The goal is “PCAPs for humans”: Overview dashboard answers what’s happening + what to click next
Domains view (hostnames first) → select a domain → jump straight to relevant flows (works even when DNS is encrypted/cached by using observed IPs from flows)
Weird stuff view surfaces common failure/latency signals (retransmits/out-of-order hints, resets, handshake issues, DNS failures when visible)
From there you can drill down: Flows → Packets → Explain (plain-English hints) / follow stream
Commands: Offline: babyshark --pcap capture.pcap
Live (requires tshark): babyshark --list-ifaces then babyshark --live en0
Repo + v0.1.0 release: https://github.com/vignesh07/babyshark
Would love feedback on UX + what “weird detectors” you’d want next.
---
Some UX bits I noticed after playing around for a few minutes:
- Esc for backwards navigation was not obvious for me. Maybe emphasize that somehow, and/or support Backspace too for backnav?
- Enter on Domains menu item does not work
- don't mention clicking if mouse is not supported. "Select" would be more appropriate
- packets screen is truncated vertically and horizontally. Probably should be scrollable
- "weird stuff" options are numbered 1-5, but pressing those keys has no effect. There's lots of little polish fixes like this.
---
And then things I wonder about as a novice user:
- Is it possible to see domain names instead of IP addresses while e.g. looking at packets?
- What does it mean to f stream?
- How do I inspect packets? Especially compressed or encrypted data? This is more a knowledge gap, like "what am I supposed to look for", "what could be in a packet", and I guess involves reverse engineering sometimes, but it's also a tooling question.
• Back navigation: good call. I'll make "Esc back" more explicit everywhere and add Backspace as an alias for back (and mention it in h help).
• Enter on Domains not working: it should drill down to flows. If you can share your OS/terminal + whether you installed from release vs cargo install, I'll try to reproduce and fix in the next release.
• "Clicking" wording: agreed — mouse isn't supported right now. I'll change Ul copy to say Select (and keep "Enter = drill down").
• Packets screen truncation: yep, needs scrolling/paging. On the list views I already do r/; l'll add page scroll and horizontal handling where it makes sense.
• Numbered weird options: great idea - I'll map 1..9 to jump-select and Enter (or open directly).
• Domain names in packets: yes, I want that. I already collect DNS/SNI/HTTP host hints; next step is showing hostname labels alongside IPs when I have more confident mapping.
• "f stream" definition: I'll clarify it as "Follow stream (reassembled payload)" and add a glossary/help entry.
• Inspecting encrypted/compressed data: totally fair. The tool can't decrypt TLS without keys, but it can make it clearer what's happening (SNI/ ALPN, sizes, timing, resets/retransmits). I'Il improve "Explain" to say what's possible vs not.
Super useful feedback thanks!
what the kids in my classes really struggle with is actually using any command line stuff (at least for a month or two), because it is so foreign to them (coming from GUI-only experience).
what specific parts are made easier with babyshark, compared to wireshark? the github readme didnt really sell me on the "easier than GUI" part, nor did your description here. is it the "explain (plan-English hints)" part? if so, i think you should focus on that. right now it looks pretty bare bones (e.g. "Weird stuff" does not seem easier or super helpful from a learning perspective)
CLI is so valuable because rather than explore a presentation of the data you plan your RE etc and then run it and it either returns the answer or it doesn't.
There are some TUIs I quite like (LNAV as a pager) but I think if you really know what you're dealing with the CLI is better almost every time.
Built it in Java, and rebuilt it in Ada, and Rust. I find it's a good exercise to learn about a programming language... bonus point, once I have a parser, plugging it live behind libpcap, dpdk, xdp, or just raw sockets is easy.
indeed! command line is great.
however, ~99.8% of 18 year old students have never used any command line tool in their lives. they do not know what grep is. they can navigate a gui because they have used a gui all of their life.
when im teaching networking for example, using a gui means i only need to teach one thing (networking), where if i use a cli i have to teach two things (cli + networking)
>I think if you really know what you're dealing with the CLI is better almost every time
to be clear, i was not making an argument that gui is better in general.
i am speaking as someone who teaches introductory networking courses at a 1st-year college level. no one i teach "really knows" what they are dealing with because it is the first time they are learning about it.
- *Overview dashboard*: immediately surfaces top talkers/flows + “what should I click next” instead of dropping you into the full packet list. - *Domains-first pivot*: `D` shows hostnames and lets you jump from a domain → the relevant flows. It also works when DNS answers aren’t visible (DoH/DoT/cached) by using observed IPs from SNI/Host flows. - *Weird stuff*: `W` is a curated set of “likely problems” (retransmits/out-of-order hints, resets, handshake issues, DNS failures when visible) with a short “why it matters” and a drill-down. - *Explain*: `?` gives plain-English hints for a selected flow + suggested next steps (follow stream, filter, pivot to domains/weird).
So it’s basically a guided triage layer on top of tshark/pcap data, with the “where do I start?” path baked in.
If you’ve got a specific teaching use-case (e.g. “why is this slow?” or “which host is generating traffic?”), I’d love to tune the Overview/Weird detectors around that. Open to PRs as well.
i think there is definitely room for something like this, it just (at first glance from the readme at least) seems like the guided part of this tool is bolted on as a bit of an after thought.
it feels like you are currently in an odd position where the user is expected to know the networking jargon already, be able to recognize that something might be "weird" at a glance, but also not know how to drill down into the data. i think that is probably a small overlap of people.
if i were you, i would lean all-in on making it a learning tool.
>If you’ve got a specific teaching use-case (e.g. “why is this slow?” or “which host is generating traffic?”), I’d love to tune the Overview/Weird detectors around that.
i will put some thought into some real-world examples of what i would be interested in, from a teaching perspective. your post caught my eye because i am starting my wireshark module next week, so it is certainly timely.
(digs out git ...)
Looking over the commit history of this project, I'm about 90% sure it was entirely done with a AI Coding Agent, and not even a very good one.
I have a business partner who sounds like a TV evangelist when it comes to vibe coding but if he gets results then I am all in. He has got quite a lot of results in a few months on a project but he has certainly put the miles in himself.
The key is to use the tool appropriately. Don't blindly allow it to do what it likes but guide it all the way using your experience and knowledge.
Anyway, we now have tsharkrs to add to gotshark!
As an aside, I was thinking about something similar to this tool for a while now after seeing this post (https://news.ycombinator.com/item?id=46723990) where someone was using Claude to troubleshoot a PCAP. It made me think that it would be nice just to have a nice collection of tshark one-liners to quickly weed out any weird stuff right off the bat. I would assume that it would be a lot more performant than using a LLM and more scalable if you have large PCAP files.
Well, they’re naturally interested in anything daddy does. Especially with the computer!
tshark is the engine; Babyshark is the guided Ul on top of it. • tshark: raw packet/field dump + powerful filters, but you have to know what fields to ask for and how to stitch the story together. • Babyshark: gives you an opinionated workflow (Overview → Domains/Weird → Flows → Packets/ Stream) with "explain/why it matters" text, curated detectors, and one-key drilldowns.
For live capture, Babyshark uses tshark -T fields to extract things like DNS qname / TLS SNI / HTTP host; for offline PCAP it parses enough to build flows + summaries.
So: if you already live in tshark one-liners, tshark is faster. If you're trying to understand what's happening or teach/debug quickly, Babyshark is a nicer front-end.
