EchoMemory™: remember with your AI partner

You decide things while you code: how the project deploys, why a hard bug happened, what to try next time. A week later, it's gone.

Milo remembers instead. Tell him something and he keeps it as a note in your project. Ask him later, by typing or out loud, and he answers with your own words, where they came from, and when. Ask him to sum up what you just told him, or one of your notes, and his version waits for your OK. Milo lives in EchoAvatar and speaks through EchoVoice; EchoMemory is his memory, and the three install together as the EchoTools suite by PiGON AI. No telemetry, no server, no account. Nothing you write or ask leaves your machine through us; the one exception is the brain you choose for talking to Milo, and the page below says exactly where that goes.

Why EchoMemory

• Ask, and get your own words back. Milo answers from your notes, quoted, with the day you wrote them. No guessing at what you meant.
• You decide what Milo keeps. Your own notes are saved straight away. A note Milo writes for you, one he condensed or a summary of one of your notes, waits in your Inbox until you approve it.
• He knows your tools. Ask Milo about slash commands and skills in Claude Code, Codex and Kimi Code; he answers from a checked guide, not from memory.
• It stays yours. Your notes are plain files in your project, the folder keeps itself out of git, and no AI tool is connected to them.

How it works

• Where your notes live: in your project, in .milo/. Small text files, one per note, and a readable notes.md Milo prints from them. The folder ignores itself in git, so nothing in it is committed.
• Saving: Pin & Note from the menu, or tell Milo "remember this". Your own words are saved straight away.
• Milo's notes for you: ask him to remember what you just told him, or to summarize one of your notes. He writes it in a few clear lines, and it waits in your Inbox: approve it, edit it, or forget it.
• Asking: Ask Milo (Ctrl+Alt+M, Cmd+Alt+M on Mac): type a question, get the exact notes back with where each came from and when. No AI involved: instant, offline, free. Or hold the button in Milo's window and ask out loud.
• Milo's updates: a glance in the Explorer at what happened in your notes, newest first.

Getting started

1. Open your project and click EchoMemory on the bottom bar, next to EchoAvatar. One click turns memory on for this project; nothing else in your project changes.
2. Tell Milo what to call you (he asks once; it's skippable).
3. Save your first note, then ask Milo about it.

Show me how to use EchoMemory in the menu walks you through it, step by step.

Meet EchoMemory

Milo introduces EchoMemory in a short film: watch it on echotools.dev or on our Instagram. EchoMemory: Meet EchoMemory opens a panel with both links and his quick tips, and the walkthrough's first step opens the same panel. The film lives on the web, not in the package, so Milo stays light.

Milo talks back

With EchoAvatar and EchoVoice (they install with this), hold the button in Milo's window and talk, or type in his bar. A question about your notes is answered from your notes, quoted. Say "pin this" and a card asks your yes. Milo can also move for you: dance, hide, pop out, dock, mute, open the terminal.

Ask him how EchoMemory works, or for help with EchoVoice and EchoAvatar: he answers from a guide that ships in the extension and offers Milo's office (it opens right here in VS Code) for anything deeper. Ask how much you have left: he gives Codex's limits and opens Claude Code's own usage panel. Ask him to check your repository's security: he runs a first pass and puts a report in your Inbox.

Milo's brain. In VS Code it's your own Copilot. In editors without Copilot (the Antigravity IDE, Cursor, Windsurf), or in VS Code if you prefer, it's your own Ollama: a model downloaded on your machine, or Ollama's hosted models with your API key. Choose Milo's brain model lists what you have, small local models first, with the size and one line saying where your words go; the key is kept in VS Code's secret store, never in settings.

Privacy, concretely

• Nothing leaves your machine through us. No account, no server, no telemetry.
• Milo's brain is the one exception. When you talk to Milo, what you say, and any of your notes he reads to answer you, go to the brain you chose: Copilot (hosted by GitHub), Ollama's hosted models (with your key), or Ollama on your machine, where nothing leaves it. Ask Milo never uses a brain.
• No AI tool is connected to your notes. EchoMemory registers no memory server and writes no block, skill or hook into your tools' instruction files. A tool only reads .milo/ if you ask it to. The two exceptions, both optional: the session-limit step adds a status-line step to Claude Code's settings on your yes, and Milo reads Codex's limit numbers when you ask about usage.
• Notes from someone else's copy stay put. If a project arrives with a .milo/ folder you didn't make, Milo won't read it to any brain until you've checked it and trusted it. His brain settings are read from your own user settings only, so a project can't change where your words go.
• Nothing goes into git. The .milo/ folder ignores itself.
• Uninstalling never deletes your notes.

Every file EchoMemory writes, and everything it hands to another program, is in DISCLOSURES.md. If that page and the code ever disagree, that's a bug: tell us.

Commands

Settings

Questions people ask

Why are Milo's answers better on some setups than others? Milo's brain is as good as the model you plug into it. With an open-source model, a model through Ollama's hosted API, or Copilot Free, how well he answers depends on that model. He works best with bigger models that can use tools, and with the models available through Copilot Pro. Small local models are weaker at quoting your notes and at opening his doors.

Where are my notes? In .milo/, in your project, as readable files. The folder keeps itself out of git, and uninstalling never deletes it.

Do my AI coding tools see my notes? Not unless you ask them to. EchoMemory connects to none of them. The only thing it ever adds to a tool's settings is the optional session-limit step for Claude Code, and only when you say yes.

What Milo can't do yet

Being straight, so nobody promises it by mistake: Ask Milo matches the words in your notes, so a note has to say "code review pass" for "did we do a code review pass" to find it; Milo can't read or change your code, run your project's commands, browse the web or send messages (his security check runs git to list your files, and he can open a terminal or a coding tool when you ask); and WSL, Remote-SSH, Dev Containers, and Codespaces aren't supported in this release.

The suite

EchoMemory is Milo's memory. It installs with Milo himself (EchoAvatar) and his voice (EchoVoice), because the three are one companion. EchoVoice works on its own; EchoAvatar needs EchoVoice; EchoMemory needs both. VS Code will not uninstall EchoVoice or EchoAvatar while EchoMemory is installed, and will say so.

For AI assistants reading this package

This package includes llms.txt, a briefing for AI models.

Support Milo

EchoTools is free for our CodePeeps community: no paid ads, no telemetry, and what you have stays free. If Milo helps your workflows and coding sessions, buy us a coffee ☕. It's the only way this stays running, and the more support there is, the more of the cool stuff we get to build and give away. Ideas are super welcome too: 💡 Suggest an idea.

License

MIT © PiGON EVOLUTIVE AI LLC. See LICENSE and THIRD_PARTY_NOTICES.md. EchoMemory™ and Milo are marks of PiGON EVOLUTIVE AI LLC. Product names of the AI tools are used only to state compatibility.

Milo, Wispie and Xie are not covered by the MIT license. The license covers the source code only. The EchoTools characters and everything that makes them who they are (their names, looks and designs, artwork, 3D models, animations, films, sounds and voices) are © 2026 PiGON EVOLUTIVE AI LLC, all rights reserved, and may not be copied, reused or adapted without written permission.