AIexporter

Automatically export your AI conversations to clean, portable Markdown archives — locally, privately, forever.

A browser extension that captures conversations from ChatGPT, DeepSeek, Gemini, and Google AI Studio, then writes them as structured Markdown + JSON bundles to a directory you control. Built as a pnpm monorepo with a WXT-based Chromium extension, background queue orchestration, and an optional local Fastify archive server.

What it does

Auto-discovery: Detects conversations on supported platforms via network interception and DOM observation — no manual clicking required.
Background queue: Schedules worker tabs to extract conversations without blocking your browsing.
Normalized Markdown: Every conversation becomes a clean .md file with YAML frontmatter, plus a .bundle.json for programmatic use.
Configurable export directory: Write to any local folder (with native host) or use the browser Downloads folder.
Optional local archive server: Sync bundles to a self-hosted Fastify + SQLite service for search and query.
Dashboard UI: Monitor queue status, view logs, and manage settings from a built-in React dashboard.

Supported Platforms

Platform	Discovery	Export	Thinking Blocks	Attachments
ChatGPT	Network interception	Full conversation	Yes	—
DeepSeek	Network + history sidebar	Full conversation	Yes	File metadata
Gemini	RPC response parsing + DOM	Full conversation	Yes	Images, links
AI Studio	Library page sweep + gRPC-Web	Full conversation	Yes	Drive links

Workspace Layout

AIexporter/
├── apps/
│   ├── extension/          WXT Chromium extension
│   │   ├── entrypoints/    Content scripts, main-world bridges, background
│   │   ├── src/            Background runtime, dashboard UI, runtime helpers
│   │   ├── native-host/    Windows native messaging host
│   │   ├── scripts/        Registration & verification scripts
│   │   └── wxt.config.ts   WXT configuration
│   └── server/             Fastify + SQLite archive service
│       └── src/
│           ├── index.ts    Server entry point
│           ├── app.ts      Fastify app setup
│           ├── db.ts       SQLite database layer
│           └── routes/     API route handlers
├── packages/
│   ├── adapter-sdk/        Shared types, defaults, fingerprinting, utilities
│   ├── adapters-chatgpt/   ChatGPT adapter (discovery + extraction)
│   ├── adapters-deepseek/  DeepSeek adapter (discovery + extraction)
│   ├── adapters-gemini/    Gemini & AI Studio adapter (RPC parsing)
│   ├── core-markdown/      Markdown serializer with frontmatter
│   └── core-schema/        Zod schemas, bundle types, title/URL normalization
├── docs/                   Reference documentation
├── docker-compose.yml      Docker deployment
└── package.json            Root workspace configuration

Quick Start

# 1. Install dependencies
corepack pnpm install

# 2. Type-check everything
corepack pnpm typecheck

# 3. Run tests
corepack pnpm test

# 4. Build all packages and apps
corepack pnpm build

# 5. Start the extension in dev mode
corepack pnpm dev:ext

# 6. (Optional) Start the archive server
corepack pnpm dev:server

Installation & Setup

Prerequisites

Requirement	Version	Notes
Node.js	>= 22.0.0	Required for ESM and modern APIs
pnpm	>= 10.0.0	Managed via Corepack (`corepack enable`)
Browser	Chromium	Edge, Chrome, Brave, etc.
OS	Windows	Recommended; native host is Windows-only

Build from Source

# Clone the repository
git clone https://github.com/Alkapuce/AIexporter.git
cd aiexporter

# Enable Corepack (if not already)
corepack enable

# Install dependencies
corepack pnpm install

# Build everything
corepack pnpm build

Load the Extension

Open your Chromium browser and navigate to chrome://extensions (or edge://extensions).
Enable Developer mode (toggle in the top-right corner).
Click Load unpacked and select the apps/extension/.output/chrome-mv3 directory.
The AIexporter icon should appear in your toolbar.

Tip: For development, use corepack pnpm dev:ext to enable hot-reload. The extension rebuilds automatically when you edit source files.

Register Native Host (Windows)

The native host enables direct filesystem operations — custom export directory, open/reveal/recycle files, and OneDrive Downloads path resolution.

