Design a plugin that installs natively across Claude Code, Codex, OpenCode, and other agent platforms
Follow a ten-rule checklist for keeping a plugin's core portable across platforms
Use the step by step guide when adding support for a new agent platform
| nagisanzenin/omniplugin | anmoln7/agent-standard-oss | dmaynard51/ultima5-ios | |
|---|---|---|---|
| Stars | 5 | 5 | 5 |
| Language | Shell | Shell | Shell |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 3/5 | 2/5 | 3/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Install routes differ per platform and most are not yet verified live by the author.
Omniplugin is a playbook, not a single tool, that explains how to build one plugin that works natively across many different AI coding assistant platforms, including Claude Code, OpenAI Codex, OpenCode, Hermes, and Antigravity. The problem it addresses is that each of these platforms has its own plugin format, its own way of discovering plugins, and its own marketplace, so building a separate version for each platform multiplies the work needed for every bug fix or update. The approach described is to keep one shared core, made up of reusable skills written in a common format plus a small command line engine with no external dependencies, and then wrap that core in the thinnest possible glue code for each platform. The idea is that the command line is the one thing almost every platform can call into, even though very few platforms share an actual plugin programming interface. The playbook was created based on a real project called engram, which the same author built and got running on six different platforms, including one that is a chat gateway rather than a coding tool. Each rule in the documentation is tied to a real bug, code review comment, or user report that led to it, rather than being purely theoretical advice. The repository is organized as a set of documentation files rather than code you run directly. It includes an overview of how a project built this way is structured, a list of ten engineering rules for keeping a single codebase portable, a step by step checklist for adding support for a new platform, notes on maintaining the project across platforms as they change over time, and a collection of real failures the author encountered along with the fixes. Omniplugin itself can also be installed as a plugin, through a component called a skill, so that an AI coding assistant can read the playbook and refer back to it automatically whenever the user is working on plugin related tasks. The author notes that most of these installation paths for other platforms have not yet been directly verified as working.
Omniplugin is a playbook and skill for building one AI-agent plugin that installs natively on Claude Code, Codex, OpenCode, and other platforms.
Mainly Shell. The stack also includes Shell, CLI, Agent Skills.
No license information is stated in the README excerpt.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.