https://github.com/bayars/router-config-rag
A fully local RAG (Retrieval Augmented Generation) agent that helps configure Nokia SR Linux/SROS/SR-SIM routers using official documentations.
https://github.com/bayars/router-config-rag
Last synced: 6 months ago
JSON representation
A fully local RAG (Retrieval Augmented Generation) agent that helps configure Nokia SR Linux/SROS/SR-SIM routers using official documentations.
- Host: GitHub
- URL: https://github.com/bayars/router-config-rag
- Owner: bayars
- License: mit
- Created: 2026-01-01T05:41:43.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2026-01-01T05:45:55.000Z (7 months ago)
- Last Synced: 2026-01-06T00:14:17.564Z (7 months ago)
- Language: Python
- Size: 22.8 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Nokia SR Linux Router Configuration Agent
A fully local RAG (Retrieval Augmented Generation) agent that helps configure Nokia SR Linux routers using official documentation. Runs 100% locally with Ollama - no cloud APIs or API keys required.
## Features
- **100% Local**: Runs entirely on your machine using Ollama
- **RAG Pipeline**: Semantic search over SR Linux documentation
- **Multi-Router Config Generation**: Generate configs for multiple routers at once
- **Interactive Chat**: Natural language interface with conversation history
- **Web Scraping**: Can ingest documentation from websites
- **Persistent Knowledge Base**: ChromaDB vector store
## Architecture Overview
```
┌─────────────────────────────────────────────────────────────────────────────┐
│ USER INTERACTION LAYER │
│ ┌─────────────┐ │
│ │ main.py │ Interactive CLI / Commands: stats, reset, quit │
│ └──────┬──────┘ │
└─────────┼───────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────────────────┐
│ RAG AGENT LAYER │
│ ┌──────────────────────────────────────────────────────────────────────┐ │
│ │ router_agent.py │ │
│ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ │ │
│ │ │ Query Analysis │───▶│ Context Builder │───▶│ LLM Generation │ │ │
│ │ │ (intent detect)│ │ (top-k chunks) │ │ (via Ollama) │ │ │
│ │ └─────────────────┘ └─────────────────┘ └─────────────────┘ │ │
│ │ │ │ │ │
│ │ │ ┌─────────────────────┐ │ │ │
│ │ └────────▶│ Config File Writer │◀─────────────┘ │ │
│ │ │ (multi-router .cfg)│ │ │
│ │ └─────────────────────┘ │ │
│ └──────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────────────────┐
│ KNOWLEDGE BASE LAYER │
│ │
│ ┌─────────────────┐ ┌─────────────────┐ ┌──────────────┐ │
│ │ vector_store.py │◀───────▶│ ChromaDB │◀───────▶│ Ollama │ │
│ │ │ │ (chroma_db/) │ │ Embeddings │ │
│ │ - add_documents│ │ │ │ │ │
│ │ - search │ │ Persisted │ │ nomic-embed │ │
│ │ - clear │ │ Vectors │ │ -text │ │
│ └─────────────────┘ └─────────────────┘ └──────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────────────┘
▲
│
┌─────────────────────────────────────────────────────────────────────────────┐
│ DOCUMENT PROCESSING LAYER │
│ │
│ ┌─────────────────┐ ┌─────────────────┐ │
│ │ pdf_processor.py│ │ web_scraper.py │ │
│ │ │ │ │ │
│ │ PDF ──▶ Text │ │ URL ──▶ Text │ │
│ │ Text ──▶ Chunks│ │ HTML ──▶ Chunks│ │
│ │ │ │ │ │
│ │ 1000 chars/chunk │ BeautifulSoup │ │
│ │ 200 char overlap │ Nav removal │ │
│ └────────┬────────┘ └────────┬────────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌─────────────────────────────────────────────────────────────────────┐ │