# Register the native host (replace <your-extension-id> with the actual ID from chrome://extensions)
pwsh -File apps/extension/scripts/register-native-host.ps1 -ExtensionId "<your-extension-id>"

Without the native host, the extension falls back to the browser Downloads API.

Run the Archive Server (Optional)

The server provides a searchable, queryable archive of all exported conversations.

# Development mode (with hot reload)
corepack pnpm dev:server
# → http://127.0.0.1:8787

# Production build & run
corepack pnpm --filter @aiexporter/server build
node apps/server/dist/index.js

Environment variables:

Variable	Default	Description
`AIEXPORTER_HOST`	`127.0.0.1`	Server bind address
`AIEXPORTER_PORT`	`8787`	Server port
`AIEXPORTER_DATA_DIR`	`./data`	SQLite & archive directory

Docker Deployment

docker compose up -d
# Server available at http://127.0.0.1:8787

The data/ directory is mounted as a volume for persistence.

How It Works

┌─────────────────────────────────────────────────────────┐
│                    AI Platform Page                       │
│  ┌──────────────┐  ┌──────────────┐  ┌───────────────┐  │
│  │ Content Script│  │ Main-World   │  │ Page Network  │  │
│  │ (DOM observe) │  │ Bridge       │  │ Interception  │  │
│  └──────┬───────┘  └──────┬───────┘  └───────┬───────┘  │
│         │                 │                   │          │
└─────────┼─────────────────┼───────────────────┼──────────┘
          │                 │                   │
          ▼                 ▼                   ▼
┌─────────────────────────────────────────────────────────┐
│              Background Service Worker                    │
│  ┌──────────┐  ┌──────────┐  ┌───────────────────────┐  │
│  │ Discovery│──▶  Queue   │──▶  Worker Tab Scheduler │  │
│  │ Events   │  │ Dedup    │  │                       │  │
│  └──────────┘  └──────────┘  └───────────┬───────────┘  │
│                                          │              │
│  ┌───────────────────────────────────────┘              │
│  │  ┌──────────────┐  ┌──────────────┐                 │
│  │  │ Extraction   │──▶ Persistence  │                 │
│  │  │ (via content │  │ (Markdown +   │                 │
│  │  │  script)     │  │  JSON bundle) │                 │
│  │  └──────────────┘  └──────┬───────┘                 │
│  │                           │                          │
└──┼───────────────────────────┼──────────────────────────┘
   │                           │
   ▼                           ▼
┌──────────────────┐   ┌──────────────────┐
│  Export Directory │   │  Archive Server   │
│  (local disk)    │   │  (Fastify+SQLite) │
└──────────────────┘   └──────────────────┘

Pipeline Steps

Detection: Content scripts observe supported conversation pages.
Discovery: Main-world bridges intercept network responses to discover conversation IDs and metadata.
Queue: The background service worker deduplicates discovery events into a per-platform queue.
Extraction: Worker tabs load conversations and content scripts extract structured data (messages, thinking blocks, attachments).
Serialization: Bundles are serialized to Markdown (with YAML frontmatter) and JSON.
Persistence: Artifacts are written to the configured export directory via native host or Downloads API.
Sync (optional): Bundles are POSTed to the local Fastify archive service.

Architecture

Extension Pipeline

The extension uses WXT (Web eXtension Toolkit) with Manifest V3:

Content Scripts (entrypoints/*.content.ts): Run in the page context of each supported AI platform. They observe the DOM and respond to extraction requests from the background.
Main-World Bridges (entrypoints/*-main-world.ts): Injected into the page's main JavaScript world to intercept network requests and access in-page state.
Background Service Worker (entrypoints/background.ts): Central orchestrator that manages the queue, schedules worker tabs, and coordinates persistence.

Background Modules

Module	Responsibility
`runtime-router.ts`	Message routing between background, content scripts, dashboard
`service-runtime.ts`	Queue orchestration, worker scheduling, platform ticks
`tab-runtime.ts`	Worker tab lifecycle, content script readiness detection
`artifact-persistence.ts`	File writes, pruning, export root resolution, remote asset fetch
`artifact-sync.ts`	Artifact verification and re-import from disk
`artifact-sync-state.ts`	Sync state tracking
`automatic-artifact-sync.ts`	Scheduled background sync on browser startup
`state-access.ts`	Queue state normalization and derived updates
`export-root.ts`	Configurable export directory support
`shared.ts`	Shared constants and archive naming

Platform Adapters

Each supported platform has:

Adapter package (packages/adapters-*/): Discovery, DOM extraction, and turndown configuration.
Content script (entrypoints/*.content.ts): Injected into the platform page.
Main-world bridge (entrypoints/*-main-world.ts): Network interception and RPC calls.

