CODING TOOLS

Connect a coding tool

Pick your tool, copy the prompt, and let your coding agent do the setup.

First, create a key in API keys and save it as YUGOS_API_KEY. The paid yugos route also needs credit in Billing.

Claude Code. A launcher starts Claude Code through Yugos.

Tested with Yugos

Paste this into your coding agent. It does the setup and checks the connection.

AI setup prompt
Connect Claude Code to the Yugos API, then confirm it works.

Facts:
- Yugos is a model router with an OpenAI-compatible API. Base URL: https://api.yugos.ai/v1
- Model ID: yugos (Yugos picks a model for each request)
- The key is in YUGOS_API_KEY. Never print it, paste it into a file I commit, or ask me to paste it here.
- Text and function tools work. Keep image options off.

Steps. Show each command and stop at the first failure:
1. Check the key is set: test -n "$YUGOS_API_KEY" && echo set || echo missing
   If it's missing, stop and tell me to create a key at https://www.yugos.ai/console/keys and run: export YUGOS_API_KEY="<my key>"
2. Check the key works: curl -s -o /dev/null -w "%{http_code}\n" https://api.yugos.ai/v1/models -H "Authorization: Bearer $YUGOS_API_KEY"
   Expect 200. If not, stop and show me the status.
3. Install the launcher with these commands, showing their output:
mkdir -p "$HOME/.local/bin"
curl --fail --show-error --location "https://www.yugos.ai/setup/yugos-clients.py" -o "$HOME/.local/bin/claude-yugo"
chmod 755 "$HOME/.local/bin/claude-yugo"
export PATH="$HOME/.local/bin:$PATH"
claude-yugo --check
4. Tell me to start new sessions with claude-yugo. Don't change my Claude Code settings.

When you're done, tell me how to confirm it: claude-yugo --check confirms the key and model ID.
Back up any config file before changing it, and don't change anything else.
Set it up yourself
Install and check
mkdir -p "$HOME/.local/bin"
curl --fail --show-error --location "https://www.yugos.ai/setup/yugos-clients.py" -o "$HOME/.local/bin/claude-yugo"
chmod 755 "$HOME/.local/bin/claude-yugo"
export PATH="$HOME/.local/bin:$PATH"
claude-yugo --check
Start
claude-yugo

Check it: claude-yugo --check confirms the key and model ID.

Not supported yet

Windsurf and Gemini CLI have no documented way to use a custom OpenAI-compatible endpoint.

What “not yet tested” means

These setups come from each tool’s own documentation, read in November 2026. We haven’t run them against Yugos yet, so tell support if one doesn’t work.