Inject prompts into major AI web chat boxes through URL parameters, with optional auto-send.
中文文档: README.zh-CN.md
Universal AI URL Prompt is a browser extension for reproducible AI workflows.
You can share a single URL and let the target AI page open with a prefilled prompt, then auto-submit if needed.
- URL-based prompt injection with
?prompt=,?q=, and?p= - Auto-send enabled by default, with
?autosend=0to disable - Per-platform enable/disable switches in the extension Options page
- Send-button detection for English, Chinese, Japanese, and Spanish
- SPA-aware retry and scheduling for delayed input rendering
- Platform-isolated configs with a shared runtime for easier maintenance
- ChatGPT, Gemini, Claude, DeepSeek, Microsoft Copilot
- Doubao, Perplexity, Kimi, Meta AI, Tencent Yuanbao
- Qwen, Grok, Wenxiaoyan, Poe, Mistral Le Chat
- ChatGLM, Baichuan, DuckDuckGo AI, HuggingChat, Z.ai
| Parameter | Purpose | Example |
|---|---|---|
prompt |
Primary prompt field | ?prompt=Write%20a%20plan |
q |
Prompt alias for some sites | ?q=Explain%20RAG |
p |
Prompt alias used by some sites (such as Kimi) | ?p=%E5%AE%89%E5%8D%93 |
autosend=0 |
Disable auto-send | ?prompt=Draft%20PRD&autosend=0 |
- ChatGPT:
https://chatgpt.com/?prompt=Write%20a%20plan - Gemini:
https://gemini.google.com/app?prompt=你好 - DeepSeek:
https://chat.deepseek.com/?q=Explain%20RAG - DeepSeek (new domain):
https://www.deepseek.com/?prompt=Explain%20RAG - Kimi:
https://www.kimi.com/?p=%E5%AE%89%E5%8D%93 - HuggingChat:
https://huggingface.co/chat/?prompt=Write%20a%20plan - Claude without auto-send:
https://claude.ai/?prompt=Draft%20PRD&autosend=0
Coming soon.
- Open
chrome://extensions/ - Enable Developer mode
- Click
Load unpacked - Select this project directory
- (Optional) Open extension
Details->Extension optionsto choose enabled platforms
Note: The GitHub Release zip can be used for Developer Mode installation (extract it, then use Load unpacked). Regular users should install from the Chrome Web Store.
npm install
npx playwright install chromium
npm run check:platforms
npm run build:manifest
npm run test:e2eAdditional commands:
npm run test:e2e:headed: run tests in headed modenpm run test:e2e:ui: run tests in Playwright UI mode
platforms/_registry.js: platform registration and schema validationplatforms/*.js: one platform definition per filecontent.js: shared runtime (URL parsing, injection, auto-send flow)scripts/platform-loader.js: shared platform-definition loaderscripts/generate-manifest.js: generatemanifest.jsonfrom platform filesscripts/validate-platform-contract.js: validate platform definitions and E2E mappings
This extension does not send data to developer-owned servers.
It keeps a short-lived in-memory prompt cache (up to about 3 minutes) in the extension runtime for SPA navigation handoff.
It stores platform on/off preferences in Chrome extension storage (chrome.storage.sync or local fallback) so your choices persist.
When auto-send is enabled, prompt content is sent only to the destination AI website as part of normal page usage.
Issues and pull requests are welcome.
Before submitting changes, run npm run check:platforms and npm run test:e2e.
See CONTRIBUTING.md, CODE_OF_CONDUCT.md, and SECURITY.md for project policies.
Inspired by gemini-url-prompt.