Platform	Discovery Method	Extraction Method
ChatGPT	Intercept `/backend-api/conversation/{id}`	Page-world API response parsing
DeepSeek	Intercept `/api/v0/chat_session/fetch_page`	Page-world history messages API
Gemini	Intercept `batchexecute` RPC	RPC payload recursive traversal + DOM fallback
AI Studio	Library page `ListPrompts` + DOM sweep	gRPC-Web `ResolveDriveResource` + DOM

Archive Format

Exports follow a consistent, human-readable directory structure:

<export-root>/
└── AIexporter/
    └── <platform>/               # chatgpt, deepseek, gemini, aistudio
        └── <conversation-folder>/ # Title + source ID suffix
            └── <revision>/        # Content hash (first 12 hex chars)
                ├── <artifact>.md           # Markdown with YAML frontmatter
                ├── <artifact>.bundle.json  # Full structured data
                └── <artifact>.assets/      # Downloaded images & attachments
                    ├── 00-image.jpg
                    └── ...

Markdown frontmatter example:

---
aiexporter: 2026-04-15.2
platform: gemini
conversation_id: abc123
title: Building a REST API
source_url: https://gemini.google.com/app/abc123
source_updated_at: 2026-04-10T12:00:00.000Z
exported_at: 2026-04-10T12:01:30.000Z
message_count: 24
revision: f7a3b2c1d4e5
---

The .bundle.json contains the complete structured data including all messages, thinking blocks, attachments, and metadata — suitable for programmatic processing.

Native Host (Windows)

The optional Windows native messaging host (apps/extension/native-host/) communicates with the extension via stdin/stdout JSON messages. It enables:

Capability	Description
Custom export directory	Pick any local folder via native folder picker dialog
Open files	Open exported `.md` with system default handler
Reveal in Explorer	Open the export folder in File Explorer
Move to Recycle Bin	Soft-delete exported files
OneDrive path resolution	Resolve OneDrive-redirected Downloads paths

Registration:

pwsh -File apps/extension/scripts/register-native-host.ps1 -ExtensionId "<your-extension-id>"

When the native host is unavailable, the extension automatically falls back to the browser Downloads API.

Server API

The Fastify archive service provides a RESTful API for ingesting and querying exported conversations.

Method	Endpoint	Description
`POST`	`/api/v1/ingest/conversations`	Ingest a conversation bundle
`GET`	`/api/v1/conversations`	List/search conversations
`GET`	`/api/v1/conversations/:platform/:sourceId`	Get a specific conversation
`GET`	`/healthz`	Health check

Query parameters for GET /api/v1/conversations:

Parameter	Type	Description
`platform`	string	Filter by platform (`chatgpt`, `gemini`, etc.)
`q`	string	Full-text search (title + source ID)
`cursor`	string	Pagination cursor
`limit`	number	Results per page (default: 50)

Dashboard

The built-in React dashboard is accessible from the browser toolbar. It provides three tabs:

Queue — View pending, processing, completed, and failed items per platform. Pause/resume individual platform services.
Logs — Real-time background debug logs with filtering and search.
Settings — Configure export root, discovery intervals, platform-specific tuning, locale (en / zh-CN), theme, and server sync.

Configuration Guide

All settings are accessible from the Dashboard → Settings tab, and stored in extension storage (browser.storage.local).

