Visit the official CodeGPT Plus website to access the Studio. To access the Studio, you must create an account at Code GPT Plus.

In the Studio, you can:

Interact with Agents: Engage with the agents and fine-tune their behavior.
Configuration and Adjustment: Easily configure and adjust agent behavior through specific context documents.
External Sharing: Access an external link to share your agents with others.
Embedding: Embed your agents seamlessly into any website.
Train your agent with specific knowledge

There are two ways to log in to CodeGPT Plus:

Creating a Username and Password:
- Enter your Fullname.
- Provide your Email account (please note that you will have access).
- Create a password (if you forget it in the future, you can restore it here).
Initiating Authentication via Google or GitHub:
- Follow the Google or GitHub authentication process.

⚠️ Remember, if you choose these two options, you must always log in using the same method.

My Account

Personal Data

This panel gives you access to the personal data associated with your account. You can edit your name and include a phone number if you wish.

If you want to delete your account, you can use the Delete Account option available here: https://app.codegpt.co/en/account

For Free-trial accounts, personal data will be removed from our database within a 30-calendar-day interval.
For paid accounts, personal data is deleted, and the payment subscription is canceled for the remaining contracted time. After this period, any data and agents created in the Studio will be lost unless the plan is renewed before the canceled plan expires.

Studio

Personalization

You can change the theme (Dark or Light), just click on your profile picture-
Send your feedback and manage your notifications on the Left menu.

Features

You can enter directly to manage your agents, create or edit the agents. However, we show you 5 main sections in this new version.

Agents
Marketplace
Files
Apikeys
Subscription
Settings
Referrals

Let's talk about all of them!

Agents

You can create many kind of agents, but we suggest 5 variants:

CodeGPT Agent: We setup the prompt of this agent to helping you on coding tasks. Of course you can add it some language programming documentation and edit the prompt.
Q&A Agent: you can edit this prompt but try to use the format of answer to maintain the effectiveness.
- Upload the list of Q&A
- Make a prompt for the format of the questions and answers
Dense summarization Agent: for large context, the prompt of this agent can help you with the summarization of large text, which mean that is prepared to receive the maximum of tokens that you can send to it

GitHub Agent: will need to add a public link repository from GitHub.
- Include mainly GitHub repositories
- Upload a GitHub Repository, get the link repo.
  If you have a paid plan account, you can import a GitHub repository using the Import GitHub Public repository button. Remember that the repository must be public, and the size is limited to 20MB per upload.
And of course, from scratch, New Agent.

New Agent: Agent from scratch

Also you can use the New Agent button available in the right superior area. Both will take you to this tab:

Agent Settings 🤖

Choose the AI model, test the model according you use case. OpenAI and Anthropic models are available into this version.

⚠️ Please be mindful of the documentation for the models, as each model may have a unique approach to constructing responses. If your use case requires a more extended response or considers additional context, review the documentation of each model thoroughly 😇.

Conversation style
In AI models, the conversation style (temperature) is a parameter used in text generation to control the randomness of output. A high temperature leads to more diverse but potentially less coherent text, while a low temperature produces more predictable and contextually fitting output. Adjusting the conversation style allows users to balance between creativity and coherence in generated text. We offer 3 levels:

- Precise: from 0 to 0.4
- Balanced: between 0.4 to 0.7
- Creative: higher than 0.7
Instructions 👩🏻‍🏫
💡Some tips to improve make a proper prompt:

-Be clear (no need to be kind) and specific
-Include examples for the result answer
-Test your prompt with main questions according your agents/system goals
-Made a struct of the response, as an example
Think the answer like a NLP Task. Your agent will be able to: summarize, text transforming, inferring, extract or classify text?.

Be like:

For develop a movie text classification agent

Develop a movie text classification agent that efficiently categorizes movie descriptions into genres. Be explicit about the criteria for classification, considering genres such as action, drama, comedy, thriller, and science fiction. Test the prompt with the following questions:

- What features does the agent focus on for genre classification?

- Can you provide examples of movie descriptions and their expected genres?

- How does the agent handle ambiguous or mixed-genre movie texts?

Extract key features influencing the genre classification.

