Problems integrating palantir-mcp

Questions
1

I am trying to integrate the palantir-mcp in a mcp-client community node as this supports only STDIO for authentication. I enter the credentials as shown in the image below. I tried installing the palantir-mcp globally in the docker image running n8n but that did not help as well.

image
image978×331 16.1 KB

What is the error message (if any)?

Problem in node ‘MCP Client‘
Failed to execute operation: The file or directory does not exist

Please share your workflow

For now I just have a trigger which is on click and it executes the get available tools from the mcp server

Information on your n8n setup

  • n8n version: 2.6.3
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main): own
  • Running n8n via (Docker, npm, n8n cloud, desktop app): Docker
  • Operating system:
2

Hi @10x_sanjeeth ,

This error (The file or directory does not exist) is extremely misleading. It sounds like n8n can’t find a file you uploaded, but usually, it just means the node cannot find the npx command itself.

Even if npx works when you SSH into the container, the n8n node runs in a restricted environment that doesn’t always have the same “PATH” variable (the list of folders where the system looks for tools).

** Quick FiX (Absolute Path)**

If you don’t want to rebuild your Docker image right now, you just need to tell the node exactly where npx lives.

  1. Find the path:
    Open your Coolify terminal (or SSH) and run this command inside your n8n container:
docker exec -it [your-n8n-container-name] which npx

(It will likely output something like /usr/local/bin/npx or /usr/bin/npx).
2. Update the Node:

  • Command: Paste that full path (e.g., /usr/local/bin/npx) instead of just npx.
  • Arguments: Keep your existing arguments (-y palantir-mcp ...).

(Note: If this still gives the “not found” error, use the absolute path: /usr/local/bin/palantir-mcp)

  • Arguments: Remove the -y part. Just keep --foundry-api-url ...
If my answer helped solve your question, would you mind marking it as the solution?
It’ll help others find it more easily—and I’d really appreciate it!

Thanks!

3

Hey @10x_sanjeeth,
Welcome to the N8n Community

From my perspective, You’re almost certainly hitting a “command not found inside the container / execution environment” issue, not something Palantir-MCP specific.

That “The file or directory does not exist” message usually shows up when n8n tries to spawn a process and can’t find the executable you typed (npx, or the MCP binary) on the PATH that the MCP Client node uses. This is super common in Docker because what works on your host doesn’t neccessarily exist (or isn’t on PATH) inside the container.

Try this method:-

Execute into the running n8n container and check whether npx actually exists:

docker exec -it sh
which npx
node -v
npx -v

If which npx returns something like /usr/local/bin/npx, then in the MCP Client node use the full path:

Command: /usr/local/bin/npx

Arguments: -y palantir-mcp --foundry-api-url https://10x.palantirfoundry.com

(Using full path fixes a lot of these spawn/path errors.)

If npx isn’t there:
Then the default n8n image you’re running doesn’t have what you need in the runtime environment, so the clean approach is a custom image where you preinstall the MCP server:

FROM n8nio/n8n:2.6.3
RUN npm i -g palantir-mcp

Then in your MCP Client node you can run it directly (again, preferably using full path from which palantir-mcp):

Command: palantir-mcp (or /path/to/palantir-mcp)

Arguments: --foundry-api-url https://10x.palantirfoundry.com

Your env vars approach is correct, just make sure they’re entered as KEY=value (one per line) in the node’s Environment section.

About EXECUTIONS_PROCESS=own
I don’t think this is an MCP limitation.

Can you share the output of which npx and which palantir-mcp inside your container, I can tell you exactly which command/path to plug into the node.

4

I tried the approach mentioned above and gave the complete path for the npx command but still I end up getting the same error. I also installed it globally and tried to run the command using both the full path and just palantir-mcp, still I get the same error. Have you had sucess implementing a STDIO based mcp client on the community mcp client node?