Export Settings

Setting	Description
Export Root Path	Custom directory for all exports (requires native host)
Revision History Mode	How to handle previous exports: `disabled`, `recycle_previous`, `archive_then_recycle`
Retain Local Revision Count	Number of old revisions to keep on disk
Open File Actions	Enable "Open Markdown" and "Show in Folder" buttons

Discovery Settings (per platform)

Setting	Description
Enabled	Enable/disable this platform
Auto-Export	Automatically export newly discovered conversations
History Backfill	Sweep the conversation history sidebar/library page
Discovery Mode	`passive_only` (intercept network) or active DOM sweep
Max Concurrency	Maximum simultaneous worker tabs for this platform

Server Sync

Setting	Description
Sync to Server	POST bundles to the local archive server
Server URL	Archive server URL (locked to localhost)

Development

Project Scripts

Command	Description
`corepack pnpm install`	Install all workspace dependencies
`corepack pnpm build`	Build all packages and apps
`corepack pnpm dev:ext`	Run extension in WXT dev mode (hot reload)
`corepack pnpm dev:server`	Start Fastify server with hot reload
`corepack pnpm test`	Run all Vitest test suites
`corepack pnpm typecheck`	Type-check all workspaces
`corepack pnpm format`	Format with Prettier
`corepack pnpm --filter <pkg> test`	Run tests for a specific package
`corepack pnpm --filter @aiexporter/extension build`	Build only the extension

Testing

Unit tests: Vitest, co-located with source (*.test.ts) or in test/ directories.
Browser testing: See docs/browser-live-testing.md for CDP/Playwright instructions.
Extension smoke test: apps/extension/scripts/verify-dashboard-file-actions.cjs validates the dashboard and native host integration.

Code Style

TypeScript ESM throughout
2-space indentation, semicolons, double quotes, trailing commas
React components: PascalCase files
Utilities/services: kebab-case filenames, camelCase symbols
Prettier for formatting; run corepack pnpm format before committing

Troubleshooting

Symptom	Likely Cause	Solution
Extension doesn't detect conversations	Not logged into the AI platform	Log into the target platform in your browser
"Native host not available" warning	Native host not registered	Run the registration script (see Native Host)
Gemini/AI Studio discovery incomplete	Google rate limiting / reCAPTCHA	Reduce concurrency, pause other platforms, try again later
Exported markdown is empty or incomplete	Extraction timed out	Increase `navigationTimeoutMs` in platform settings
DeepSeek discovery sidebar not ready	Sidebar not open or page still loading	Ensure the DeepSeek sidebar is visible; try refreshing
Dashboard white screen	JavaScript error in React render	Check browser console; report the error with logs
Server connection refused	Server not running	Run `corepack pnpm dev:server`
Download was interrupted (SERVER_FAILED / SERVER_BAD_CONTENT)	Network issue or bad remote asset	The item will be retried; check network connectivity

Limitations

Google platform discovery depends on what the logged-in UI exposes during DOM sweeps. Gemini's history sidebar has a known cap (~587 conversations).
Native host is Windows-only. On other platforms, the extension uses the Downloads API and cannot customize the export directory.
Background scheduling requires the browser to be online and authenticated into the target AI platforms.
No dedicated archive browsing UI yet — the server focuses on ingestion and query. Use your favorite Markdown editor to browse exports.
DeepSeek attachment files are recorded as metadata (name + size) but cannot be downloaded automatically (requires per-file signed URLs).
AI Studio Drive resources are stored as links requiring Google account permissions.

Contributing

See CONTRIBUTING.md for development setup, code style, and pull request guidelines.

License

MIT. See LICENSE for details.

Name		Name	Last commit message	Last commit date
apps		apps
docs		docs
packages		packages
.gitignore		.gitignore
.prettierrc		.prettierrc
AGENTS.md		AGENTS.md
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md
README.zh-CN.md		README.zh-CN.md
docker-compose.yml		docker-compose.yml
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
pnpm-workspace.yaml		pnpm-workspace.yaml
tsconfig.base.json		tsconfig.base.json

Latest commit

History

Folders and files

Repository files navigation

AIexporter

Table of Contents

What it does

Supported Platforms

Workspace Layout

Quick Start

Installation & Setup

Prerequisites

Build from Source

Load the Extension

Register Native Host (Windows)

Run the Archive Server (Optional)

Docker Deployment

How It Works

Pipeline Steps

Architecture

Extension Pipeline

Background Modules

Platform Adapters

Archive Format

Native Host (Windows)

Server API

Dashboard

Configuration Guide

Export Settings

Discovery Settings (per platform)

Server Sync

Development

Project Scripts

Testing

Code Style

Troubleshooting

Limitations

Contributing

License

About

Resources

Contributing

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages