v0.2.1

What's Changed

• Support integration with qdrant vector database by @hezhaozhao-git in #207.
• Support integration and execution on github codespaces by @LeonZh0u in #186.
• Add integration with multiple web search engines (duckduckgo_search/googleapi) by @shenchucheng in #161.
• Enhance web search result parsing capability, enabling targeted web content retrieval and result parsing by @shenchucheng in #69.
• Add timeout settings and network anomaly alerts for the OpenAI interface by @wangjie5540 in #112.
• Incorporate fault tolerance in update_cost and update_usage calculations, and reduce instances of code execution errors caused by network anomalies by @alitrack in #138.
• Introduce and refine the role of qa engineer, enabling abilities such as writing test cases, running code, and debugging errors by @garylin2099 in #77.
• Fix the bug that can not use proxy and update the default setting check mechanism by @flyi in #121
• Introduce and refine the role of researcher, enabling abilities such as online searching, analyzing and summarizing web search results, and writing reports by @shenchucheng in #161.
• Add custom loop policy for Windows by @bocandrei in #257
• Bug fix for ProductManager import by @denniseilander in #221
• Handle exception when using Anthropic_API_KEY instead of OPENAI_API_KEY by @flyi in #194
• Add MetaGPT quickstart doc links by @voidking in #139
• Fix typo problem in run_code.py by @jansalvador in #144
• Correct requirements and install instructions by @talkingtoaj in #238
• Add action write docstring, supporting sphinx/google/numpy styles by @shenchucheng in #124
• Update the Dockerfile, translate Chinese comments within the code, and perform other typo corrections, thx @voidking @LeonZh0u @flyi.
• Fix bug when calling faiss_store write by @seehi in #84
• Fix bug for incorrect parse_tasks and parse_workspace by @femto in #116
• Support installing mmdc locally and skipping downloading Chromium by @alitrack in #83
• Implement code review to enhance the executability of the generated code, optimize the code-writing prompt to improve the executability of the generated code by @qa6300525 in #73
• Add a retry mechanism to reduce the probability of code generation failure by @qa6300525 in #73
• Add memory_storage using ann to avoid similar idea repetitive execution by @better629 in #70
• Add a new api key field and create a wrapper class for serper API by @LeonZh0u in #50
• Format import according to sort and clean up existing unused code, add .pre-commit-config.yaml to automatically detect code issues when code commit by @seehi in #67
• Add requirements.txt save and api_spec_and_tasks.md by @Hallimede in #57
• Bug fix: delete special characters when creating a directory by @voidking in #137
• Add support for anthropic Claude api by @LeoXV1 in #65

Other updates:

1. Thanks to @geniuslzh in #136, for FAQ-EN
2. Thanks to @eltociear in #143, creating REAME_JA.md
3. Thanks to @hezhaozhao-git in #208, adding ruff for code linter
4. Thanks to @voidking and @alitrack for their work on updating the docker image and optimizing the Dockerfile in #106 and #109.
5. Thanks to @stellaHSR in #63, for providing the ui role and enabling the use of the stable diffusion text2image api.
6. Thanks to @wangjie5540 in #179, for optimizing the Dockfile.
7. Thanks to @martcpp in #156, for providing English comments.

v0.1.0

v0.1.0

1. Establish a separate output queue, differentiated from the message queue
2. Support Azure asynchronous API
3. Make gpt-3.5-turbo available (HARD)
4. Support Docker
5. Support pip
6. Make a stable requirements.txt
7. Support streaming mode
8. Add Japanese README
9. Resolve config conflicts
10. Remove langchain in main path
11. Support non-git-clone project
12. Adapt pydantic

Attached files are some games generated by MetaGPT. ~$0.7 for one game.