ai-citizen/solidgpt

Developer AI Persona Search Agent

solidgpt's Issues

开启代理执行python3 quickstart.py system后出现如下错误

openai.error.APIConnectionError: Error communicating with OpenAI: HTTPSConnectionPool(host='api.openai.com', port=443): Max retries exceeded with url: /v1/chat/completions (Caused by SSLError(SSLEOFError(8, 'EOF occurred in violation of protocol (_ssl.c:1129)')))

onboard project阶段：上传代码文件后点击start时报错

Unexpected token '<', "<!DOCTYPE "... is not valid JSON

Fix README grammar, spelling, and phrasing.

I have created a pull request that fixes many of these issues. Find it here.

OpenAI API key

How to change the openai base url?

Cant upload folder ?

TypeError: Failed to fetch

Unable to Initiate `chat` with Repository on Publicly Hosted Web Application

I am encountering difficulty utilizing the recently released chat feature on your publicly hosted website. It seems there isn't a clear method to initiate this feature.

Could you kindly assist me or direct me towards the corresponding instructions or guides?

Error during onboarding, stuck in a bad state. How to reset everything?

I setup the project and loaded my repo. The repo was too large and so I drilled into a library and selected that instead. Upon resubmitting I received a network error. Now I've uninstalled all packages and reset the project to main. Yet, I still seem to be stuck in this blank state with errors.

How can I nuke this from orbit and start over?

Graph has an error.

Hi Team,

When ever I tried to onboard a project I get the onboard success
[2023-12-11 19:10:01,589: INFO/MainProcess] Task solidgpt.src.api.celery_tasks.celery_task_onboard_repo_graph[0e3c67f4-1e02-4fc0-b62a-c3d264ff44df] succeeded in 127.07799999997951s: '---****-

However when I start chatting I this error
raise ValueError(f"Collection {collection_name} not found")
ValueError: Collection d2d1a8a9-1067-429f-9525-0e6090045e8e not found

Complete log
[2023-12-11 19:10:48,574: WARNING/MainProcess] Directory 'D:\SolidGPT-0.2.7\solidgpt../localstorage\workspace\out\d32ab2ce-1e97-4447-854c-3aa669a6d8f0\Query_code_0' created.
[2023-12-11 19:10:48,575: WARNING/MainProcess] Directory 'D:\SolidGPT-0.2.7\solidgpt../localstorage\workspace\out\d32ab2ce-1e97-4447-854c-3aa669a6d8f0\Repo_Chat_1' created.
[2023-12-11 19:10:48,575: INFO/MainProcess] Executing SolidGPT...
[2023-12-11 19:10:48,575: WARNING/MainProcess] Node begins Query code task...
[2023-12-11 19:10:48,577: INFO/MainProcess] Start to search best five code...
[2023-12-11 19:10:49,580: ERROR/MainProcess] Task solidgpt.src.api.celery_tasks.celery_task_repo_chat_graph[56417a7d-fb8c-499f-acd1-e79e375d97c6] raised unexpected: ValueError('Collection d2d1a8a9-1067-429f-9525-0e6090045e8e not found')
Traceback (most recent call last):
File "\AppData\Local\Programs\Python\Python310\lib\site-packages\celery\app\trace.py", line 477, in trace_task
R = retval = fun(*args, **kwargs)
File "\AppData\Local\Programs\Python\Python310\lib\site-packages\celery\app\trace.py", line 760, in protected_call
return self.run(*args, **kwargs)
File "D:\SolidGPT-0.2.7\solidgpt\src\api\celery_tasks.py", line 117, in celery_task_repo_chat_graph
g.execute()
File "D:\SolidGPT-0.2.7\solidgpt\src\workgraph\workgraph.py", line 115, in execute
self.__execute_node(node)
File "D:\SolidGPT-0.2.7\solidgpt\src\workgraph\workgraph.py", line 122, in __execute_node
node.skill.execute()
File "D:\SolidGPT-0.2.7\solidgpt\src\workskill\workskill.py", line 52, in execute
self.execution_impl()
File "D:\SolidGPT-0.2.7\solidgpt\src\workskill\skills\query_code_local.py", line 51, in execution_impl
top_five = self.__find_top_five()
File "D:\SolidGPT-0.2.7\solidgpt\src\workskill\skills\query_code_local.py", line 61, in __find_top_five
search = self.client.search(
File "\AppData\Local\Programs\Python\Python310\lib\site-packages\qdrant_client\qdrant_client.py", line 315, in search
return self._client.search(
File "\AppData\Local\Programs\Python\Python310\lib\site-packages\qdrant_client\local\qdrant_local.py", line 154, in search
collection = self._get_collection(collection_name)
File "\AppData\Local\Programs\Python\Python310\lib\site-packages\qdrant_client\local\qdrant_local.py", line 113, in _get_collection
raise ValueError(f"Collection {collection_name} not found")
ValueError: Collection d2d1a8a9-1067-429f-9525-0e6090045e8e not found

When I checked the UI I can see the message - "Graph has an error."

Let me know what may be the issue. I am trying to onboard JAVA code base.

