We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/unionai-oss/union-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
deployment_v2.md•1.61 KiB
# Union v2 MCP deployment
This file contains instructions for deploying the Union v2 MCP server.
## Setup
Clone the repo:
```bash
git clone https://github.com/unionai-oss/union-mcp
```
Install `uv` and the `union-mcp` package:
```bash
pip install uv
```
## Running a local server
```bash
DISABLE_AUTH=1 uv run --with '.[v2]' mcp run examples/v2/server.py --transport sse
```
## Deploy app to Union
Create a config file:
```bash
uv run --with union union create login --host <host>
```
Then export environment variables:
```bash
export FLYTE_ENDPOINT=dns:///demo.hosted.unionai.cloud
export FLYTE_ORG=demo # use custom org name
export FLYTE_PROJECT=union-mcp # use custom project name
export FLYTE_DOMAIN=development # use custom domain name
export APP_NAME=union-mcp-v2 # use custom app name
export APP_SUBDOMAIN=mcp-v2 # use custom app subdomain
export APP_PORT=8000 # use custom app port
export APP_NAME=union-mcp-v2-test # use custom app name
export APP_SUBDOMAIN=union-mcp-v2-test # use custom app subdomain
```
Create a Flyte v2 config file:
```bash
uv run --with '.[v2]' flyte create config --endpoint $FLYTE_ENDPOINT --project $FLYTE_PROJECT --domain $FLYTE_DOMAIN
```
Register dummy workflows for testing (optional):
```bash
uv run --with '.[v2]' flyte deploy --project $FLYTE_PROJECT --domain $FLYTE_DOMAIN examples/v2/workflows.py env
```
Register the build and run tasks:
```bash
uv run --with '.[v2]' flyte deploy --project $FLYTE_PROJECT --domain $FLYTE_DOMAIN tasks_v2.py
```
Then deploy the app:
```bash
uv run --with '.[v2]' flyte deploy --project $FLYTE_PROJECT --domain $FLYTE_DOMAIN app_v2.py app
```