Let an agent debug chip designs in a live Verdi session
A developer built a connector letting an AI agent run Synopsys Verdi, a chip-debugging program, to trace design bugs; he reports testing it on four designs.
Done with Other
What they did
A developer built a connector that lets an AI agent control a running copy of Synopsys Verdi, a tool for debugging chip designs. It works on the same computer or a remote one. He asked the agent in plain language to investigate bugs, such as why a buffer's "full" signal stayed on. The agent picked signals to watch, jumped back to before the error, studied the signal timelines and found the design code causing it.
How it went
The author shares a 1m38s demo where the agent, not a pre-recorded script, chooses each step. He reports it worked on four real chip designs, with recordings up to 2.7 GB, and includes a blind debugging trial.
Worth knowing
It is experimental (version 0.3.3) and needs a paid Verdi license. Connecting to any already-open Verdi window has not yet been tested on its own.
Try it yourself
Using the Verdi Bridge MCP tools connected to my Verdi session for [design name], investigate [describe failure, e.g. why signal X stayed high]. Read the simulation log to find the error time. Add the relevant signals, set the cursor [N] cycles before the error, and inspect their transitions. Trace the drivers of the suspicious signal into the RTL source. Report your findings with times, signal values and file/line references. Do not modify any RTL.
Swap the highlighted [parts] for your own details.
Source: GitHub · RuihongY · Oct 13, 2026
Written with AI assistance from the original post and checked against our editorial guidelines. Results are as reported by the source, not independently verified.