API call fail, please retry

I can use the py script to call the openai API

有關於工作分流

SolidGPT可以監察我的notion 多個page的task然後自動代分配給不同的員工嗎? task由我不定期新增在notion

VS Code Extension (zsh: exec format error: ..../aict.solidgpt-0.0.6/server/run_api_x86)

Hi,

Having issues with the VS Code Extension. I noticed it was stuck on "Initializing Please Wait." for a long time so I looked at the Terminal output and saw it was throwing: zsh: permission denied: /home/<me>/.vscode/extensions/aict.solidgpt-0.0.6/server/run_api_x86. So I chmodded it with exec privileges but now it's throwing: zsh: exec format error: /home/<me>/.vscode/extensions/aict.solidgpt-0.0.6/server/run_api_x86. I also tried executing it using bash but doing that it throws a different error: /home/<me>/.vscode/extensions/aict.solidgpt-0.0.6/server/run_api_x86: /home/<me>/.vscode/extensions/aict.solidgpt-0.0.6/server/run_api_x86: cannot execute binary file.

Not quite sure what the issue is. Any help would be appreciated.

AxiosError: Network Error

i can visit the web of openai but still hold the problem

Question about Notion Support?

In the previous version I had the capability to upload output to Notion and sync it back seamlessly. I'm wondering whether this functionality is still supported in the current version.
Additionally, I'm interested in knowing if there are plans to further focus on and enhance Notion integration in future updates.

API call failed when onboarding

I keep get the api failed message when I try to onboard my project, block on 11/51 😢

Deploy with Llama2

Hi SolidGPT team,

Are there any ways I can deploy your product with Llama2 or any other open source model?

I saw you have Llama manager but didn't find a way to use it.

BR,

{"detail":"Not Found"}

I get the following when I start the server: {"detail":"Not Found"}.

I am on a Mac M1. When I run the shell script I'm routed to http://0.0.0.0:8000/ where I receive {"detail":"Not Found"}.
I'm confused, what gives?

Comparison and questions about SolidGPT

Comparison of AI-SDE projects

This issue compares three open source projects that aim to automate aspects of software development using AI:

• SolidGPT ↗
• DevOpsGPT ↗
• MetaGPT ↗

Overview

All three projects integrate large language models (LLMs) with existing tools to generate code and other artifacts from natural language requirements. Some key differences:

SolidGPT focuses on human-AI collaboration using Notion and Lowdefy. It aims to automate entire workflows and integrate with collaboration tools.

DevOpsGPT emphasizes automating software development and deployment processes. It claims to support analyzing existing codebases and multiple DevOps platforms.

MetaGPT takes a "software company" approach, assigning AI agents different roles like PM, architect, engineer. It generates a full project from an initial idea.

Feature comparison

Feature	SolidGPT	DevOpsGPT	MetaGPT
Supports custom agent skills	✅	✅	✅
Analyzes existing codebases	❌	✅	❌
Supports private data	✅	✅	✅
Integrates with tools	Notion, Lowdefy	Multiple DevOps tools	None
Automates workflows	✅	Partial	Partial
Generates full projects	Partial	Partial	✅

Room for improvement

All three projects could benefit from:

• More robust testing of generated code quality
• Support for additional languages/frameworks
• Fine-tuning models on more domain-specific datasets
• Improved handling of edge cases in requirements
• Open sourcing pre-trained models for customization

They also have room to learn from each other, such as MetaGPT integrating collaboration tools or DevOpsGPT taking a more "software company" paradigm. Overall, these projects show promise for advancing AI-driven development.

Questions

After examining the SolidGPT project and playing with demos, I have a few questions that might help myself and future users understand and utilize it with better precision:

1. Private Data Integration: Could you elaborate more on how SolidGPT handles private data? Is there any specific security measure implemented to ensure data privacy and protection? The project mentions using private data, but does not provide details. How do users connect their own private datasets? What format does the data need to be in?

2. Agent Customization: In reference to agent and skill customization—Is there a limit to what an agent can do, or could it feasibly perform any task with enough customization? It says users can customize agents, but documentation is lacking. How exactly can agents be customized - via prompt editing, fine-tuning, or something else? Examples would be helpful.

3. Notion Integration: How essential is Notion to the functioning of SolidGPT? Are there alternative platforms we can use, or do we plan to support them in the future?

4. SolidGPT Schema: Could we get a better understanding of the solidgpt/src/configuration/Configuration.yaml file? What are in general the best practices when configuring this file?

5. Framework Versatility: While the current framework is optimized for software development, how easy or hard would it be to adapt this solution to other fields like research, marketing, or education?

6. Scalability & Performance: Could you provide some guidance on scaling this solution? How does it perform under various workload sizes—small, medium, large?

7. Output formats: What file formats do agents generate output in? The demos only show Notion integration - can outputs also be saved in other formats like Markdown, JSON etc?

8. Production usage: The project focuses on proof-of-concept demos. How far has the framework been validated for production software development workflows? What testing/QA processes are in place?

9. Language support: The demos only show code generation in JavaScript/TypeScript. Can SolidGPT agents support automation for other languages like Python, Java, C# etc.?

I'm excited to learn more about the AI-Citizen/SolidGPT project and would greatly appreciate any insights into these aspects. Please feel free to share any additional information that could help in understanding SolidGPT better. Getting clarification on these questions would help better understand SolidGPT's capabilities and limitations. The project shows promise as an open framework for human-AI collaboration --- please let me know if you need any help with development!