---
title: "Hermes Agent coding plan: config.yaml setup"
description: "Nous Research's Hermes Agent on a coding plan from $10/mo: base_url and API key in config.yaml for Sonnet 5.5, Opus 5.5 and DeepSeek. No surprise charges."
url: https://codingpass.ai/tools/hermes
language: en
updated: 2026-11-17
---
# The coding plan for Hermes.

Keep Hermes Agent as it is. Point it at CodingPass with a custom base\_url and API key in config.yaml, and run Claude Sonnet 5.5, Opus 5.5 and DeepSeek on one monthly coding plan.

- Setup in 60s
- No surprise charges
- Full refund if you've used 5% or less

~/.hermes/config.yaml

```
model:
  provider: custom
  default: claude-sonnet-5-5
  base_url: https://codingpass.ai/v1
  api_key: cp-live-…
```

Model ID for your plan

**Go**

claude-sonnet-5-5

**Plus**

claude-opus-5-5

**Pro**

claude-fable-5-2

## An agent that remembers and writes its own skills.

Nous Research's open-source agent, reachable from your chat apps and any server. All of it runs on CodingPass.

### Long-term memory

Remembers your projects, preferences and setup across sessions, and searches its own past conversations when it needs the details. Memory has a size cap: when it fills up, Hermes merges or drops old entries.

### Writes its own skills

A self-improving agent: after a finished task it saves the steps as a skill and refines it with each use. Every skill becomes a slash command in the CLI and your chat apps.

### An AI agent in Telegram, Discord and Slack

One gateway connects the CLI with Telegram, Discord, Slack, WhatsApp and Signal, and the conversation carries over between them.

### Always on, on any server

Runs locally, in Docker, over SSH, or on Modal, Daytona and Vercel Sandbox; serverless backends hibernate when idle. Describe a recurring job in plain words and the built-in scheduler runs it.

## Which plan for Hermes?

Every plan works with Hermes. Pick the Claude flagship you want as your default; all open-weight models come with every plan and share the same monthly allowance.

- [Full plan details](https://codingpass.ai/pricing#compare-plans)
- [Allowances for all 24 models](https://codingpass.ai/models)

### Go: Claude Sonnet 5.5

Best for Daily tasks, with open-weight models for scheduled jobs

$10/mo

$60 Sonnet allowance a month

**Requests / month**

≈2,050

**Per 5-hour window**

≈342req

**Each open-weight model**

$40/mo

**Model ID**

claude-sonnet-5-5

### Plus: Claude Opus 5.5

Best for Long tasks and several chat channels

$30/mo

$200 Opus allowance a month

**Requests / month**

≈3,420

**Per 5-hour window**

≈570req

**Each open-weight model**

$125/mo

**Model ID**

claude-opus-5-5

### Pro: Claude Fable 5.2

Best for Parallel subagents and the hardest problems

$100/mo

$500 Fable allowance a month

**Requests / month**

≈5,090

**Per 5-hour window**

≈848req

**Each open-weight model**

$300/mo

**Model ID**

claude-fable-5-2

Estimates from CodingPass traffic: about 100K tokens per request at roughly 5% input, 1% output and 94% cache read. A single Hermes task usually makes several requests. Allowances are metered at official API prices.

## Three steps. About a minute.

Hermes takes a custom endpoint in `config.yaml`. Set it once and every channel runs through CodingPass.

1. ### Create a key
   Sign in with Google, GitHub or email and pick a plan. Then go to Dashboard → Keys → New key. Keys start with `cp-live-`.
2. ### Set the endpoint
   Paste the block into `~/.hermes/config.yaml` with your key. Or run `hermes model`, choose Custom endpoint and enter the same values.
   `~/.hermes/config.yaml`
3. ### Run hermes
   Start chatting. On Plus or Pro, set `default` to your plan's flagship; switch any time with `hermes model`. [Get your request ID](https://codingpass.ai/tools#check).
   `hermes`

~/acme-ops Example session zsh

```
$ hermes

> Every morning at 8, post failed CI runs from last night to Telegram.

● Ran gh run list --status failure
● Scheduled daily 08:00 · telegram
● Saved skill ci-failure-digest
  First digest goes out tomorrow at 08:00.
```

Sonnet 5.5

$2.80 / $60.00 resets in 11d

## Get more out of every allowance.

One monthly allowance covers every model, with 5-hour and weekly caps. The right model for the task makes it go further.

### Give scheduled jobs an open-weight model

Unattended jobs add up. On Go each open-weight model can use up to $40 a month from the shared allowance, and costs less per request. Hand them digests and routine checks; keep Sonnet for the work you watch.

Go · up to, per model/mo ≈ req

Claude Sonnet 5.5 $60 ≈2,050

Kimi K2.7 Code $40 ≈1,570

DeepSeek V4.1 Flash 1017 $40 ≈13,600

GLM-5.3-Flash $40 ≈10,100

+ 4 more open-weight models, up to $40 each.

### Save repeated workflows as skills

Walked Hermes through a deploy or a weekly report once? Run `/learn` and it saves the steps as a skill. Next time call it by its slash command instead of explaining again: each run sends fewer tokens and your allowance lasts longer.

```
/learn how I just deployed the staging server
```

## Questions about Hermes.

**Is this allowed? Could my account get banned?**

You use only your CodingPass key. You never log in to, share or relay anyone's account, so your own Claude or ChatGPT account isn't involved. Your requests follow each vendor's usage policy.

**Does it work with Hermes Desktop?**

Yes. Desktop uses the same Hermes install and config, so the custom endpoint applies there too. Subagents can use their own CodingPass model under Settings → Model → Subagents.

**Do web search and image tools come with it?**

No. CodingPass serves the models. Hermes tools that need their own service, such as the Nous Portal tool gateway, are set up separately.

**What happens when I hit the 5-hour cap?**

Each 5-hour window allows 1/6 of the monthly allowance (on Go, $10 of Sonnet 5.5, about 342 requests). The cap counts all models together, so CodingPass then returns 429 with the reset time for every model. Wait for the reset, buy a reset card for the current window, or upgrade. [How caps and 429s work](https://codingpass.ai/usage-limits#response).

## Same key, every agent.
