Skip to content

Add Databricks service support#3

Open
weishi-imbue wants to merge 2 commits intomainfrom
wz/databricks
Open

Add Databricks service support#3
weishi-imbue wants to merge 2 commits intomainfrom
wz/databricks

Conversation

@weishi-imbue
Copy link
Contributor

@weishi-imbue weishi-imbue commented Jan 30, 2026

Summary

  • Add Databricks service support with cookie-based authentication
  • Capture browser session cookies (DBAUTH) and CSRF tokens for API access
  • Handle dynamic workspace URLs (*.cloud.databricks.com) via dynamic service creation
  • Update services command to include databricks in the list
  • Add comprehensive unit tests for credentials and URL matching

Test plan

  1. Manually tested with latchkey curl 'https://dbc-b28fe787-b68d.cloud.databricks.com/ajax-api/2.0/mlflow/experiments/list'
  2. unit tests, linting

🤖 Generated with Claude Code

weishi-imbue and others added 2 commits January 29, 2026 15:58
- Document minimum Node.js (22.9.0) and npm (11.6.0) versions
- Add note for nvm users about global package reinstallation
- Add prepare script to run build before npm install
- Clean up package-lock.json peer dependency markers

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Implement cookie-based authentication for Databricks workspaces using
browser session cookies (DBAUTH) and CSRF tokens. Unlike other services
with fixed URLs, Databricks uses dynamic workspace URLs (*.cloud.databricks.com)
so it's handled via dynamic service creation rather than static registry.

- Add DatabricksApiCredentials class for cookie + CSRF token storage
- Add Databricks service with BrowserFollowupServiceSession for login flow
- Update registry to detect Databricks URLs and create service dynamically
- Update services command to include databricks in the list
- Add unit tests for credentials serialization and URL matching
- Document Databricks usage in README

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
@hynek-urban
Copy link
Collaborator

Thanks, @weishi-imbue !

I think we want the tool to work with public APIs. Mostly because the internal ones are typically not documented (and the skill definition asks agents to look up documentation online). Some services make this easy because their clients are built on top of public APIs. When that's not the case, our approach has been to use browser automation for to generate the API token through the service's web interface (like here). Do you think that would be a viable approach for Databricks?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants