InfraLens v2.0.0: see what’s actually talking to what in your cluster, via eBPF

I maintain InfraLens, an open-source (Apache-2.0) tool that uses eBPF to discover and visualize service-to-service traffic on Kubernetes clusters and plain Linux servers. No sidecars, no SDKs, no code changes — the agent hooks kernel functions like tcp_v4_connect and udp_sendmsg, so it sees every TCP/UDP connection a node makes and puts it on a live map within seconds of deploying.

Just tagged v2.0.0. What’s new:
• Demo mode (DEMO_MODE=true) — a built-in topology simulator so you can try the full UI in about 30 seconds, no Linux/eBPF/agents required. Runs fine on macOS/Windows via Docker Compose.
• UDP tracing — DNS, StatsD, syslog and other UDP flows now show up as dashed edges with a protocol badge, alongside TCP.
• Topology search — press /, search by name, IP, technology, or node.
• Delta WebSocket updates — used to be a full topology snapshot every 2s, now it’s incremental per-entity deltas with periodic re-sync. Much lighter on the backend and your browser tab.
• Graph export — pull the live topology out as Mermaid or Graphviz DOT.
• Agent auth that actually works — –api-key / INFRALENS_API_KEY, plus HTTPS backend URLs, so API_KEY on the backend is finally enforceable end to end.

A few other things it does, for context: fingerprints services (Postgres, Redis, Kafka, Mongo, Nginx, etc.) from ports and process names, can probe deeper for HTTP headers/DB handshakes, reads go.mod/package.json/requirements.txt for dependency info, and has an optional per-service AI doc generator (OpenAI/Anthropic/Gemini, or fully local through Ollama/LM Studio if you’d rather not send anything off-box).

Stack: Go for the agent and backend (using cilium/ebpf), React Flow on the frontend. The agent is CO-RE, so it’s compiled once and runs across kernel 5.8+.

It’s a young project and there’s a real roadmap ahead — no historical/time-travel view yet, no anomaly detection, no L7 request tracing. Bug reports, feedback, and PRs are genuinely welcome, and if it’s useful to you, a star helps other people stumble onto it.

Repo + install instructions: https://github.com/Herenn/Infralens

Happy to answer questions about the eBPF side, the k8s IP-to-pod resolution, or anything else.