See in the print how much screen I have available, your plugin is unique, but it could be improved just by increasing the response box, not to mention that you shouldn't push the text up if you get to the bottom since you didn't even have time to read the first line when he concludes.

:D

Hi @CocoSgt,

Thank you for the excellent update, which is working perfectly. However, I've encountered an issue with the extension not saving larger prompts or text in context.

I've noticed that when I create a new prompt with larger text, it appears to be created and saved successfully. However, upon refreshing the page or opening a new one, the extension loses or doesn't retain that particular prompt.

Could you please provide guidance on how to create bigger or larger text prompts, or how to increase the extension's memory to accommodate larger prompt data?

Thank you.

Originally posted by @webmasterfsd in #35 (comment)

Currently the chrome plugin only supports gpt3.5, I am used to using gpt4, can you guys do it? When I fill in my own key, I hope to be able to use gpt4

Hi I have been using your tool with ChatGPT to create unique contents but all manual using some custom prompt. I am keen on automating this task where webpilot extracts H2s from top 10 serp pages and then chatGPT will generate contents from it.

Is this possible and if so pls tell me what to so.

For this link： https://www.toutiao.com/article/7256707596743082530/?log_from=5f74c06a18593_1689644495307

It got the content correctly, but the summary and content are completely irrelevant. The same link and Prompt tried several times, and each time the content was different and looked "random".

Try again：

does anyone encounter the below issue?

