Open-Source AI That Speaks 3GPP Fluently

Tools
, ,
1

“Ask Anything About 3GPP -This AI Has the Answers”

“Turn Claude into a Senior Telecom Consultant (Open Source).”

I have developed a tool that I wish I had years ago in the telecom industry.

Introducing the 3GPP Expert Skill for Claude — an open-source plugin that transforms Claude AI into a senior telecom consultant. You can ask it anything about mobile networks, and it provides standards-grounded answers with actual specification references.

Here’s what it covers:

  • Every generation: 2G/GSM through 6G
  • Every release: Phase 1 (1992) to Release 21
  • Full protocol stacks: PHY, MAC, RLC, PDCP, SDAP, RRC, NAS — including LTE vs NR differences
  • Core network architecture: EPC to 5GC/SBA, encompassing all network functions
  • Security: EPS-AKA, 5G-AKA, SUPI/SUCI, IMSI catcher analysis
  • Deployment: Network planning, spectrum strategy, NSA/SA migration paths, O-RAN

Some examples of its capabilities include:

  • “How does a UE scan and select the best cell from power-on to RRC Connected?” — It details the entire flow, including PSS/SSS, SIBs, cell selection criteria, RACH, RRC setup, measurement events, all with TS references.
  • “How does 5G protect against IMSI catchers?” — It explains SUPI/SUCI, compares 5G-AKA to EPS-AKA, and addresses the remaining gaps.
  • Deployment planning: “We’re migrating from LTE/EPC to 5G on a budget. Walk me through the NSA vs SA options.”

This tool is completely free and open source (MIT license).

If you work in telecom, I encourage you to give it a try and share your thoughts. Feedback, contributions, and stars are all welcome.

Open-Source AI That Speaks 3GPP Fluently
Open-Source AI That Speaks 3GPP Fluently1024×1536 448 KB

Github: :backhand_index_pointing_down:

LinkedIn: :backhand_index_pointing_down:

1 Like
2

Last week I shared a 3GPP Expert Skill for Claude and the response was amazing. :raising_hands:

Engineers tried it, broke it, and gave me great feedback.

Biggest catch? Ask “What sequence is used for PSS in 5G NR?” and it answered Zadoff-Chu. Wrong — NR PSS uses m-sequences. ZC is for PRACH in NR, not PSS.

Classic AI hallucination — sounds right, uses correct jargon, wrong signal.

v1.1.0 is out now:

  • Fixed the PSS bug with hard-coded critical PHY facts
  • New PHY layer reference (sync signals GSM→NR, channels, RACH, reference signals)
  • New Working Groups reference (RAN/SA/CT structure, spec ownership)
  • 8 targeted test cases including trap questions — all pass

Open source, free. Keep the feedback coming.

GitHub - lugasia/3gpp-skill · GitHub

LinkedIn: :backhand_index_pointing_down:

1 Like
3

EDUNAVO team and Cafetele is already developed this kind of Personal AI agent ..

But keep in mind guys AI tools is for domain expert not for beginners.

Why I am saying this?
If you know the prompt expected output then make sense to use as you can ask AI to fix incase wrong response.

Else if you blindly rely on then you r filling the garbage.

Currently no existing AI reach such a intelligence where itcan solved critical stack level issue..

So use it carefully..

1 Like