[Feature] Add `get_app_logs` tool for Databricks Apps

[epistoteles]

When developing Databricks apps there is one big drawback: There is no MCP tool for accessing the App logs.

This means my AI agent thinks that the deployment was successful while actually there was a post-startup crash which is only visible from the logs (the app deploy command shows "successful" but the app actually immediately had a startup error).

While many of these crashes can be caught with local testing, some only occur on the Databricks compute (e.g. lacking env variables).

Could you add a get_app_logs tool or something similar to the ai-dev-kit which allows my agent to check the logs post-deployment?

Thanks!

[CheeYuTan]

This feature already exists! The get_app MCP tool supports an include_logs parameter that retrieves deployment logs:

# Get app details with logs
get_app("my-app", include_logs=True)
# Returns: {"name": "my-app", ..., "logs": "..."}

# Get logs for a specific deployment
get_app("my-app", include_logs=True, deployment_id="dep-123")

See databricks-mcp-server/databricks_mcp_server/tools/apps.py lines 132-177 — the get_app tool accepts include_logs: bool = False and deployment_id: Optional[str] = None parameters.

This should address the use case of checking post-deployment crash logs. Could this issue be closed?