D:\python_project_2\Webpilot>yarn dev
yarn run v1.22.19
$ plasmo dev
file:///D:/python_project_2/Webpilot/node_modules/plasmo/dist/index.js:74
Visit https://docs.plasmo.com for documentation and more examples.)}});var qi={};Ie(qi,{default:()=>vm});import{paramCase as hm}from"change-case";import{resolve as gm}from"path";import{cwd as ym}from"process";async function wm(){re();let t=q("--exp"),e=await Di(),r=ym(),n=gm(r,hm(e)||e);m("Project directory:",n);let o=je(n);if(m("Package name:",o.packageName),t&&!o.packageName.startsWith("with-"))throw new Error("Example extensions must have the with- prefix");await _n(n);let i=await Mt();m(Using package manager: ${i.name} ${i?.version}),await new Bt(o,i,t).create(),await Ti(n,i),await Ni(o,n),await ji(n,r,o,i)}var vm,Wi=d(()=>{pe();V();b();Ct();xr();De();Ii();Oi();$i();Ai();Ri();vm=wm});import{WebSocket as Pm,WebSocketServer as xm}from"ws";var Sm,Bi,Ui=d(()=>{Sm=JSON.stringify({type:"build_ready"}),Bi=async t=>{let e=new xm({port:t+1});return{triggerReload:()=>{for(let n of e.clients)n.readyState===Pm.OPEN&&n.send(Sm)}}}});var He,Vt=d(()=>{Ae();He=()=>{let t=Pe(),{target:e,tag:r}=t,[n,o]=e.split("-");return{tag:r,target:e,browser:n,manifestVersion:o}}});import{AssertionError as Em}from"assert";function Je(t){throw new Error("Statement should be unreachable")}function Ur(t,e="Value is invalid"){if(!t)throw new Em({message:e});return t}var lt=d(()=>{});function Lm(t){switch(t){case 0:return"Web";case 1:case 2:case 3:return Gi[t];default:Je(t)}}var Hr,Jt,Jr,ft,bm,_m,Vi,km,Cm,Mm,Vr,Gr,Ce,Me,X,Hi,Ji,Xd,Gi,Ht,Zd,eh,zi,Ki,th,le,ke,Fm,rh,Im,nh,oh,ih,sh,Yi=d(()=>{lt();Hr="en",Jt=!1,Jr=!1,ft=!1,bm=!1,_m=!1,Vi=!1,km=!1,Cm=!1,Mm=!1,Gr=Hr,Me=typeof self=="object"?self:typeof global=="object"?global:{};typeof Me.vscode<"u"&&typeof Me.vscode.process<"u"?X=Me.vscode.process:typeof process<"u"&&(X=process);Hi=typeof X?.versions?.electron=="string",Ji=Hi&&X?.type==="renderer",Xd=Ji&&X?.sandboxed;typeof navigator=="object"&&!Ji?(Ce=navigator.userAgent,Jt=Ce.indexOf("Windows")>=0,Jr=Ce.indexOf("Macintosh")>=0,Cm=(Ce.indexOf("Macintosh")>=0||Ce.indexOf("iPad")>=0||Ce.indexOf("iPhone")>=0)&&!!navigator.maxTouchPoints&&navigator.maxTouchPoints>0,ft=Ce.indexOf("Linux")>=0,Vi=!0,Vr=navigator.language,Gr=Vr):typeof X=="object"?(Jt=X.platform==="win32",Jr=X.platform==="darwin",ft=X.platform==="linux",bm=ft&&!!X.env.SNAP&&!!X.env.SNAP_REVISION,km=Hi,Mm=!!X.env.CI||!!X.env.BUILD_ARTIFACTSTAGINGDIRECTORY,Vr=Hr,Gr=Hr,_m=!0):console.error("Unable to resolve platform.");Gi={[1]:"Mac",[2]:"Linux",[3]:"Windows"};Ht=0;Jr?Ht=1:Jt?Ht=3:ft&&(Ht=2);Zd=Object.values(Gi),eh=Lm(Ht),zi=Jt,Ki=ft,th=Vi&&typeof Me.importScripts=="function",le=Ce,ke=Gr;(n=>{function t(){return ke}n.value=t;function e(){return ke.length===2?ke==="en":ke.length>=3?ke[0]==="e"&&ke[1]==="n"&&ke[2]==="-":!1}n.isDefaultVariant=e;function r(){return ke==="en"}n.isDefault=r})(Fm||={});rh=(()=>{if(typeof Me.postMessage=="function"&&!Me.importScripts){let t=[];Me.addEventListener("message",r=>{if(r.data&&r.data.vscodeScheduleAsyncWork)for(let n=0,o=t.length;n<o;n++){let i=t[n];if(i.id===r.data.vscodeScheduleAsyncWork){t.splice(n,1),i.callback();return}}});let e=0;return r=>{let n=++e;t.push({id:n,callback:r}),Me.postMessage({vscodeScheduleAsyncWork:n},"*")}}return t=>setTimeout(t)})(),Im=!!(le&&le.indexOf("Chrome")>=0),nh=!!(le&&le.indexOf("Firefox")>=0),oh=!!(!Im&&le&&le.indexOf("Safari")>=0),ih=!!(le&&le.indexOf("Edg/")>=0),sh=!!(le&&le.indexOf("Android")>=0)});var Qi,Xi=d(()=>{Yi();Qi=zi?"windows":Ki?"inotify":"fs-events"});async function ge(t,e){for(let r of t)if(await e(r))return r}var dt=d(()=>{});import{basename as Dm,extname as Zi,posix as Om,win32 as Nm}from"path";var Z,es,Ge=d(()=>{Z=t=>t.replaceAll(Nm.sep,Om.sep),es=t=>Zi(Dm(t,Zi(t)))});import{relative as $m,resolve as ts}from"path";var Gt,rs,ns=d(()=>{Gt=U(B(),1);b();Ge();rs=async({indexFilePath:t="",withMessaging:e=!1,withMainWorldScript:r=!1},n)=>{m("Creating BGSW entry");let o=ts(n.commonPath.staticDirectory,"background"),i=ts(o,"index.ts"),s=$m(o,t),a=[e&&'import "./messaging"',t&&import "${Z(s).slice(0,-3)}",r&&'import "./main-world-scripts"'].filter(Boolean).join(

SyntaxError: missing ) after argument list
at Loader.moduleStrategy (internal/modules/esm/translators.js:141:18)
error Command failed with exit code 1.
info Visit https://yarnpkg.com/en/docs/cli/run for documentation about this command.

This browser extension seems to be very great, just curious, is there any plan to port it to Firefox?

Thanks.

https://vant-ui.github.io/vant/#/zh-CN/list

When I close the Webpilot plugin in Edge browser, it is normal. When it is opened, it will display the problem in the image.

A button that will toggle the theme to dark mode from light mode and vice versa!

