Programming Custom GPTs
Last reviewed
May 11, 2026
Sources
17 citations
Review status
Source-backed
Revision
v3 · 2,253 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 11, 2026
Sources
17 citations
Review status
Source-backed
Revision
v3 · 2,253 words
Add missing citations, update stale details, or suggest a clearer explanation.
See also: Custom GPTs, GPT Store, ChatGPT, and OpenAI
Programming Custom GPTs are specialized versions of ChatGPT configured to help with writing, debugging, reviewing, and learning code. They live in the Programming category of the GPT Store, one of the eight main category sections introduced when the store launched on January 10, 2024. The category is also one of the most active corners of the store: by mid 2024, coding-focused tools like Grimoire and Code Copilot had each collected hundreds of thousands of conversations, and computer programming as a topic accounted for roughly 4 percent of total GPT Store traffic.[1][2][3]
This page is a category landing page. It lists the most notable programming GPTs that have been verified through the official ChatGPT GPT page (chat.openai.com/g/<id> or chatgpt.com/g/<id>) and well-known third-party trackers such as gptstore.ai and featuredgpts.com.
A Custom GPT is a tailored version of ChatGPT built with the GPT Builder, the no-code creation tool that OpenAI introduced at its first DevDay on November 6, 2023. A builder configures the GPT with three pieces of customization: a set of instructions that shape its behavior, optional knowledge files, and optional Actions that call external APIs. Once published, the GPT runs on top of GPT-4, GPT-4 Turbo, or, more recently, GPT-4o and later models.[4][5]
A GPT lands in the Programming category when the creator chooses that category at publish time and when its instructions and Actions point at coding tasks. In practice, programming GPTs do one or more of the following:
Many programming GPTs combine the built-in Code Interpreter (also called Advanced Data Analysis) and file uploads so the assistant can run Python sandboxed on OpenAI's servers, read uploaded source files, and return generated artifacts.[5]
The GPT Store is organized into eight top-level sections: Top Picks, DALL-E, Writing, Productivity, Research and Analysis, Programming, Education, and Lifestyle. The Programming section surfaces both first-party GPTs from OpenAI and third-party GPTs built by independent developers and companies. Anyone with a verified builder profile and a paid ChatGPT plan can publish a GPT here.[2]
Users reach a Programming GPT in three ways: through a direct link of the form chatgpt.com/g/g-<id>, through the Explore GPTs picker inside ChatGPT, or through the Programming category browse page. Once a GPT is opened, the conversation runs in the regular ChatGPT chat interface, with the GPT's custom instructions injected into the system prompt and any Actions available as callable tools.
Programming GPTs are gated behind a paid plan: ChatGPT Plus, Team, or Enterprise. Free-tier users have been able to use a small number of featured GPTs since the May 2024 rollout of limited GPT access to free accounts, but most coding GPTs still require a paid subscription.[2][4]
The table below lists widely used and well-documented programming GPTs. Conversation counts and ratings are taken from third-party tracker pages (gptstore.ai, featuredgpts.com, gptshunter.com) and from official OpenAI store listings where available. Numbers shift over time and should be treated as approximate.
| GPT | Developer | Focus | Capabilities | Link |
|---|---|---|---|---|
| Grimoire | Mind Goblin Studios | Full stack web coding from prompts | Generates websites and apps from natural language or image mockups, 20+ hotkey shortcuts, 75 starter projects, Netlify deploy Action, Python tool, file uploads | chat.openai.com/g/g-n7Rs0IK86 |
| Code Copilot | promptspellsmith.com | General multi-language coding assistant | Debugging, code review, snippet generation, optimization advice across Python, JavaScript, TypeScript, and other languages | chat.openai.com/g/g-2DQzU5UZl |
| Code Tutor | Khan Academy | Learning to code | Socratic questioning that guides learners to the answer rather than writing the code; branded as Khanmigo Lite | chat.openai.com/g/g-HxPrv1p8v |
| Python | Nicholas Barker | Advanced Python work | Code review and refactoring against PEP 8, algorithm design, data analysis, machine learning scaffolding, debugging | chat.openai.com/g/g-cKXjWStaE |
| Wolfram | Wolfram Research | Math, science, and computational coding | Calls Wolfram Alpha and the Wolfram Language for accurate math, symbolic computation, curated data, and code execution | chat.openai.com/g/g-0S5FXLyFN |
| ChatPRD | Claire Vo | Product specs for engineering teams | Drafts and improves product requirement documents, coaches on roadmap and stakeholder questions, custom personas | chat.openai.com/g/g-G5diVh12v |
| WebGPT | plugin.wegpt.ai | Web access plus no-code building | Browses the web through Web Requests, builds in no-code playgrounds, calls external APIs through Actions | chat.openai.com/g/g-9MFRcOPwQ |
| Diagrams: Show Me | helpful.dev | Software diagrams | Sequence, ERD, flowchart, timeline, and architecture diagrams using Mermaid and PlantUML, returns editable source | chat.openai.com/g/g-5QhhdsfDj |
| Whimsical Diagrams | Whimsical (Kaspars Dancis) | Visual planning for engineers | Flowcharts, mind maps, sequence diagrams, links back to Whimsical for further editing | chat.openai.com/g/g-vI2kaiM9N |
| AskTheCode | askthecode.ai | GitHub repository Q and A | Reads repo structure, commit history, files, and issues through GitHub Actions; supports private repos with user auth | chat.openai.com/g/g-3s6SJ5V7S |
| AskYourCode | askyourcode.ai | Whole codebase chat | Indexes uploaded source, returns code maps, summaries, and snippets in response to natural language queries | chat.openai.com/g/g-lZd4pBo59 |
[Sources: 1, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]
