Emulo plugin
A plugin and CLI that mine your messages to coding agents into a personal you.md profile, with every rule backed by a dated quote
- 301 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The plugin reads full session logs, which may hold secrets, client names and third-party code
- With mining on a hosted model, selected fragments of your messages go to the provider
- A profile with verbatim quotes installed in AGENTS.md easily ends up in a shared repository
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/emuloDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The plugin reads full session logs, which may hold secrets, client names and third-party code
- With mining on a hosted model, selected fragments of your messages go to the provider
- A profile with verbatim quotes installed in AGENTS.md easily ends up in a shared repository
This is third-party code. Review the repository files before installing.
What it does
Emulo reads the session logs of Claude Code, Codex, Copilot CLI, OpenCode and Google Antigravity and picks out the messages you wrote to agents. From them it builds a you.md profile. It records what "done" means to you and what you reject on sight. A separate part covers how you debug, design interfaces and write. Each rule carries dated quotes from your own messages. The mode launched by an agent writes four separate files for work, design, writing and video, so a host loads only the fitting one. The `emulo verify you.md` command searches the mined sessions for every quote in the profile and exits with an error if a quote is missing. Then the profile is installed where the agent reads it. For Codex and OpenCode that is the project AGENTS.md, for Claude Code a skill named you that opens with `/you`, and for other hosts the file goes where they look for it. The plugin is invoked with the phrase "run emulo" or the emulo:mine skill.
Who it is for. For people who work with coding agents a lot and switch between Claude Code, Codex and OpenCode.
Good fit when
- You have corrected the agent on the same things many times and want to turn that into written rules
- You want to be sure that every rule in the profile is backed by your own words
- You want separate profiles for work, design, writing and video and load only the needed one
- You move between Claude Code, Codex and OpenCode and want to keep one profile for all
- You want to preview what will be read: `emulo --dry-run` prints the number of sessions, messages and tokens without writing
Not a fit when
- You work on a client's closed code: session logs hold its fragments, and a profile with quotes would carry them into a file
- You need memory of project facts: Emulo describes your habits, and project-memory tools fit facts
- You need a guaranteed effect: better agent work has not been proven yet
Example request
run emulo. Show me the full-history mining plan and its cost, and after I approve it, build the work and writing profiles.Limitations
Profile loading by the host has not been verified everywhere. In a test on September 25, 2026 the profile reached the model through AGENTS.md in Codex and was loaded into OpenCode instructions. Claude Code does not always open the you skill by itself and needs the `/you` command. For Cursor, Gemini, OpenClaw and Hermes Agent the file is placed, but loading there is unconfirmed.
In a series of 60 runs the profile built from your messages did not separate from an invented one of the same length on any of nine comparisons. In a September rerun it separated in 5 of 9 comparisons on Fable 5.1 and in 1 of 9 on Opus 5. Mining spends model tokens, so the agent shows the cost and waits for your approval.
How to disable. Remove the plugin with `/plugin` in Claude Code or the codex plugin command in Codex. The installed profile sits in the project AGENTS.md or in the you skill directory; delete the profile block there. The emulo-out directory with the corpus, chunks and RUN_ME.md can be deleted by hand.
FAQ
How do I install Emulo in Claude Code?
As a plugin: `/plugin marketplace add ohad6k/emulo`, then `/plugin install emulo@emulo`. The other route is `npx skills add ohad6k/emulo@emulo` and the phrase "run emulo". The plugin adds five skills: mine, work, design, write and video.
Which logs does Emulo read?
Claude Code, Codex, Copilot CLI and Google Antigravity sessions in JSONL, and for OpenCode the SQLite database or JSON session files. Only messages you typed are used. AGENTS.md, CLAUDE.md, memory files and self-descriptions are rejected as sources.
Where do my messages go?
Extraction, secret redaction, caches, the profile itself and the adapters run on your machine. Mining is the exception: with a hosted model the selected fragments go to the provider, and with a local model everything stays with you.
How do I check that the profile quotes are real?
Run `emulo verify you.md`. The command searches the mined sessions for every quote, reports those it cannot find and exits with an error. Quotes resting on one session are flagged separately: that is context, too early to call a rule.
Does Emulo have an MCP server?
Yes, `uvx emulo mcp` starts a stdio server with one tool, load_emulo_profile, which returns the work, design, writing or video profile. It makes no network requests.
Discussion
No questions or reviews yet.
Related
PluginsA self-improvement plugin for Claude Code, Codex and OpenCode: turns corrections into rules the agent follows in later sessions
A skill for messaging between Claude Code, Codex, Gemini CLI and other CLI agents through a shared local SQLite database, written in bash
Alook gives local Claude Code, Codex, Cursor and OpenCode agents a handle and an inbox so a team can talk to them in shared rooms like to people
Memory for coding agents in plain Markdown: the agent keeps specs, decisions and lessons itself, and each new session starts from them. No vector database
Local memory for Claude Code, Codex and Cursor Agent: learns from your activity on a Mac and gives agents context with source receipts over MCP
A Claude Code plugin that connects the agent to shared organizational memory with access control, audit and signed erasure
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| Emulo (this entry) | no | 301 | |
| Writing DNA Distiller | no | 2,755 | derives a style profile from a set of an author articles |
| Claude-Mem | no | 104,106 | saves project session observations and injects them into new sessions |
README in short
The project's claims are split into three levels: quote tracing, profile loading by the host and better agent work, and each states what has been verified. The workflow, what is read and what goes to the model are laid out in docs/PROFILE-FLOW.md. The Emulo Proof v1 methodology is written but has not been run yet.