I misunderstood the "turbo mode" to use the better version of ChatGPT (the 3.5 for plus users), but it is not, it is for popping up the window once you select something.
Maybe we can change a name of the mode to clarify this.

options.html:1 Uncaught (in promise) {}[[Prototype]]: Objectconstructor: ƒ Object()hasOwnProperty: ƒ hasOwnProperty()isPrototypeOf: ƒ isPrototypeOf()propertyIsEnumerable: ƒ propertyIsEnumerable()toLocaleString: ƒ toLocaleString()toString: ƒ toString()valueOf: ƒ valueOf()defineGetter: ƒ defineGetter()defineSetter: ƒ defineSetter()lookupGetter: ƒ lookupGetter()lookupSetter: ƒ lookupSetter()proto: (...)get proto: ƒ proto()set proto: ƒ proto()
Promise.then (async)
setup @ options.2a742e8f.js:1
p @ options.2a742e8f.js:1
(anonymous) @ options.2a742e8f.js:1
rF @ options.2a742e8f.js:1
W @ options.2a742e8f.js:1
Y @ options.2a742e8f.js:1
B @ options.2a742e8f.js:1
ec @ options.2a742e8f.js:1
mount @ options.2a742e8f.js:1
t.mount @ options.2a742e8f.js:1
(anonymous) @ options.2a742e8f.js:1

Hey there,

We are absolutely thrilled with your plugin and highly enthusiastic about the prospect of collaborating with you for some exciting new projects. Would it be possible for us to get in touch with you? Please let me know how we can do so.

My information is as follows:
Mail: [email protected]

Thank you so much!

After entering content in the webpilot input box and performing a search, when I close webpilot and reopen it, the old input box content remains, but the response content is lost. I expect one of the following two interactions:

1. When reopened, the input box content and the corresponding response content are retained.
2. When reopened, all content is cleared.
   I lean towards the second option because when I close webpilot, it means I no longer need the current content. Therefore, when I reopen it, I would prefer to see a reset input box.

Please evaluate whether there is a need to optimize this interaction. If this interaction needs to be optimized, I'm more than happy to do it.

Thank you!

I am currently requesting a list of cons to be retrieved from a web page as follows:

{
"content": "Please follow the link to the product reviews and extract me a list of cons https://www.trustradius.com/products/onecause/reviews?qs=pros-and-cons#reviews"
}

The results are always good quality, just variable. I've returned lists of between 14 and 27 cons.

Sometimes this variability can be useful but in this case I need a fixed out for my downstream completions. Is there a way to fix the temperature to retain consistency?

I'm using the API btw.

输入

全选

请问如何解决？谢谢！

Currently the plugin prevents access to https://webreader.webpilotai.com/openapi.yaml for unauthorized IPs. If you removed this requirement us in the Langchain Opensource community will be able to incorporate the plugin in some cool projects!

How to disable Fluentify temporarily? Now whenever I select texts, Fluentify will work on it.
But I just want to use it when I really need to, not every time I select texts.

Please add a privacy policy

Hello,

I would like to request a new button in the question box, he would be responsible for reading the page and already suggesting questions that I could ask for that page.

For example:

1. I visit the article https://www.washingtonpost.com/world/2023/09/04/putin-kim-meeting-weapons-deal-russia-north-korea/
2. I press the button and your extension asks GPT: What questions do you suggest I ask for this page?
3. The GPT answer must be in threads without explanations, only the suggested questions
4. Which should be interpreted as Shortcuts to the next question
5. Somehow it would be nice to keep the suggestions so that the user can continue to explore.

That it :D

Thx so much!

试了几次都没有给出正确的结果

https://chat.openai.com/share/5dabde35-d414-4a24-b6f6-a55bd102cc8e

目前markdownWrap块显示空间太小了，如果能自定义height和width就太好了。

Like below/如下图：

The web page/网页地址: https://blog.csdn.net/Javon_huang/article/details/122252177.

Disable this extension is OK/关掉这个扩展就没问题了.

Windows平台，edge浏览器，最新插件版本，输入了自己的key和自托管host之后，当时能用，关闭浏览器重新打开，key没了。
请问是什么原因？

Hello everyone,
This is not an issue but I couldn't fine other places for this post, so...
Please forgive me (non-tech guy) to ask such a layperson question. I really like the great search capability of Webpilot. Can anyone show me where (which file) to see? (I checked most, but I could find out the answer...)

