[Open Source] An AI copilot focused on IT operations, based on Rasa and LLM technologies.
![[Open Source] An AI copilot focused on IT operations, based on Rasa and LLM technologies.](https://yy123.ink/api/img/1782591198852-1ad0kv.jpg)
1. Project Overview
OpsPilot

OpsPilot is an open-source AI copilot developed by the WeOps team, based on Rasa and LLM technologies and focused on IT operations. It supports integration with web applications in the form of a ChatBot.
2. Open Source License
Licensed under Apache-2.0
3. Screenshots

4. Feature Overview
OpsPilot is an open-source AI copilot developed by the WeOps team, based on Rasa and LLM technologies and focused on IT operations. It supports integration with web applications in the form of a ChatBot.
The main capabilities include:
- IT Operations Knowledge Accumulation: Consolidates operational knowledge, skills, and troubleshooting procedures. When solving problems, it guides users through conversations in the form of an AI copilot to help resolve IT operations issues.
- Local Knowledge Q&A: Indexes local and internet-based knowledge and combines it with LLM capabilities to answer various IT operations questions.
- LLM Chat: When a question exceeds OpsPilot's capabilities, LLM technology is used to handle various long-tail questions.
5. Tech Stack
Deployment
export OPENAI_ENDPOINT= export OPENAI_KEY= cd ./support-files/ docker-compose up -d
Development Environment Setup
Use Python version 3.10.
virtualenv venv -p python3.10 pip install -r ./requirements.txt pip install -r ./requirements-test.txt python ./ops_pilot_cli.py init_data
Start the OpsPilot NLU service.
rasa run --enable-api --cors "*"
Start the OpsPilot Actions service.
rasa run actions --auto-reloa
Frequently Asked Questions
How to Enable JWT Authentication for WebSocket
Modify credentials.yml and add the jwt_key and jwt_method configurations.
socketio: user_message_evt: user_uttered bot_message_evt: bot_uttered session_persistence: true jwt_key: key jwt_method: HS256
How to Convert Docusaurus Web Pages into PDF Files
npx docusaurus-prince-pdf -u xxx
How to Install requirements.txt on Mac
export HNSWLIB_NO_NATIVE=1 pip install -r requirements.txt
Parameter Description
Parameters
Description
Optional Configuration
FALLBACK_LLM
Use an LLM to generate responses when OpsPilot cannot handle a request.
OPENAI
OPENAI_ENDPOINT
Endpoint of the model deployed on OpenAI.
OPENAI_KEY
API key used by the model deployed on OpenAI.
JENKINS_URL
Jenkins URL. Required to enable Jenkins automation capabilities.
JENKINS_USERNAME
Jenkins username. Required to enable Jenkins automation capabilities.
JENKINS_PASSWORD
Jenkins password. Required to enable Jenkins automation capabilities.
BING_SEARCH_URL
Bing Search endpoint.
BIND_SEARCH_KEY
Bing Search password.
VEC_DB_PATH
Vector database path.
RUN_MODE
Whether to run in development mode.
FALLBACK_CHAT_MODE
Whether the LLM uses local knowledge base mode or casual chat mode.
REDIS_HOST
Redis IP address.
REDIS_PORT
Redis port number.
REDIS_DB
Redis database number.
REDIS_PASSWORD
Redis password.
Version Description
0.2
- Prompt configurations are stored in Redis.
- Uses both inverted indexing and semantic search to improve LLM inference accuracy.
0.1
- Completed the basic framework setup.
- Supports ChatGPT casual chat mode.
- Supports indexing target websites and local PDF documents to provide local knowledge-based Q&A.
- Supports Intent-based integration of IT operations capabilities.