Example:

Input: "A group of friends embark on a thrilling adventure to uncover a hidden treasure." 
Output: { "Summary": "Friends on a thrilling treasure hunt", "Text Transformation": "Adventure", "Inference": "The adventure genre is identified based on the thrilling treasure hunt.", "Extraction": "Key features include 'friends,' 'thrilling,' and 'treasure hunt."}

Chat with the agent

🗨️ Chat with the agent - now with 3 more features!

Reset the chat - this will clear the screen for a fresh start.
Send your message - communicate your thoughts or queries.
Interact with an ephemeral document - upload a document and ask questions about it.

Please remember:

🚫 You can only upload one document at a time.
📏 Respect the document upload limits, as described in the Files section.

Advanced

Agent Knowledge
In this menu you can Add a new document or plug documents available on your Files.
- You can get the files from the list
- Check the files that you want to plug to your agents. Don't forget to click on "Plug" to save it.
- Check the number of tokens and if you want to know how we calculate this, go to this section (⚠️ Regarding File Size and Storage Capacity).
- Get the id document from "⋮" button
- Only .PDF, .DOC, .XLS, .PAGES, 20MB Max, and .TXT, .CSV 2MB Max.
- For URL, be aware that this include a scrapping and crawling process to keep in mind that some websites have a tricky structure or security mechanism that does not allow to get the data.
- Or Upload a new one (you would find it in the Files section too)

Done! Test your agent on the Studio.

You can also share it by getting the link and seeing a preview of the button. Enable the option Publish in Private.

If you want to work with this agent on VS Extension just follow the instructions available in button.

⚠️ If you disable the option to share the agent link, the agent's status will change from Private to Public

Conversation Starters

Conversation starter buttons on a chatbot are pre-defined options or prompts presented to users to kick-start the conversation. They help guide users on what to ask or discuss with the bot. Conversation starter buttons streamline the user experience by offering predefined options, reducing the need for users to formulate their own queries from scratch.

Chat Skin 🖌️
You can change the skin of your agent, we made 4 combinations available.

Chats History
Here, you can effortlessly 📊 monitor the number of messages, dates, and intricate details of all your interactions. However, ⚠️ please be aware that the extent of this feature's functionality may be subject to the limitations of your chosen payment plan.

Save chats in the left panel, name each chat, or organize them into folders.
Double-click on the message (select a day), view the status and token usage (both prompt and completion), and gain access to the following:
- Messages: This displays all the messages and responses within that conversation.
- Debugging: This allows you to inspect the prompt (instructions) and chunks related to the questions.
- Embeddings: This enables you to examine the embeddings by chunk, but only when you have used RAG.
- Duration: This lets you review the times for embedding creation, semantic search, start completion time, and start conversation time, all measured in seconds.

Marketplace

You can also try an agent deployed by us or our community on Marketplace. Just look the one which fits with you and install it on your Studio. Or request to create one expert on your request.

All Apps

Examine the agents that you can connect to Discord, Slack, or utilize the Chat Widget.
For the Chat Widget, you can only link one of your agents, as the widget ID is exclusive to a single agent.

Activate the agent and embed the code into your webpage's HTML code.

Look this example:

File

The Files feature is designed to provide specific knowledge to the agents. This feature benefits from the ability to train an agent with specific information, thus enriching the agent's prompt. You can fine-tune the model to perform optimally in specific situations and contexts, which can be challenging to achieve with more general models—all of this without the need to generate lines of code.

Upload a File

Allowed formats for uploading a file include .txt, .csv, .xls, .xlsx, .pdf, .doc, .docx, .pages, .dot, .dotx, .log, .odt, .rtf. You can drop the file onto the panel or click the Upload button.

Now you can Refine your data, managing the Content, metadata and how to make the chunks.

Refine the Content, you can edit the content manually.

You can ✏️ edit and 💾 save the file.

Introduce Metadata showing the advance settings. Please ensure that the file titles are contextually related to their content, making it easier for agents to 🔍 search for answers within the files. If ideas don't come to you, we've integrated an 🤖 AI helper to figure out the titles for you ✨., just click on Generate button.

