Hermes Agent

Nội dung tài liệu kỹ thuật hiện được cung cấp bằng tiếng Anh.

Hermes Agent is an open-source, terminal-native coding and task agent from Nous Research. Uptech API is not a built-in Hermes provider — configure it as a custom endpoint (or OpenAI API + OPENAI_BASE_URL).

Prerequisites

  • Hermes installed (Hermes docs)
  • An Uptech API API key from Workspace → API Keys
  • A model with enough context for agent tools (Hermes typically needs large context windows — prefer 64K+)
hermes model
  1. Choose Custom endpoint (self-hosted / OpenAI-compatible).
  2. Enter:
    • API base URL: https://api.uptech.vn/v1
    • API key: your sk-apigo-... key
    • Model name: a Chat Completions slug from the Model Catalog, for example gpt-5.6-sol
  3. Start a session:
hermes
# or
hermes --tui

Manual configuration

1. Store the API key

Add to ~/.hermes/.env:

OPENAI_API_KEY=sk-apigo-...
OPENAI_BASE_URL=https://api.uptech.vn/v1

Or keep the key only in the custom-endpoint block below.

2. Edit ~/.hermes/config.yaml

Custom endpoint:

model:
  provider: custom
  default: gpt-5.6-sol
  base_url: https://api.uptech.vn/v1
  api_key: sk-apigo-...

OpenAI API provider with base URL override:

model:
  provider: openai-api
  default: gpt-5.6-sol

With OPENAI_API_KEY and OPENAI_BASE_URL set in ~/.hermes/.env as above.

3. Launch

hermes chat --provider custom --model gpt-5.6-sol

Model IDs

Use an exact catalog slug that supports the OpenAI-compatible endpoint:

  • gpt-5.6-sol
  • qwen3.8-max
  • deepseek-v4-flash

Reliability tips

Troubleshooting

Symptom Fix
No API key / provider not found Re-run hermes model or verify ~/.hermes/.env and config.yaml
401 / 403 Confirm the key in Workspace; check spend caps
Context length errors Switch to a larger-context model on Uptech API
Model not found Copy the slug from Mô hình without renaming

Resources