│ │ docs/ Directory │ │
│ │ srlinux.pdf, sros.pdf, routing_protocols_guide.pdf, etc. │ │
│ └─────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────────────────────┘
```
## RAG Pipeline Flow
```
┌──────────┐ ┌───────────────┐ ┌─────────────┐ ┌──────────────┐
│ User │ │ Embedding │ │ Vector │ │ Top-K │
│ Query │───▶│ (Ollama) │───▶│ Search │───▶│ Chunks │
└──────────┘ └───────────────┘ └─────────────┘ └──────┬───────┘
│
▼
┌──────────┐ ┌───────────────┐ ┌─────────────┐ ┌──────────────┐
│ Response │◀───│ LLM (Ollama) │◀───│ Prompt │◀───│ Context │
│ │ │ │ │ + Context │ │ Building │
└──────────┘ └───────────────┘ └─────────────┘ └──────────────┘
```
## Video Presentation
## Quick Start
### Prerequisites
- Python 3.8+
- [Ollama](https://ollama.ai/) installed and running
- 4GB+ RAM (8GB+ recommended)
### Installation
```bash
# 1. Clone and enter directory
cd nokia-srlinux-agent
# 2. Create virtual environment
python -m venv .venv
source .venv/bin/activate # Linux/Mac
# .venv\Scripts\activate # Windows
# 3. Install dependencies
pip install -r requirements.txt
# 4. Start Ollama
ollama serve
# 5. Pull required models (in another terminal)
ollama pull nomic-embed-text # Embedding model (required)
ollama pull llama3.2 # Default LLM
```
### Download Documentation
Place Nokia SR Linux PDFs in the `docs/` directory:
```bash
mkdir -p docs
# Download PDFs from Nokia documentation portal
# See docs section below for download links
```
### Run the Agent
```bash
# First run (builds vector database)
python main.py
# Force rebuild after adding new documents
python main.py --rebuild
```
## LLM Model Selection
### Model Comparison
| Model | Parameters | RAM Required | Speed | Quality | Best For |
|-------|------------|--------------|-------|---------|----------|
| `tinyllama` | 1.1B | 2-4 GB | Fastest | Basic | Testing, low-end hardware |
| `phi3:mini` | 3.8B | 4-6 GB | Fast | Good | Edge devices, quick responses |
| `gemma2:2b` | 2B | 4-6 GB | Fast | Good | Balanced performance |
| `llama3.2` | 3B | 6-8 GB | Fast | Good | **Default - recommended** |
| `mistral` | 7B | 8-12 GB | Medium | Better | General use |
| `llama3.1:8b` | 8B | 10-16 GB | Medium | Better | Multi-router configs |
| `gemma2:9b` | 9B | 12-16 GB | Medium | Better | Complex configs |
| `llama3.1:70b` | 70B | 48+ GB | Slow | Best | Maximum accuracy |
### Using Different Models
```bash
# Environment variable
export LLM_MODEL="tinyllama"
python main.py
# Or inline
LLM_MODEL="phi3:mini" python main.py
# For multi-router generation (need larger models)
LLM_MODEL="llama3.1:8b" python main.py
```
### Smallest Models for Low-End Hardware
```bash
# Absolute minimum (1.1B params, ~2GB RAM)
ollama pull tinyllama
LLM_MODEL="tinyllama" python main.py
# Good balance for low memory (3.8B params, ~4GB RAM)
ollama pull phi3:mini
LLM_MODEL="phi3:mini" python main.py
# Google's compact model (2B params)
ollama pull gemma2:2b
LLM_MODEL="gemma2:2b" python main.py
```
## Usage
### Interactive Commands
| Command | Description |
|---------|-------------|
| `stats` | Show knowledge base statistics |
| `reset` | Clear conversation history |
| `quit` / `exit` / `q` | Exit the agent |
### Example Session
```
$ python main.py
SR Linux Router Configuration Agent
============================================================
I can help you configure Nokia SR Linux routers!
You: How do I configure a BGP neighbor?
