OnCall NetOps 🚀

In this repository, you can find the code for a Langgraph-based agents that collaborates to diagnose network issues.

It’s part of a larger demo using a TIG stack for monitoring and alerts, along a pyATS server to interact with the network devices.

To view the rest of the components:

• Full demo: https://github.com/jillesca/oncall-netops-tig-pyats-demo
• pyATS Server: https://github.com/jillesca/pyats_server
• Observablity stack: https://github.com/jillesca/observablity_tig

Graph ⚙️

When receiving a request, the node_coordinator validates the info provided,
then passes it to the node_orchestrator, which decides which network agents to call. Each agent connects to devices, gathers data, and returns a report.

Network agents:

• agent_isis
  • Retrieves ISIS info.
• agent_routing
  • Retrieves routing info.
• agent_log_analyzer
  • Checks logs.
• agent_device_health
  • Retrieves device health.
• agent_network_interface
  • Retrieves interfaces/config.
• agent_interface_actions
  • Performs interface actions.

When all agents finish, their reports go to the node_root_cause_analyzer, which determines the root cause. If more info is needed, it requests it from the node_orchestrator. Otherwise, final findings go to the node_report_generator.

Short memory is implemented (per thread) that lets agents recall past investigations to resume work, so users can ask about events.

Requirements ⚠️

• Python 3.11 (Only for the Langgraph Studio Desktop version).
• OpenAI Key.
• Langsmith Key, Create a token and copy the langsmith env vars.
• pyATS Server URL.

Create an .env file and set your keys there.

LANGSMITH_TRACING=true
LANGSMITH_ENDPOINT="https://api.smith.langchain.com"
LANGSMITH_API_KEY=<langsmith_token>
LANGSMITH_PROJECT="oncall-netops"
OPENAI_API_KEY=<openai_token>

Setup 🛠️

There are two options to run the graph.

1. Run the langgraph server cli on the terminal, no container. You can use the web version of langgraph studio (a bit slower).
2. Using the Langgraph Desktop version (Only Mac). Container based.

Set environment variables

• PYATS_API_SERVER: Connects the Langgraph API server to the pyATS server. You must set this environment variable. Add it to an .env file. Default port in the pyats_server project is 57000.
• LANGGRAPH_API_HOST: Links the grafana-to-langgraph-proxy with the Langgraph API server. Defaults to http://host.docker.internal:56000, adjust if needed.

See the .env.example file to find the rest of environment variables used. These are set by the Makefile.

Build 🏗️

Validate and build the base environment. If an env var is missing, the script fails.

make build-environment

Start the grafana-to-langgraph-proxy component in a separate terminal. See Appendix for more info.

make build-proxy

Option 1. Langgraph Server CLI 💻

Install the dependencies listed on the requirements file, use a virtual environment if possible.

Start the server with.

make run-environment

❯ make run-environment
langgraph up --port 56000 --watch --recreate
Starting LangGraph API server...
For local dev, requires env var LANGSMITH_API_KEY with access to LangGraph Cloud closed beta.
For production use, requires a license key in env var LANGGRAPH_CLOUD_LICENSE_KEY.
- Pulling...Error response from daemon: get langgraph-data: no such volume
Ready!
- API: http://localhost:56000
- Docs: http://localhost:56000/docs
- LangGraph Studio: https://smith.langchain.com/studio/?baseUrl=http://127.0.0.1:56000

Open the LangGraph Studio URL using Chrome (Firefox doesn't work).

If you have issues with the web version make sure:

• You are logged in langsmith.
• Refresh your browser.

If you don't want to use the web version, you can still see the operations in the terminal, but is hard to follow with so much output.

Option 2. Langgraph Studio Desktop 🍏

Download the desktop version (only for Mac).

Before you start opening the project, set the target port in the bottom bar. This project uses port 56000, if you set a different one, update the env var LANGGRAPH_API_PORT

On langgraph studio, select the project and open it. This imports the code from this repo and install everything in a dedicated container.

Run 🚀

You can interact with the graph via GUI or API:

1. GUI. Use the User Request field to ask about network devices. For a first-time query, provide the device hostname (one thing to improve).

2. API. This cURL starts a background thread, you won’t see the progress until it finishes. Check logs for real-time updates.

Once completed, select the corresponding thread to view the graph’s execution. If the thread is empty in the GUI, switch to another thread to refresh the view.

curl http://localhost:56000/runs \
  --request POST \
  --header 'Accept: */*' \
  --header 'Content-Type: application/json' \
  --data '{
    "assistant_id": "oncall-netops",
    "input": {
      "incident_description": "How cat8000v-2 is doing?"
    },
    "on_completion": "keep",
    "on_disconnect": "continue",
    "after_seconds": 1
  }'
{"run_id":"1eff2ee5-c3a9-6234-9836-3cc708aa151c","thread_id":"30771945-13eb-4530-88e8-678755547f0e","assistant_id":"dd5719bd-2607-51d4-b079-9621fe24d378","created_at":"2025-02-24T20:31:44.535531+00:00","updated_at":"2025-02-24T20:31:43.535531+00:00","metadata":{"assistant_id": "dd5719bd-2607-51d4-b079-9621fe24d378"},"status":"pending","kwargs":{"input": {"incident_description": "How cat8000v-2 is doing?"}, "config": {"metadata": {"graph_id": "oncall-netops", "created_by": "system", "assistant_id": "dd5719bd-2607-51d4-b079-9621fe24d378"}, "configurable": {"run_id": "1eff2ee5-c3a9-6234-9836-3cc708aa151c", "user_id": "", "graph_id": "oncall-netops", "thread_id": "30771945-13eb-4530-88e8-678755547f0e", "assistant_id": "dd5719bd-2607-51d4-b079-9621fe24d378", "langgraph_auth_user": null, "langgraph_auth_user_id": "", "langgraph_auth_permissions": []}}, "command": null, "webhook": null, "subgraphs": false, "temporary": false, "stream_mode": ["values"], "feedback_keys": null, "interrupt_after": null, "interrupt_before": null},"multitask_strategy":"reject"}%

Traces 🔍

Since I'm using langsmith, you can inspect the traces of the graph execution. There you can find state, runs, inputs and outputs.

• Graph triggered by an automatic alert: Trace
• Graph triggered by an user request following up on the alert: Trace

Appendix 🧩

grafana-to-langgraph-proxy 🔗

The grafana-to-langgraph-proxy translates Grafana’s alert webhook to a format compatible by a langgraph API server, launching a background job. The proxy listens using the environment variable LLM_PROXY_URL.

The environment variables LANGGRAPH_ASSISTANT_ID and LANGGRAPH_API_ENDPOINT are used to create a new payload and connect to the langgraph server.

See the http://<SERVER_API>:56000/docs to get more information about the Langgraph API.

See the README file of the proxy for more details.

• grafana-langgraph-proxy logs

  docker logs -f grafana-langgraph-proxy

Useful resources 📚

• Building effective agents by Anthropic
• Course. Introduction to LangGraph
• Scientific paper agent using LangGraph