Manage the chunks, change the split settings. Managing the chunks and adjusting split settings is important for optimizing processing efficiency and ensuring smooth data handling. Here's why:

Efficiency: Breaking data into manageable chunks prevents overwhelming the system with large volumes of information at once, improving processing speed and resource utilization.
Resource Allocation: By controlling the size of data chunks, system resources such as memory and processing power can be allocated more effectively, preventing bottlenecks and enhancing overall performance.
Error Handling: Splitting data into smaller chunks reduces the risk of errors or failures during processing. If an error occurs with one chunk, it's easier to identify and handle, minimizing the impact on the entire process.
Scalability: Adjusting split settings allows systems to scale more efficiently to accommodate growing data volumes. Managing chunks effectively becomes crucial for maintaining performance and reliability as data size increases.

Upload a URL content

Incorporate content from websites, keeping in mind that websites often have multiple links for each page's content. ⚠️Ensure you include all the links if you intend to capture all the data from a website.

For instance, consider the website "https://www.codegpt.co/"; it contains various links for each feature. If you require all the data, retrieve all the links and include them individually. For specific information, like Pricing details, make sure to upload the link for that content separately, such as "https://www.codegpt.co/pricing/#info1-r".

Upload a GitHub Repository

If you have a paid plan account, you can import a GitHub repository using the Import GitHub button. Remember that the repository must be public, and the size is limited to 10MB per upload.

⚠️ Regarding File Size and Storage Capacity

Remember that depending on the contracted plan, you have a maximum file storage limit. While the maximum file upload size is 20MB, the vectorization process's resulting file size depends on various factors:

The Language Model, for example, if you're using GPT-3, GPT-4, etc. The model has its tokenization strategy, hence the number of tokens varies based on the model.
Knowing the Text Length, the number of tokens required depends on the text length. You can divide your text into tokens to estimate. Note that some models may have limits on the maximum number of tokens they can process in a single call.

Example:

If you are using the GPT-3 model from OpenAI, according to the documentation available on their website, this model would divide the following sentence into 23 tokens:

"The artificial intelligence is transforming many aspects of our daily lives. From virtual assistants to recommendation systems, AI is present in various applications."

GPT-3 uses "tokens" that are not just words but smaller segments of text.

If we assume each token occupies approximately 4 bytes in memory, we can calculate the conversion weight as follows:

23 tokens * 4 bytes/token = 92 bytes

In other words, the longer the sentence, the more bytes the vector will occupy in this case.

For more information about tokens you can use this app.

Apikeys

Now we generate the Organization ID, you will used to link other user to your agents. With this version, you can create a Secret API Key, be sure to saved it properly.

Subscription

See the detail of your current plan, on this tab you can change your subscription, or unsubscribe.

You can manage your subscription: plan, info (in case that you want the bill in the name of your company for example) in the Manage subscription button.

Settings

Set your profile

Organization

If you wish to customize the name and company logo, this menu allows you to personalize the agents.

Also, invite new members to try the platform, with same Organization customization.

Profile

⚠️ Manage your account settings, Full Name, user name, and password. You can also delete your account using this option.

Referrals

We're thrilled to announce the launch of our brand new Referral Program, designed to reward YOU for spreading the word about our amazing products/services! It's our way of saying thank you for being a part of our community and helping us grow.

Here's how it works:

👉 Step 1: Get Your Unique Referral Code

Upon signing up or logging into your account, go tothe Referrals tab to get an exclusive referral code. This code is your ticket to unlocking incredible rewards!

👉 Step 2: Share with Friends & Family

Share your referral code with friends, family, colleagues, and anyone who could benefit from our offerings. Whether through social media, email, or good old-fashioned word of mouth, spread the love far and wide!

For more details, please check here.

Error Messages

Error 400: Possible reasons include CodeGPT Plus API Key issues, often due to an expired plan. You can update your payment method or transition from a free trial to a payment plan.
Server Error: Communication with the model server is disrupted. This can occur if users change the API key server in Studio Settings, and their API Key server lacks credit or is unavailable.
Agents Not Responding: Check your prompt configuration; perhaps you added curly brackets or the agent's name doesn't match the current one.