Customized Plugins

Beyond the Claude and Codex plugins, HyperMemory has plugins made for other agents and tools — and because the plugins are open source, you can build a customized one for your own team.

Plugins for other tools

Tool	Plugin	Install
OpenClaw	Native `hypermemory-cli` plugin: recalls before each prompt, saves after each run	`hm setup openclaw` — or the skill from ClawHub: `clawhub install hypermemory`
Hermes Agent	HyperMemory as Hermes’s memory provider	`hm setup hermes`
Kimi Code	Skill, hooks and memory-writer	`hm setup kimi`
ComfyUI	`Extract Text - Hypermemory` node: recalls brand context from HyperMemory while building prompts	Clone hypermemory-cumfyui-extract-text-node into `ComfyUI/custom_nodes`
Cursor, Windsurf, Cline	The HyperMemory skill for coding agents	hypermemory-skill, or `hm_skill` with `variant: "cursor"`
Your own app	TypeScript SDK with React and Svelte components and graph visualizers	hypermemory-sdk

The hm setup plugins are installed by the CLI. Install it, sign in, and set up every agent on your computer at once:

pipx install hypermemory-cli
hm login
hm setup all

Build your own plugin

Every HyperMemory plugin is a thin layer over the same MCP server and tools, so a customized plugin for your team is mostly a matter of changing the skill and the hooks.

Start from our plugin

Fork hm-plugins-claude for Claude, or hm-plugins-openai for Codex. Each is a plugin marketplace: one repository that lists the plugins inside it.

Customize it

Common changes:

The skill (skills/hypermemory/SKILL.md) — add your team's rules: what to remember, how to name nodes, which node types to use for your domain.
The hooks — make recall and saving stricter or lighter, or add your own checks.
The memory-writer agent — change what it records at the end of each turn.
Extra skills — bundle your team's own skills next to the HyperMemory one.

Keep the MCP server URL, https://api.hypermemory.io/mcp, so your plugin signs in to HyperMemory with OAuth.

Publish it to your team

Rename the marketplace and plugin, push the repository (it can be private if everyone on your team can read it), and share the install commands:

/plugin marketplace add your-org/your-hypermemory-plugin
/plugin install hypermemory@your-marketplace

You may not need a fork at all. If what you want is shared instructions, publish them as organization skills: every member’s agent receives them through hm_skill, alongside the standard HyperMemory plugin.

See GitHub for all of HyperMemory’s public repositories.