What's the best Vibe-Code set up for foundry/repo dev?

I was given an opportunity to try out Foundry developer account a few days ago and the platform’s been pretty powerful and amazing.

I discovered AI FDE so I was making Workshop module with that for a bit, and then I realized, I should probably leverage OSDK via developer console. But tbh, the Continue extension has been quite slow in its outputs and task speed, overall clunky feeling.

Now I am thinking of just cloning it locally, and using Palantir MCP with my own Claude Code or Cursor, but would be eating up my weekly token quota…

What are y’all’s set up for this??

There are many options. You can also use the foundry provided language models in your local Claude code to route LLM calls through foundry.

Palantir-MCP is getting better every week - I got great feedback from my org. Make sure to either explicitly block tools or activate tool search otherwise your context is full with the Palantir-MCP tools.

I have recently got into AI FDE and it’s been great experience so far ! It’s handling most of the in-built platform features with great accuracy, equipped with its 100+ dedicated tools.

However today we can’t extend/augment its capacity with custom tools for integration with other MCP/tools.

Would it work better with Claude code/ Codex? Do they have access to the same harness as AI FDE via Palantir-MCP ?

For me, i would be interested to plug-in few ideas for ML exploration:

  • huggingface/ml-intern: :hugs:

  • Unsloth

  • Confluence

  • Jira

@jmoisan