An Obsidian plugin for browsing, creating, and editing AI agent skills, commands, and agents across 17 coding tools like Claude Code and Cursor.
rlaope/oh-my-hermes
A skills, routing and workflow layer for Hermes Agent that picks models per task lane and runs multi-step graph jobs in parallel.
- Stars
- 5,157
- Forks
- 412
- Open issues
- 15
- Last commit
- 19 hours ago
What is it?
oh-my-hermes is an add-on package for Hermes Agent, an open-source AI agent. You keep chatting with Hermes, and OMH decides which skill and which model each piece of work uses. Work is tagged by category, such as deep reasoning or quick edits, and each category has an ordered model chain with a reasoning effort setting. If a provider rejects a model, the chain moves to the next one.
For larger goals, the `/dag-ulw` workflow turns a request into a graph of small nodes. Nodes whose inputs are ready run in parallel, nodes running at the same time do not write to the same files, and a separate node checks the goal before the report. The package also includes more than a hundred specialist skills that the router loads on demand, plus a long-term project memory that Hermes loads without modifying Hermes' own memory.
OMH provides its own terminal interface on top of Hermes' gateway, shows live lane status with model, effort and token counts, and labels each status by stage and confidence, such as 'Plan · not run' or 'Test · verified'. It installs through a shell script, Homebrew, Bun or npm.
Key features
- 01Routes each work category to an ordered model chain with adjustable reasoning effort.
- 02Runs /dag-ulw goal graphs with parallel nodes and a separate goal-verification node.
- 03Provides over a hundred on-demand specialist skills for frontend, backend, Rust and security review.
- 04Offers long-term project memory with written admission decisions and token-budgeted recall.
- 05Shows live lane status, cost and tokens in its own terminal interface.
Use cases
- Running a large refactor or feature as parallel lanes that do not edit the same files.
- Assigning cheaper or faster models to simple edits and stronger models to hard reasoning.
- Keeping project conventions and decisions available across Hermes sessions.
- Checking that an agent's claimed completion is backed by a passing test or review.
Quick start
- 1
Install the CLI with npm
npm install -g oh-my-hermes - 2
Run OMH setup
omh setup
Things to know
- It is an add-on layer for Hermes Agent and does not work without it.
- Model routing depends on access to model providers configured in your Hermes setup.
In collections
Topics
FAQ
- What is oh-my-hermes?
- A skills, routing and workflow layer for Hermes Agent that picks models per task lane and runs multi-step graph jobs in parallel. oh-my-hermes is an add-on package for Hermes Agent, an open-source AI agent. You keep chatting with Hermes, and OMH decides which skill and which model each piece of work uses. Work is tagged by category, such as deep reasoning or quick edits, and each category has an ordered model chain with a reasoning effort setting. If a provider rejects a model, the chain moves to the next one.
- Is oh-my-hermes free to use? What license does it have?
- oh-my-hermes is open source under the MIT license. Check the license terms before using it in commercial products.
- Is oh-my-hermes still maintained?
- Actively maintained. The last commit to oh-my-hermes was on Nov 5, 2026.
- What can I use oh-my-hermes for?
- Running a large refactor or feature as parallel lanes that do not edit the same files. Assigning cheaper or faster models to simple edits and stronger models to hard reasoning. Keeping project conventions and decisions available across Hermes sessions. Checking that an agent's claimed completion is backed by a passing test or review.
- How do I install oh-my-hermes?
- Install the CLI with npm: npm install -g oh-my-hermes Run OMH setup: omh setup
- Where can I get oh-my-hermes?
- The source code and installation instructions are on GitHub: https://github.com/rlaope/oh-my-hermes
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source platform for building versioned agent skills from visual workflows or natural-language descriptions, exportable to Claude Code, Cursor, and APIs.
An open-source agent skill that helps AI agents understand existing software, plan a reconstruction, and validate it, working alongside the REA tool.
A VS Code extension that puts Claude Code, Codex and other coding agents in one conversation, so they can give separate opinions, hand off work and debate solutions.
A modular runtime and orchestration system that coordinates AI coding agents through gated pipelines for tasks, review, shipping and docs across Claude Code, Codex, Cursor and more.
A CLI that one-command sets up Claude Code and Codex with bilingual prompts, workflows, API provider presets and MCP configuration.