Grimoire is the most cited programming GPT in independent rankings of the store. It was built by Mind Goblin Studios, the developer behind the GPTavern collection, and positions itself as a coding wizard for what its creator calls Prompt-gram: building software by describing it in plain English. The GPT ships with more than 20 hotkey shortcuts for common coding flows and around 75 starter projects, ranging from landing pages to mini games. It has the Python tool enabled so it can run scripts, accepts file uploads, and includes a Netlify Action for one-click site deployment. Tracker data put Grimoire well into the millions of conversations, making it one of the highest-traffic GPTs in any category.[1][6]
Code Copilot is the most popular general-purpose coding GPT after Grimoire. It is built by an independent developer at promptspellsmith.com and aims at experienced programmers rather than learners. Its system prompt steers it toward concise, production-grade code, with built-in steps for debugging, code review, and performance optimization. Featuredgpts.com lists Code Copilot at roughly 72,000 users with a 4.2 star rating; the GPT is not affiliated with GitHub's separate product also called Copilot.[7]
Code Tutor is one of the few programming GPTs published by a major organization rather than an individual. Khan Academy released it as Khanmigo Lite, a slimmed-down version of the Khanmigo tutoring platform. Unlike most coding GPTs, Code Tutor deliberately refuses to write working code for the user. It uses Socratic questioning to draw the answer out of the learner, suggesting pseudo-code, asking clarifying questions, and pointing back at Khan Academy lessons. Featuredgpts reports a 4.4 star rating across more than 3,200 reviews.[8]
The GPT simply titled Python, built by Nicholas Barker, is aimed at production Python work. It is tuned for canvas and notebook output, runs Python through the Code Interpreter sandbox, and provides PEP 8 review, debugging help, and algorithm design. It has accumulated more than 55,000 users at a 4.2 star rating and is one of the most frequently linked Python tutoring GPTs on third-party rankings.[9]
The Wolfram GPT is the official successor to the Wolfram Alpha ChatGPT plugin from March 2023, after OpenAI retired its plugin system in April 2024. The GPT routes computation, math, and science questions through Wolfram Alpha and the Wolfram Language, returning exact symbolic answers and visualizations. For programming specifically, it is useful for verifying numerical libraries, generating reference data, and writing or executing Wolfram Language code. Stephen Wolfram has written extensively about why pairing the model with a symbolic engine produces more reliable math than the model on its own.[10][11]
ChatPRD was created by Claire Vo over Thanksgiving week in 2023 and quickly became one of the most-talked-about side projects of the early GPT Store era. It does not write code directly; it writes the spec that engineers will code from. The GPT drafts and refines product requirement documents, gives feedback on a draft, and provides coaching on roadmap and stakeholder management. ChatPRD has since grown into a standalone product at chatprd.ai with paid tiers, but the original GPT remains available in the store. Vo has said the project produced six-figure revenue while she was running product and engineering organizations at LaunchDarkly.[12]
WebGPT (the Custom GPT, not the 2021 OpenAI research project of the same name) is built by plugin.wegpt.ai. It combines web browsing through the Web Requests Action with a no-code playground for building small products and an API-calling layer. It was one of the earlier GPTs to top the Programming category and is often used for quick API exploration and scraping tasks.[13]
These two GPTs cover diagramming work that often comes up alongside code. Diagrams: Show Me, built by helpful.dev, generates Mermaid and PlantUML diagrams (sequence, ERD, flowchart, mindmap, timeline) and returns the source so the engineer can edit it. Whimsical Diagrams, built by Whimsical's founder Kaspars Dancis, focuses on flowcharts, mind maps, and sequence diagrams, with hand-off back to Whimsical's own canvas for further editing. Both are used heavily in software design discussions and onboarding documentation.[14][15]
These two are repository-aware GPTs that fill a gap the base ChatGPT cannot: pulling actual code out of a project before answering. AskTheCode connects to GitHub through an Action, can read file structure, commit history, and issues, and supports private repos when the user authenticates. AskYourCode takes a different angle: instead of pulling from GitHub, it indexes uploaded source files and returns code maps and summaries from that index. AskTheCode reports around 3,300 reviews at a 4.1 star rating.[16][17]
The GPT Builder is the tool every Custom GPT in the Programming category is built with. It is itself a GPT, accessible at chatgpt.com/gpts/editor for paid users, and it offers two modes: a conversational Create flow that drafts the GPT through dialogue, and a Configure view that exposes raw instructions, knowledge file uploads, capability toggles (Browsing, Python, DALL-E, Canvas), and the Actions builder for hooking up external APIs.[5]
For programming GPTs specifically, three configuration choices matter most:
Programming GPTs share most of the constraints of plain ChatGPT: they can hallucinate APIs that do not exist, mix up version-specific syntax, and miss subtle bugs in larger code. Three issues are specific to the Custom GPT layer:
Another common complaint, especially for the more popular GPTs, is that the system prompt has leaked publicly. The repositories at github.com/LouisShark/chatgpt_system_prompt and github.com/linexjlin/GPTs include extracted prompts for Grimoire, Code Tutor, Whimsical Diagrams, and others. OpenAI considers leaked instructions a known limitation: any user who can talk to the GPT can ask it to reveal its instructions, and only weak defenses against that are possible.