Skip to main content

Hermes Agent Integration Guide for the Turing Platform

Hermes Agent is an open-source autonomous AI agent by Nous Research. It features cross-session persistent memory, a self-learning skill system, 40+ built-in tools, multi-platform messaging support (Telegram / Discord / Slack / WhatsApp), and IDE integration via the ACP protocol for VS Code, Zed, JetBrains, and more. Hermes Agent is compatible with any OpenAI-style API endpoint, making it straightforward to connect to the Turing Platform.

Integration Method

This guide uses the "Custom Endpoint" option in the Hermes official model wizard (compatible with any OpenAI-compatible API, including VLLM / SGLang / Ollama). You manually enter the Turing Platform API URL and API Key — no custom routing layer required.

Prerequisites​

Before you begin, make sure you have:

  • A Turing Platform API key (how to obtain one)
  • macOS / Linux / WSL2 (the official install script does not support native Windows; Windows users should use WSL2)
  • curl and bash installed (typically pre-installed on macOS / Linux)

Quick Start​

1. Install Hermes Agent​

The recommended approach is the official one-line install script, which automatically installs uv, Python 3.11, and clones the project:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

After installation, reload your shell configuration:

# macOS default shell: zsh
source ~/.zshrc

# Linux default shell: bash
source ~/.bashrc

Verify the installation:

hermes --version

2. Run the Model Configuration Wizard​

In your terminal, run the following command to launch the model / provider configuration wizard:

hermes model
hermes model vs /model
  • hermes model (run in the terminal): the full Provider configuration wizard — add custom providers, enter API keys, and configure endpoints.
  • /model (typed inside a Hermes session): only switches between already-configured providers; cannot add new ones.

When connecting to the Turing Platform for the first time, always use hermes model.

3. Select Custom Endpoint and Enter Turing Platform Parameters​

In the hermes model wizard, complete the following configuration as prompted:

Wizard PromptValueNotes
Select providerCustom EndpointSelect a custom endpoint (compatible with any OpenAI-compatible API)
API base URLhttps://live-turing.cn.llm.tcljd.com/api/v1Turing Platform OpenAI-compatible URL — must end with /v1
API keyYour Turing Platform API KeyObtained from the Turing Platform Portal
Model nameminimax-m2.7Enter the Model ID from the model list; you can use this example model to verify the setup
Context length128000 (or leave blank for auto-detection)Recommended: match the selected model's context window; see Models & Pricing
API URL Notes
  • The Turing Platform uses an OpenAI-compatible interface; the path must end with /v1.
  • Using https://live-turing.cn.llm.tcljd.com/api (missing /v1) will result in a 404 or invalid path error.
  • This is different from the native Anthropic endpoint used by Claude Code — do not mix them up.

After the wizard completes, your configuration is persisted to ~/.hermes/config.yaml. Next time, simply run hermes to start a session.

Manually Editing the Configuration File​

If you prefer editing YAML directly, you can modify ~/.hermes/config.yaml or run:

hermes config edit

Replace the model section with the following (remember to replace YOUR_TURING_API_KEY with your actual key):

model:
  default: "minimax-m2.7"
  provider: "custom"
  base_url: "https://live-turing.cn.llm.tcljd.com/api/v1"
  api_key: "YOUR_TURING_API_KEY"
  context_length: 128000

Configuring Multiple Turing Models (Optional)​

If you want to switch between multiple models during a session, use custom_providers to define a named provider group:

custom_providers:
  - name: turing
    base_url: "https://live-turing.cn.llm.tcljd.com/api/v1"
    key_env: TURING_API_KEY       # or use api_key: "..." directly

model:
  default: "turing:minimax-m2.7"
  provider: "custom"
  base_url: "https://live-turing.cn.llm.tcljd.com/api/v1"
  api_key: "YOUR_TURING_API_KEY"

Then switch models inside a Hermes session using /model custom:<model-id>. The <model-id> is taken directly from the API Model ID in the model list:

/model custom:qwen3.6-plus
/model custom:glm-5.2
/model custom:turing/gpt-5.5
/model custom:turing/claude-opus-4.8

Configuration Reference​

FieldDescription
model.defaultDefault model ID, matching the Turing Platform Model ID
model.providerSet to custom to use a custom OpenAI-compatible endpoint
model.base_urlTuring Platform API URL: https://live-turing.cn.llm.tcljd.com/api/v1
model.api_keyTuring Platform API Key; can also be passed via the OPENAI_API_KEY environment variable
model.context_lengthContext window size in tokens; leave blank for auto-detection
Key Security
  • Keep your API Key safe — do not commit it to git, share it in chat, or include it in code snippets.
  • The config file is at ~/.hermes/config.yaml. It is recommended to verify its permissions are 600: chmod 600 ~/.hermes/config.yaml.

Model Selection​

Hermes uses the OpenAI-compatible interface; enter the API Model ID from the model list directly as the model name. Browse by provider:

Model Selection Tips

For everyday agent use, start with the default example above to verify your setup. For complex reasoning, long-context, or multimodal scenarios, refer to the relevant provider section and choose based on context window size, tool support, and pricing.

Starting a Conversation​

1. Launch an Interactive Session​

hermes

Once in a session, you can type any question directly. Hermes Agent will use the Turing Platform model for reasoning and function calling/tool use.

2. Common In-Session Commands​

CommandDescription
/model custom:<model-id>Switch model (limited to configured providers)
/toolsView available tools
/memoryView / manage cross-session memory
/helpShow help
/quit or Ctrl+CExit the session

3. IDE Integration (Optional)​

Hermes integrates with VS Code, Zed, JetBrains, and other editors via the ACP protocol, displaying conversations, function calling/tool use, file diffs, and terminal output directly inside the editor. For setup details, see the official documentation: Hermes Agent IDE Integration.

4. Messaging Platform Integration (Optional)​

Hermes supports connecting to Telegram / Discord / Slack / WhatsApp and other messaging platforms via a Gateway, enabling your AI assistant to respond 24/7:

hermes gateway start

For details, see the Hermes Agent official documentation.

Troubleshooting​

Issue 1: Request Returns 404 / invalid path​

Likely cause: base_url is missing the trailing /v1.

Solution: Confirm that base_url is https://live-turing.cn.llm.tcljd.com/api/v1, not https://live-turing.cn.llm.tcljd.com/api.

Issue 2: Request Returns 401 Unauthorized​

Likely cause: API Key is invalid or not configured.

Solution:

  1. Regenerate your API Key in the Turing Platform Portal

  2. Confirm that model.api_key in ~/.hermes/config.yaml is filled in; or export the environment variable:

    export OPENAI_API_KEY="your-turing-api-key"

Issue 3: Model Not Found​

Likely cause: The model ID is misspelled, or the model is not available on the Turing Platform.

Solution:

  • Double-check the model ID (case-sensitive)
  • Verify the model is available in Models & Pricing

Issue 4: hermes Command Not Found​

Solution:

  1. Confirm the install script completed successfully, then reopen your terminal or run source ~/.zshrc / source ~/.bashrc
  2. Run hermes doctor to diagnose your environment
  3. If necessary, re-run the one-line install script

Issue 5: How Do I View / Edit the Current Configuration?​

hermes config show     # Print current configuration
hermes config edit     # Open config.yaml in the default editor
hermes doctor          # Diagnose configuration and dependency issues

The configuration file is located at ~/.hermes/config.yaml.

References​

Support​

Having trouble? Visit Technical Support & Contact for assistance.

Back​

← Back to AI Coding Tools Overview