context via databricks jobs get and databricks clusters list output as described in SKILL.md and references/examples.md.
Boundary markers: The instructions require using --output json and specific jq filters to maintain structural boundaries for ingested data.
Capability inventory: The agent has the capability to write to pyproject.toml, modify bundle YAML sources, and execute commands through the uv package manager.
Sanitization: All state-changing operations require explicit user approval following a displayed plan from a --dry-run execution.
[COMMAND_EXECUTION]: The skill uses the databricks CLI and uv package manager to execute environment setup, authentication, and diagnostic commands.
[EXTERNAL_DOWNLOADS]: Mentions the installation of the uv package manager and references official Databricks GitHub repositories for issue reporting. The skill documentation specifically advises against the use of silent or remote installers for security purposes and provides guidance on safely redacting reports.

databricks-setup-local