Skip to content
Codex Agent edited this page Jun 2, 2026 · 1 revision

Quick Start 🚀

This page gets you from installation to your first AI-ready zip archive for ChatGPT, Claude, Cursor, VS Code, or another LLM coding workflow.


1. Install Export2AI

Method Use when Steps
Open VSX / Cursor normal installation Open Extensions, search Export2AI, install avnsx.export2ai.
GitHub Release VSIX marketplace is unavailable or you want a specific version Download export2ai-x.y.z.vsix from Releases, then install from VSIX.
Local build contributor workflow Run npm install, npm run compile, npm run package, then install build/export2ai-x.y.z.vsix.

Requirements for running the extension: VS Code / Cursor ^1.105.0.

Requirements for building from source: Node.js 24+ is recommended to match the release workflow.


2. Create an AI Context Zip

  1. Open a project folder in VS Code or Cursor.
  2. In the Explorer, right-click the folder you want to export.
  3. Choose Export2AI → Zip Folder.
  4. Wait for the progress notification to finish scanning and writing.
  5. Upload the generated zip to ChatGPT, Claude, Cursor, or your AI coding tool.

The zip is written to the workspace root:

{folder}-{model}-context-{YYYY-MM-DD-HHMMSS}.zip

Example:

backend-gpt-5.5-context-2026-06-02-201700.zip

The filename uses:

Segment Source
backend selected folder basename, sanitized and capped
gpt-5.5 export2ai.llmModel
context stable marker for AI handoff archives
timestamp compact local creation time

3. Check the Token Estimate

When token counting is enabled, Export2AI shows an estimate in:

Location Example
status bar gpt-5.5 · (est. 47,382 tokens)
zip notification Export2AI created for gpt-5.5: ... (est. 47,382 tokens)
optional Explorer folder badges compact 2-character folder badges when explicitly enabled

A ~ prefix means approximate. No ~ means an exact offline tokenizer was used for that model family.

More detail: Token Estimates & AI Model Support.


4. Upload the Zip to an AI Tool

Use the archive as a project snapshot. Good prompts usually state what the archive represents:

I uploaded an Export2AI project context zip. Please inspect the codebase, docs, tests,
and CI files first. Treat _EXPORT2AI_MANIFEST.txt as archive metadata, not source code.
Do not assume omitted .git internals or credentials are part of the project.

For coding agents, also tell them to read AGENTS.md when it is present.


5. Common First-Time Settings

Open Export2AI → Settings from the menu, Command Palette, or by clicking the status bar token estimate.

Goal Setting
Change target AI model export2ai.llmModel
Disable token counting export2ai.enableTokenCounting
Keep token badges out of Explorer leave export2ai.showExplorerTokenBadges as false
Add project-specific excludes export2ai.excludePatterns or export2ai.excludePaths
Preserve all code comments leave export2ai.removeComments as false
Preserve formatting leave export2ai.compressCode as false

Full reference: Settings & Configuration.


Quick Sanity Check

After creating a zip, open it once and verify:

  • _EXPORT2AI_MANIFEST.txt exists unless you disabled it.
  • node_modules/, dist/, build/, .env*, private keys, and .git/ internals are absent.
  • .github/, .gitignore, docs, tests, and AGENTS.md are present when they exist in the project.
  • Large or binary files are represented by short placeholders, not raw binary content.
  • The filename includes the target model you intended to use.

Clone this wiki locally