This article shares the author's practical experience of developing an MCP service at zero cost and successfully monetizing it for 1,000 yuan in just five days of spare time, using AI tools such as Co...
| Tool | Comprehensive Recommendation | stability | cost | performance | Access Restrictions |
|---|---|---|---|---|---|
Tencent CodeBuddyWebsite | 4.9/5 | high | free | High efficiency, quickly adapted to GPT-5 | none |
 Claude CodeWebsite | 4.0/5 | There are many code bugs | Depend on transit station | Fast response but easy to deviate | Depend on transit station |
CursorWebsite | 3.0/5 | Poor stability, frequent errors | Restricted Free | powerful but limited | Restrict domestic access |
Recently, I used CodeBuddy + GPT-5/Gemini 2.5 Pro to develop 10 MCP services in just 5 days of spare time and successfully monetized ￥1000! The entire process cost me nothing. AI tools helped me efficiently complete the entire workflow from market research, product design, to development and testing. The project duration was reduced from an initial 4 hours to just 45 minutes (primarily dependent on GPT-5's response speed, which has been extremely popular lately). You don't need to spend much time on it; you can completely step away to focus on other important tasks or check the stock market. Of course, these insights are the result of countless trials and errors.
If you also want to use AI technology to earn money in your spare time, feel free to check out my practical experience, and I’ll also share my recent hands-on experiences with CodeBuddy, Cursor, and Claude Code.
Let me briefly introduce MCP (Model Context Protocol). MCP is an open standard launched by Anthropic in late November 2024, aimed at unifying the communication protocols between large models and external data sources and tools. The primary purpose of MCP is to address the current challenge where AI models cannot fully leverage their potential due to data silos. MCP enables AI applications to safely access and manipulate local and remote data, providing a universal interface for AI applications to connect with everything.
Let me give you a concrete example. I’ve selected one of the optimized workflow steps from a real-world case study to share with you, so you can directly experience the elegance of CodeBuddy and the powerful capabilities of GPT5+Gemini 2.5 Pro.
Let’s explore the collaborative power of CodeBuddy+GPT5+Gemini 2.5 Pro through a real-world case study. This case is one of the 10 MCP services I developed — a “one-source-multiple-use” MCP tool for content creators, which took only 45 minutes from conception to self-testing completion.
2. Output PRD.md and tasks.md documents (5 minutes)
Here, you need to configure the prompt words for project rules (general prompt words for product and development that have been optimized through dozens of project iterations; the stronger the model, the more effective it is). If you need them, you can leave a message in the comments section to obtain them.
3. GPT5 Core Code Generation (18 minutes)
4. GPT5 self-test (5 minutes)
5. Write README.md (3 minutes)
The entire process felt like working with a professional development team: Gemini handled product design, GPT5 took the lead in development, and CodeBuddy seamlessly filled the role of technical director, automatically fixing two potential bugs in GPT5's code.
From a vague idea to a fully functional MCP service with complete deliverables, the entire process took only 45 minutes, generating a revenue of 100 yuan for me. This is the speed and excitement of monetizing technology in the AI era.
In choosing AI programming tools, I went through an interesting exploration process. Initially, I primarily used Cursor, leveraging its features through various means. However, at the end of last month, Cursor suddenly restricted access from within China. While it was possible to bypass this by switching proxies or reconfiguring via GitHub tools, stability was significantly compromised, and frequent errors became a major headache.
Just the next day, Claude Code began to gain popularity. I tested multiple intermediary platforms, including Any Router, Kimi, and Qwen. Among these, Kimi cc offered a particularly smooth user experience, especially with Claude 4 Opus playing a significant role in development. However, after a blogger revealed that Qwen had issues with excessive billing, I immediately ceased using it.
By late July, I obtained an invitation code for Tencent CodeBuddy. This further boosted my AI programming efficiency, especially since it quickly adapted to support GPT5 the day after its release. During this period, I focused on optimizing several overseas sites, achieving significant results—the best part being that it was free.
During my first experience with GPT-5, I used the Agent mode, carefully verifying each step. Although the speed was slow, by midday testing, the overall performance was satisfactory, and the coding capabilities were indeed powerful.
On the second day when using GPT-5 again, the response speed in the afternoon was noticeably slower, and failures occurred frequently. I then adjusted my strategy: using Gemini 2.5 Pro's Chat mode to generate documentation, followed by GPT-5 for coding. Although GPT5 is slower, the output quality is very high, and it can implement the functionality in as few as two rounds, with a clear and accurate implementation approach. In contrast, Claude 4 has a fairly fast execution response speed, but its performance is somewhat “unrestrained,” with the generated code containing many bugs that require constant adjustments. More troublesome is that if the problem description is not accurate enough, it may completely deviate from the direction, or even clear the original text or configuration content.
The submission deadline for the event is August 22. Why not take advantage of your weekend free time to give it a try?
✅ Earn extra income at no cost
✅ Quickly improve your AI programming practical skills
✅ Master the latest tool combinations
Feel free to leave any questions, and the event link is: https://yunyinghui.feishu.cn/wiki/EM0xwhRFwiIljqkpDDucI4gAnxj.
Loading comments...