Jawf

Hermes Claude Plan — Git skill for Claude Code

Git community

Hermes agent plugin that routes Anthropic API requests through anthropic claude plan.

How to install Hermes Claude Plan

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Jawf/hermes-claude-plan and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Hermes Claude Plan does

Hermes agent plugin that routes Anthropic API requests through anthropic claude plan. zero modifications to hermes source code. Fully compatible with `hermes update` and `git pull` upgrades

Alternatives in Git

  • Openclaude — Open Claude Is Open-source coding-agent CLI for OpenAI, Gemini, DeepSeek, Ollama, Codex, GitHub Models, and 20 22.5k ★
  • Anthropic 课程 — by anthropics - 官方课程材料 19.7k ★
  • PR Review Toolkit — Specialized agents for reviewing pull requests 14k ★

README

hermes-claude-plan

[English](./README.md) | [中文](./README_CN.md)

A [Hermes Agent](https://github.com/NousResearch/hermes-agent) plugin that routes Anthropic API requests through the Claude subscription channel for improved rate-limit allocation.

**Non-invasive design** -- installs entirely under `~/.hermes/plugins/`, zero modifications to hermes source code. Fully compatible with `hermes update` and `git pull` upgrades. Install and uninstall without side effects.

Prerequisites

  • Hermes Agent v0.9.0+
  • Claude Code installed and logged in (claude CLI available)
  • An active Claude Max / Team / Enterprise subscription
  • Python >= 3.11

Install

hermes plugins install Jawf/hermes-claude-plan

Or clone manually:

git clone https://github.com/Jawf/hermes-claude-plan.git ~/.hermes/plugins/anthropic_plan

Configure

Set `anthropic_plan` as your default provider in `~/.hermes/config.yaml`:

model:
  default: claude-opus-4-6      # or claude-sonnet-4-6, etc.
  provider: anthropic_plan

Or switch at runtime:

# Inside a hermes chat session
/model anthropic_plan:claude-opus-4-6

How It Works

hermes chat
    |
    v
[Anthropic Messages API request]
    |
    v
[localhost:28765 proxy]
    |   - Reads OAuth token from ~/.claude/.credentials.json
    |   - Rewrites headers (billing, user-agent, betas)
    |   - Rewrites body (system prompt prefix, tool name prefix)
    v
[api.anthropic.com/v1/messages]
    |
    v
[SSE response streamed back]

The plugin starts a lightweight HTTP proxy on `127.0.0.1:28765` (loopback only). The proxy:

  1. Reads the current OAuth token from ~/.claude/.credentials.json on every request (picks up token refreshes automatically)
  2. Adds Claude Code identity headers (x-anthropic-billing-header, user-agent, OAuth betas)
  3. Prepends the Claude Code identity block to the system prompt
  4. Prefixes tool na