Who We Are

ContextSwitchAI is a free Chrome browser extension. It has no corporate entity, no paid tier, no subscription, and no business model that depends on your data. The extension is developed and maintained as a project and distributed through the Chrome Web Store at no cost.

This privacy policy describes how the extension handles information when you use it.

What Data We Collect

We do not collect any personal data.

The extension does not collect, store, transmit, or share any personal information, usage data, browsing history, conversation content, or any other data. There is no analytics integration, no crash reporting service, no remote logging of any kind.

The extension has no backend server, no database, and no network requests of its own. Every operation — parsing conversations, compressing text, building the JSON export, attaching files to AI chat inputs — happens entirely inside your browser process and never leaves your device.

How Your Conversation Data Is Handled

When you click "Export Chat Memory," the extension reads the text content of the current AI chat page using the browser's DOM APIs. This content is:

• Processed entirely in memory within your browser tab
• Compressed and formatted into a CCMF JSON structure
• Saved to your local Downloads folder via the Chrome Downloads API
• Optionally stored temporarily in chrome.storage.local so it can be automatically attached when you navigate to another AI platform in the same session

This locally stored data never leaves your browser. It is not synced via chrome.storage.sync. It is not sent to any remote endpoint.

You can clear this data at any time by clicking the reset button (↺) in the extension popup, or by clearing your browser's extension storage in chrome://settings.

Permissions Explained

The extension requests the following Chrome permissions. Each one is used for a specific, limited purpose:

No permission grants access to your browsing history, bookmarks, passwords, cookies, or any other browser data. The extension only ever reads content from the specific AI chat pages it is explicitly authorised to access.

Third-Party Services

The extension does not use, load, or communicate with any third-party services, APIs, analytics platforms, advertising networks, or CDNs at runtime. All code that executes is bundled within the extension package itself.

The extension interacts with the AI platforms you visit (ChatGPT, Claude, Gemini, etc.) only insofar as it reads their page content and attaches files to their chat input. This interaction is subject to each platform's own privacy policy and terms of service. ContextSwitchAI has no affiliation with any of these platforms.

Children's Privacy

This extension is not directed at children under the age of 13. We do not knowingly collect any information from children. Since the extension collects no information from anyone, there is no special handling required for minors.

Data Security

Since no data leaves your device, the security of your conversation exports is entirely within your control. The exported JSON files are stored in your local Downloads folder and are subject to the same security as all other files on your computer.

We recommend treating exported conversation files with the same care you would apply to any sensitive document, particularly if the conversations contain personal, financial, or confidential information.

Changes to This Policy

If we make material changes to this privacy policy — for example, if a future version of the extension adds any form of network functionality — we will update the "Last updated" date above and increment the extension version. We will not introduce data collection without first updating this policy and making the change visible in the Chrome Web Store listing.

The current version of this policy always applies to the current version of the extension.

Your Rights

Because we collect no personal data, there is no data to access, correct, export, or delete. If you wish to remove all locally stored extension data from your browser, you can do so by:

• Clicking the reset button (↺) in the extension popup to clear the current session's exported data
• Going to chrome://settings/siteData and clearing storage for the extension
• Uninstalling the extension, which removes all associated local storage

Contact

If you have any questions about this privacy policy or the extension's data practices, you can reach us at:

📦 Version History

⬆ Export Not Working

The extension only works on supported AI pages. Make sure you're on one of: ChatGPT, Claude, Gemini, Grok, DeepSeek, Perplexity, Microsoft Copilot, Quark AI, or Doubao.

Common fixes:

• Reload the page — then click the extension icon again. Some AI pages load dynamically and the extension needs to re-inject.
• Scroll through the conversation first — lazy-loaded messages may not be captured if you haven't scrolled to them.
• Check Downloads folder permissions — make sure your browser has permission to save files to Downloads.
• Try a different format — if JSON fails, try MD. This can reveal whether the issue is with compression or file writing.

🔀 Auto-Inject Not Working

Context injection works by opening the target AI platform and automating the file attachment. If it fails:

• Export first — you must export before switching. The extension loads the most recent export file.
• Allow popups — your browser may be blocking the new tab. Check the address bar for a blocked popup icon.
• Platform UI changes — AI platforms update frequently. If injection broke recently, let us know and we'll push a fix quickly.
• File size limit — very long conversations compressed to JSON can be large. Try using MD format or the Select feature to export fewer messages.

🧠 Smart Context Search Issues

Smart Context Search uses an in-browser semantic embedding model to index and search all your saved exports by meaning — not just keywords. Here's how to set it up and fix common problems:

🔗 First-time setup

• Open the extension and scroll to the Smart Context Search card at the bottom of the popup.
• Click "📁 Link Folder" and select your Downloads/ContextSwitchAI/ folder (or wherever your exports are saved). You only need to grant this permission once — it persists across sessions.
• Wait for the first scan — the extension will read all JSON files in the folder and build a local semantic index. This may take 3–10 seconds depending on how many files you have. You'll see the count update in the badge.

🔍 Search tips

• Search by topic, not filename — the engine understands intent. Try phrases like "that React hooks memory leak", "marketing email campaign", or "database schema design".
• Use filters — narrow results by AI platform (ChatGPT, Claude, etc.), date range, or whether the export contains code blocks.
• Check the relevance score — results are ranked 0–100%. Anything above 60% is usually a strong match. A low top score means the topic may not be in your library yet.
• Load and inject — click a result card to select it (it gets a ✓ checkmark), then hit "🚀 Load into [AI]" to auto-inject it into a new chat on your chosen platform.

⚠️ Common problems & fixes

• No results / empty index — your folder may not be linked yet, or the folder contains no .json files. Make sure you've exported at least one conversation in JSON format first.
• New export not appearing — click the ↻ Refresh button in the Smart Context Search panel. The index doesn't watch the folder in real-time; a manual refresh picks up new files.
• Folder permission lost after browser restart — some Chrome builds revoke folder access between sessions. Simply click "📁 Link Folder" again and re-select the same folder — your existing index is preserved.
• "Indexing…" spinner stuck — this can happen if a JSON file is corrupt or very large. Try clicking ↻ Refresh. If it persists, delete the problematic file from the folder and re-scan.
• Wrong folder linked — click ↻ and then "📁 Link Folder" again to select a different directory. The index will rebuild from the new location.
• Search returns unrelated results — the semantic model runs locally in your browser (no internet required) and works best with CCMF JSON exports. CSV and MD exports are indexed as plain text and may match less precisely.

✉️ Get in Touch

If none of the above helped, we'd love to hear from you. The fastest way to reach us is by email — we typically respond within 24 hours.