Skip to content

Hermes Agent Integration

1. Prerequisites

codesw writes directly to ~/.hermes/config.yaml and ~/.hermes/.env (mode 0600), setting up the provider, base URL, default model and OpenAI key — no proxy, no middleman.

2.1 Install

bash
npm i -g @bumocc/codesw
# or
yarn global add @bumocc/codesw

2.2 Run the interactive mode

Just execute:

bash
codesw

Follow the prompts:

  1. Select provider: pick bumo.ai (BUMO API); existing profiles can be reused across clients.
  2. Profile name / API Key / Model / Small model: fill in as prompted. Get the key from the API Keys page.
  3. Apply to which client: choose Hermes Agent.

codesw writes the following into ~/.hermes/config.yaml:

  • model.provider: custom
  • model.base_url: https://api.bumo.ai/v1
  • model.default: <your chosen model>
  • model_aliases.codesw-<profile>: a convenient alias pointing at the chosen model.

It also writes OPENAI_API_KEY=<your key> into ~/.hermes/.env.

Once it finishes, start Hermes Agent as usual. To switch models or providers later, run codesw again and pick the existing profile.


Appendix: Manual Configuration (Alternative)

Edit ~/.hermes/config.yaml and merge:

yaml
model:
  provider: custom
  base_url: https://api.bumo.ai/v1
  default: claude-sonnet-4-6

model_aliases:
  bumo-claude-sonnet-4-6:
    model: claude-sonnet-4-6
    provider: custom
    base_url: https://api.bumo.ai/v1

Set the API key in ~/.hermes/.env (recommended mode 0600):

env
OPENAI_API_KEY=YOUR_API_KEY

Replace YOUR_API_KEY with your platform API Key.