# ohad6k/emulo [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/ohad6k/emulo  
**GitHub Stars:** 290  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/ohad6k-emulo

## Description
Loads your personal profile, mined from your local Claude Code/Codex/OpenCode logs, so your agent works like you instead of a cold start. One tool, loademuloprofile. Run with uvx emulo mcp.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "emulo": {
    "command": "npx",
    "args": ["-y","skills"]
  }
}
```

## Documentation

## What ohad6k/emulo MCP server does

The ohad6k/emulo MCP server gives an agent access to a profile derived from your previous AI coding-agent sessions. Instead of treating every conversation as a blank start, it can load observed preferences such as how you define completion, what kinds of changes you reject, when you ask for evidence, and how you communicate while working.

The profile is mined from selected evidence in local logs rather than from `CLAUDE.md` or another rules file. Emulo organizes the result into separate layers for work, design, writing, and video, so an agent can use the part relevant to the task.

## How it works

Emulo reads session histories from Claude Code, Codex, Copilot CLI, OpenCode, and Google Antigravity by default. The mining process extracts and redacts selected evidence, then produces a private working profile with dated receipts from the source sessions. The profile, extracted data, redaction results, caches, and adapters remain on the local machine.

The ohad6k/emulo MCP server exposes the resulting profile through one MCP tool named `loademuloprofile`. The profile can load natively in Claude Code and Codex, or through documented adapters for Cursor, Gemini, and OpenCode. OpenClaw and Hermes Agent use it as a standard skill.

Mining can use a hosted model or a local model. With a hosted model, selected evidence is sent to that provider; using a local model keeps the mining run on the machine. Emulo also includes a usage report that analyzes the user's own messages without making a model call, including repeated requests, re-explained context, rephrased requests, and corrections to prior answers.

## Setup and configuration

The README documents a CLI launch command: `uvx emulo mcp`. Claude Code users can alternatively add the `ohad6k/emulo` marketplace and install the `emulo@emulo` plugin. Codex users can add the same marketplace and plugin with the documented marketplace commands.

After installation, run `emulo:mine` and provide the session history to process. The usage report can be run with `python emulo.py --coach`, or limited to Claude Code with `python emulo.py --coach --source claude`.

## Tools and capabilities

The ohad6k/emulo MCP server provides:

- `loademuloprofile` for loading the mined profile into an agent context.
- Session mining for supported local agent histories.
- Separate work, design, writing, and video profile layers.
- Redaction and local storage of extracted data, caches, profiles, and adapters.
- A local usage report based only on messages written by the user.

## Limitations and notes

Emulo is not conventional explicit memory: it derives behavior from what session history demonstrates. Its proof methodology is described as unexecuted unless a separately approved evidence release exists. Hosted-model mining sends selected evidence to the chosen provider, so fully local processing requires a local model. The README states that the project is free, requires no account or sign-in, and is MIT licensed.

_Full upstream README: https://allmcps.com/mcp/ohad6k-emulo/readme_