I have reviewed the website, and it explained to uncheck Web Browsing and instead use WebPilot as an action, and there is no compelling reason or reason at all on why there are benefits to doing this. Would love that to be added, or even a response here with some bullet points possibly (?) on what benefits exist.

eg;

• why is it better than native browsing feature
• does it provide any unique capabilities
• does it skirt requirements or limits that restrict native web browsing functionality

Thanks!

Thank you very much to the developers for creating this excellent tool. During my usage, I found that the message box generated by AI cannot be resized. I'm wondering if there are any plans to support resizing of the output box?

Hi experts,

could you help take a look the below issue?

"Is it possible to add a custom API key and interface address to the web version of your official website (https://www.webpilot.ai/)?"

Please support third-party API services such as API2D, Thanks!

Your plugin seems very useful, have you considered making a Chrome new tab extension? It can be used when opening a new tab.

Azure Openai API support needed also

Hello there,

I'm really enjoying this extension, but I've noticed that the prompts aren't appearing when I use the shortcode or click the extension button in Chrome. They only seem to show up when I select some text.

Would you happen to know how I can resolve this issue? Or perhaps let me know when the new version with this feature will be released?

Screenshot attached.

Thank you for your help!

The console is reporting an error. The extension can be installed but cannot run. I am using the ChatGPT API key and I don’t know where the error occurred.
![Uploading Snipaste_2023-11-04_19-38-53.png…]

()

The plugin suddenly became unverified and has been taken off the plugin store, as shown in screenshots attached.

Description:

Only while the WebPilot Chrome Extension is active, there is a JS error in the console when focusing a text field with type=number that runs this.select() on focus/click/etc

Console Error:

Uncaught DOMException: Failed to execute 'setSelectionRange' on 'HTMLInputElement': The input element's type ('number') does not support selection.

Steps to replicate it:

• Make sure the Chrome Extension is active for all websites
• Create a HTML page with the following content: <input type="number" onclick="this.select()" value="1234" />
• Click into the text field. You'll get a JS error, and the number is not selected
• To verify it's WebPilot: Disable the browser extension and try again

Test URL: https://output.jsbin.com/jacaret

Possible source of problem:
https://github.com/webpilot-ai/Webpilot/blob/0874189cade6add978473d9d6146782fda3150f2/src/utils/index.ts#L85C2-L85C2

Related StackOverflow discussion:
https://stackoverflow.com/q/33406169/313501

Screenshot (gif)

Support more ai like AI2C and cohere and etc

阿里云和azure openai都有类似function call的功能。因为用量大，plus限制顶不住所以一直用的api，但是api一直没给联网功能，所以希望用function call功能实现。如果能直接提供类似一键导入的openai.yaml就好了

在 https://platform.openai.com/account/api-keys 获取 openai key，输入后有下面的报错：

请问是因为我当前使用的不是 paid plan，并且已经用完了免费额度吗？

When there is too much output content, the output window is too small and you need to constantly adjust the scroll bar to view more content.

Hopefully, a window zoom button can be added, or the window can be automatically adjusted based on the length of the output content.

          Hi @CocoSgt 

Thanks. It's saving the longer prompts now, but the prompt context length is still short for me; how can I increase its context length to 1000 words or 6000 characters?

Thanks,

Originally posted by @webmasterfsd in #87 (comment)

要用那个开发工具编译啊

Hi Team,

I'm getting time-outs for prompts via the API which work fine on OpenAI's UI with Webpilot as a plugin. Here is an example:

{
"content": "Please research innovations in charity fundraising for me. I want summaries of 3 successful fundraising campaigns used by charities in the last 3 years. Please provide the name of the charity, a summary of their campaign, and links to your source."
}

Via the API, it times out (5+ attempts and no problems with other smaller prompts). On OpenAI I get an exemplary response from Webpilot's plugin.

Any ideas why?

Cheers,

Dorian

Will add text-to-speech function?

配置的代理接口如下图

在划词选中部分文本后调用，可以正常返回结果。但直接快捷键调出窗口，输入“总结网页内容”之类的问题就会报错