| Platforms | |
| Transport | |
| Integrations |
RAG-powered network QA investigation tool for multi-vendor networks.
Run your tests with any framework. When something fails, YANA investigates - it queries live devices, searches protocol specs, follows diagnostic decision trees, and tells you why it failed and what to fix.
Supported models:
- Haiku 4.5, Sonnet 4.6, Opus 4.6 (default, best reasoning)
Documentation:
- WORKFLOW.md - operational flow
- GUARDRAILS.md - security controls
What's new:
- CHANGELOG.md - latest version
| Technology | Role |
|---|---|
| Python | Core language |
| FastMCP | MCP server exposing 8 tools |
| Claude | Reasoning, context, investigation |
| LangChain | RAG pipeline (chunking, embedding, retrieval) |
| ChromaDB | Vector database for knowledge base |
| Scrapli | Multi-vendor SSH transport |
| JUnit XML | Test results format |
| Protocol | What's Checked |
|---|---|
| OSPF | Adjacencies, area and area types, config, LSDB |
| Routing | Table, route maps, prefix lists, PBR, ACLs, ECMP |
| Interfaces | Up/down state, expected operational status |
Network diagram:
Lab environment:
- 16 devices defined in TOPOLOGY.yml
- 5 x Cisco IOS, 3 x Cisco IOS-XE, 4 x Arista cEOS, 2 x MikroTik CHR, 1 x Juniper JunOS, 1 x Aruba AOS-CX
- See lab_configs for my test network's configuration
Prerequisites: Python 3.11+
Step 1 - Install and ingest:
sudo apt install git make python3.12-venv -y
cd ~ && git clone https://github.com/pdudotdev/YANA
cd YANA && make setupStep 2 - Configure credentials:
cp .env.example .env
# Edit .env with your device credentialsStep 3 - Authenticate with Claude:
claude auth loginStep 4 - Register the MCP server:
claude mcp add yana -s user -- /home/<user>/YANA/yana/bin/python /home/<user>/YANA/server/MCPServer.py| Tool | Description |
|---|---|
get_status |
Report active data sources (inventory, intent, KB) |
list_devices |
List inventory devices, optionally filtered by CLI style |
get_ospf |
Query OSPF state (neighbors, LSDB, process config) on a live device |
get_interfaces |
Query interface up/down status on a live device |
get_routing |
Query routing table, route maps, prefix lists, PBR, and ACLs on a live device |
query_intent |
Retrieve network design intent from local JSON |
traceroute |
Trace the forwarding path from a device to a destination IP |
search_knowledge_base |
Search network knowledge base (RFCs, vendor guides) with protocol/vendor/topic filters |
YANA is designed to work with your own test topology. Bring your own:
| What | Where | Format |
|---|---|---|
| Device inventory | data/NETWORK.json |
JSON dict: device name -> host, platform, cli_style |
| Design intent | data/INTENT.json |
JSON: router roles, OSPF areas, links, neighbors |
| Protocol docs | docs/*.md |
Protocol data (re-ingest with make ingest) |
| Test results | results/*.xml |
JUnit XML from any test runner |
| Credentials | .env |
ROUTER_USERNAME and ROUTER_PASSWORD |
- Run your tests with any framework (pytest, pyATS, Ansible, etc.)
- Place JUnit XML results in
results/ - Investigate failures with the
/qaskill:
claude
> /qa
The /qa skill loads the latest JUnit XML, lists all failures, and investigates each one using live device queries, network intent, and RFC context.
Protocol documentation lives in docs/ as Markdown files. Each file is tagged with vendor, topic, and protocol metadata during ingestion.
To update after editing docs:
make ingestRAG optimizations in place:
protocolmetadata field - filters search by protocol (ospf, bgp, eigrp), eliminating cross-protocol noise- Contextual chunk headers - source and protocol prepended to each chunk for better embedding quality
- Compound filtering - combine vendor, topic, and protocol filters in a single query
See OPTIMIZATIONS.md for the full optimization roadmap.
YANA/
├── server/
│ └── MCPServer.py # FastMCP server (8 tools)
├── tools/ # MCP tool implementations
│ ├── ospf.py # get_ospf
│ ├── operational.py # get_interfaces, traceroute
│ ├── routing.py # get_routing
│ ├── intent.py # query_intent
│ ├── status.py # get_status
│ ├── inventory_tool.py # list_devices
│ └── rag.py # search_knowledge_base
├── core/ # Infrastructure
│ ├── inventory.py # Device dict (from data/NETWORK.json)
│ └── settings.py # Credentials + SSH config
├── data/ # Data files
│ ├── NETWORK.json # Device inventory
│ ├── INTENT.json # Network design intent
│ └── chroma/ # ChromaDB vector store (generated)
├── transport/ # SSH execution (Scrapli)
│ ├── __init__.py # Command dispatcher + semaphore
│ └── ssh.py # Multi-vendor SSH executor
├── platforms/ # Vendor CLI command mapping
│ ├── platform_map.py # OSPF, routing, interface + traceroute commands (6 vendors)
│ └── definitions/ # Custom Scrapli definitions
├── input_models/
│ └── models.py # Pydantic validation (OspfQuery, KBQuery, etc.)
├── docs/ # Knowledge base (RFCs + vendor guides)
├── results/ # QA test results (JUnit XML)
├── skills/
│ ├── ospf/ # OSPF diagnostic decision tree
│ └── routing/ # Routing diagnostic decision tree
├── testing/
│ ├── automated/ # Unit + integration tests
│ ├── live/ # Live lab tests
│ └── run_tests.sh # Test runner (--live for lab tests)
├── metadata/
│ ├── guardrails/ # Security controls documentation
│ ├── scalability/ # RAG optimization roadmap
│ ├── topology/ # Test network diagram
│ └── workflow/ # RAG pipeline walkthrough
├── ingest.py # RAG ingestion pipeline
├── Makefile # Setup automation (make setup)
├── CLAUDE.md # Investigation workflow + tool reference
├── TOPOLOGY.yml # Containerlab topology definition
├── .env.example # Credential template
├── CHANGELOG.md # Version history
├── LICENSE
├── requirements.txt
└── README.md
- EIGRP and BGP support (tools, skills, knowledge base docs)
You are responsible for defining your own network inventory and design intent, building your test environment, and meeting the necessary prerequisites (Python 3.11+, Claude CLI/API, network device access).
Licensed under GNUv3.0.
Interested in collaborating?
- Email: [email protected]
- LinkedIn: LinkedIn