gokeshenzhen

gokeshenzhen / waveform_mcp

Public

MCP server for Claude Code to debug simulation failures via log parsing and waveform analysis (FSDB/VCD)

15
3
100% credibility
Found Mar 02, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This repository creates a specialized assistant for AI coding tools to automatically analyze hardware simulation logs and waveform files, extracting errors and signal behaviors to enable autonomous debugging of verification test cases.

How It Works

1
🔍 Struggling with test failures

You're running hardware tests and they keep failing with confusing error messages in logs and signal charts.

2
📥 Get the analysis helper

Download this simple helper that makes sense of those test logs and signal drawings.

3
🔗 Connect to your AI buddy

Link the helper to your friendly AI coding assistant so it can use special analysis powers.

4
▶️ Run a failing test

Start one of your test cases, and when it fails, note where the logs and signal files are saved.

5
💬 Ask AI to fix it

Simply tell your AI: 'Debug this test until it passes,' and it takes over.

6
🧠 AI investigates deeply

Your AI reads the errors, searches signals in the charts, and figures out exactly what went wrong around each failure moment.

✅ Tests pass perfectly

The AI tweaks your code step by step until everything runs smoothly with no errors left.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 15 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is waveform_mcp?

Waveform_mcp is a Python MCP server that plugs into Claude Code, letting AI autonomously debug hardware simulation failures by parsing VCS or Xcelium logs and analyzing FSDB or VCD waveforms. Tell Claude to "develop case0 and debug until it passes," and it fetches paths, extracts errors like assertion fails or UVM_FATALs, searches signals by name, and correlates log timestamps with waveform snapshots or histories. You configure once via ~/.claude.json with tool paths, then invoke tools like parse_sim_log or analyze_failures through mcp server calls.

Why is it gaining traction?

It stands out by closing the loop on AI-driven verification: Claude iterates simulations, diagnoses root causes with joint log-waveform analysis, and suggests RTL or testbench fixes without manual Verdi dives. Custom error patterns load from YAML—no code changes—and it handles GB-scale FSDB via efficient indexing. Solid pytest coverage and a clear mcp server tutorial make onboarding fast for mcp github python users.

Who should use this?

ASIC/FPGA verification engineers running UVM regressions with VCS/Xcelium, tired of sifting irun.log and top_tb.fsdb manually. Teams using Claude Code in VSCode for RTL debug workflows, especially those chasing assertion fails or signal mismatches in testcase/case0.sv.

Verdict

Promising niche tool for mcp server ai in waveform mcp analysis, with thorough docs, unit tests, and easy adaptation via config.py constants—but at 13 stars and 1.0% credibility, it's early-stage; test it on your flow before production reliance.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.