TokenLab Provider for Hermes Agent

Use TokenLab models from Hermes Agent through TokenLab's OpenAI-compatible API. The plugin registers TokenLab in the Hermes provider picker, discovers models from the live TokenLab catalog, and keeps a curated agent-capable fallback list for offline setup.

Install

pip install tokenlab-hermes-provider
export TOKENLAB_API_KEY="your-tokenlab-api-key"
hermes plugins enable tokenlab-provider --no-allow-tool-override
hermes model

Choose TokenLab, then select a model returned by https://api.tokenlab.sh/v1/models.

For a custom gateway URL, set the OpenAI-compatible /v1 base:

export TOKENLAB_BASE_URL="https://api.tokenlab.sh/v1"

Hermes uses TokenLab's OpenAI-compatible Chat Completions surface for agent execution. TokenLab also exposes native OpenAI Responses, Anthropic Messages, and Gemini generateContent APIs for applications that need those request contracts.

Verify

hermes plugins list
hermes model

The plugin appears as tokenlab-provider, and the provider picker shows TokenLab. Hermes keeps third-party plugins opt-in, so the plugins enable step is required once after installation.

Development

python -m venv .venv
. .venv/bin/activate
python -m pip install -e '.[test]'
pytest

Links

• Website: https://tokenlab.sh
• API compatibility guide: https://docs.tokenlab.sh/zh/guides/ide-sdk-compatibility
• Hermes Agent provider plugin guide: https://github.com/NousResearch/hermes-agent/blob/main/website/docs/developer-guide/model-provider-plugin.md