Kato

Local-first AI conversation capture

Own your AI conversations.

Kato captures supported AI chats from IDEs, CLIs, and local apps into durable Markdown files you control. Your prompts, decisions, dead ends, and breakthroughs stop being disposable UI state and start becoming reusable project memory.

View on GitHub Download releases
  • Keep chats in files you own, not locked inside one assistant.
  • Export clean, vendor-agnostic Markdown that fits team workflows.
  • Give humans and agents the same durable source of context.

Quickstart

npm install -g @spectacular-voyage/kato
kato init
kato start
kato workspace init
kato workspace register --alias default
# in your next chat, ::capture-default

What changes with Kato

Disposable chat becomes durable project memory.

Capture once. Reuse across docs, handoffs, reviews, and the next agent session.

Built for AI-assisted development

Local daemon Local web console Terminal console Markdown output JSONL output In-chat commands

Local-only by default

Markdown-native output

Codex, Claude Code, and Gemini support

Operator Console

Keep discovered sessions, recordings, and workspace actions in one local view.

Kato's operator console gives you a grounded, local-first control surface for active chats. You can see session inventory, recording state, and the next action to take without digging through vendor UIs.

Kato operator console sessions view showing session inventory, recording state, and session actions.

Why it matters

AI-assisted development gets much more valuable when the conversation itself becomes part of the project.

Most teams are already doing serious work inside AI chats: debugging, planning, trying fixes, explaining tradeoffs, and handing work between people and agents. Kato makes that work durable instead of ephemeral.

01

Stop losing expensive context

The useful part of an AI session is rarely just the final code. Kato preserves the reasoning trail so you can revisit why a decision happened instead of reprompting from scratch.

02

Keep your workflow vendor-agnostic

Supported chats land in plain Markdown files you control. That keeps your project memory portable across tools, repos, and future assistant choices.

03

Make handoffs cleaner for humans and agents

Durable conversation history is easier to search, review, quote, commit, and reuse. Team memory stops living only in one developer's IDE sidebar.

What Kato Changes

Kato treats AI conversations like first-class work product.

If you are already building with AI, you are already generating a steady stream of useful project history. For most people, that history ends up dispersed across workstations and vendor silos. Kato lets you archive and organize it on your terms: per project folder, per repository, in shared storage, or into multiple destinations when the workflow calls for it.

That matters for solo builders who want better recall, and even more for teams who need continuity across sessions, machines, contributors, and assistants.

Without Kato

  • Important prompts disappear into tool-specific history.
  • Handoffs rely on screenshots, copy-paste, or memory.
  • Switching assistants means losing continuity.

With Kato

  • Sessions become files you can keep, search, and reuse.
  • Conversation outputs can land in local folders, checked-out repos, or shared team folders. They can fan out to multiple destinations when you need that.
  • Conversation history stays yours as tools change.

Compatibility

Works with the assistants people are already using.

Kato currently supports the following capture targets:

Codex

VS Code, CLI, local app

Claude Code

VS Code, CLI, local app

Gemini

VS Code, CLI

Install

Start with a prebuilt binary. No local compile step required.

The npm package installs prebuilt binaries for Windows, MacOS or Linux. You can also download a bundle from GitHub Releases and place it on your PATH.

Primary install

npm install -g @spectacular-voyage/kato

Or with pnpm

pnpm add -g @spectacular-voyage/kato

Quickstart

Get from install to your first captured conversation fast.

The public command is kato. Bundled kato-daemon and kato-web are managed for you.

Setup

kato init
kato start
mkdir chats-default
cd chats-default
kato workspace init
kato workspace register --alias default
kato init

Create Kato's base local configuration.

kato start

Start the local daemon that watches supported chats.

mkdir chats-default

Create a folder to hold your first workspace outputs.

cd chats-default

Move into that workspace folder before initializing it.

kato workspace init

Create the workspace-local Kato config.

kato workspace register --alias default

Register this workspace under the alias used by ::capture-default.

two easy control mechanisms

Capture from chat

Start a supported AI chat and put ::capture-default on its own line in a user message. It also helps to tell the model to ignore lines that start with ::.

See In-Chat Commands for the full command set.

Or use the local web UI

kato web init --username <username>
kato web start

Then open http://127.0.0.1:5173/login in your browser and start recordings from the Sessions page.

Learn more

Use Kato if AI conversations are part of how you build.

The repository has current install details, release artifacts, and the rest of the command surface. Start there, then make your AI workflow durable.

Open the repository Browse releases