Agent: To configure a BGP neighbor on SR Linux, use the following commands:
enter candidate
/network-instance default {
protocols {
bgp {
autonomous-system 65001
router-id 10.0.0.1
neighbor 10.0.0.2 {
peer-as 65002
peer-group ebgp-peers
}
}
}
}
commit now
You: write configs for 3 routers with OSPF
Agent: I'll generate configurations for 3 routers...
[Writes router1.cfg, router2.cfg, router3.cfg to configs/]
```
## Documentation Sources
### Included PDFs (download to `docs/`)
| Document | Description | Download |
|----------|-------------|----------|
| `srlinux.pdf` | SR Linux base documentation | Nokia portal |
| `sros.pdf` | Service Router OS documentation | Nokia portal |
| `config_basics_guide_24.7.pdf` | Configuration basics | [Link](https://documentation.nokia.com/srlinux/24-7/books/pdf/Configuration_Basics_Guide_24.7.pdf) |
| `routing_protocols_guide.pdf` | BGP, OSPF, IS-IS | [Link](https://documentation.nokia.com/srlinux/23-3/books/pdf/Routing-Protocols-Guide_23.3.pdf) |
| `interface_config_guide.pdf` | Interface configuration | Nokia portal |
| `system_mgmt_guide_24.7.pdf` | CLI, gNMI, JSON-RPC | [Link](https://documentation.nokia.com/srlinux/24-7/books/pdf/System_Management_Guide_24.7.pdf) |
| `vpn_services_guide_25.7.pdf` | EVPN, MPLS, VPN | [Link](https://documentation.nokia.com/srlinux/25-7/books/pdf/VPN_Services_Guide_25.7.pdf) |
### Adding New Sources
Edit `main.py` to add PDFs or websites:
```python
PDF_PATHS = [
"docs/srlinux.pdf",
"docs/your_new_doc.pdf", # Add new PDF
]
WEB_URLS = [
"https://documentation.nokia.com/srlinux/24-7/", # Add website
]
```
Then rebuild: `python main.py --rebuild`
## Project Structure
```
.
├── main.py # CLI entry point
├── router_agent.py # RAG agent implementation
├── vector_store.py # ChromaDB + Ollama embeddings
├── pdf_processor.py # PDF text extraction/chunking
├── web_scraper.py # Website scraping
├── requirements.txt # Python dependencies
├── tests/ # Test scripts
│ ├── test_agent.py
│ ├── test_write_configs.py
│ ├── test_12_routers.py
│ └── test_parsing.py
├── docs/ # PDF documentation (gitignored)
├── examples/ # Example prompts and outputs
├── configs/ # Generated .cfg files (gitignored)
└── chroma_db/ # Vector database (gitignored)
```
## Technology Stack
| Component | Technology |
|-----------|------------|
| LLM Inference | [Ollama](https://ollama.ai/) |
| Embeddings | nomic-embed-text via Ollama |
| Vector Database | [ChromaDB](https://www.trychroma.com/) |
| PDF Processing | pypdf |
| Web Scraping | BeautifulSoup4 |
## Troubleshooting
### Connection Errors
```bash
# Ensure Ollama is running
ollama serve
# Check Ollama status
curl http://localhost:11434/api/tags
```
### Model Not Found
```bash
# List installed models
ollama list
# Pull missing models
ollama pull nomic-embed-text
ollama pull llama3.2
```
### Out of Memory
```bash
# Use a smaller model
LLM_MODEL="tinyllama" python main.py
# Or use quantized versions
ollama pull llama3.2:1b-q4_0
```
### Slow Performance
- Use smaller models (tinyllama, phi3:mini)
- Reduce chunk retrieval: edit `router_agent.py` to lower `top_k`
- Ensure you're using GPU acceleration if available
## License
MIT License - See LICENSE file for details.
## Resources
- [Nokia SR Linux Documentation](https://documentation.nokia.com/srlinux/)
- [Learn SR Linux](https://learn.srlinux.dev/)
- [Ollama](https://ollama.ai/)
- [ChromaDB](https://www.trychroma.com/)