The tools could predict Update PCB but someone had to press Validate, read the check column and press Execute. Running the real ECO showed the prediction got two things wrong. The PR drives the dialog through Windows UI Automation.
Recurring agent failure issue in altium-mcp (coffeenmusic)
What engineers report
A recurring agent failure problem in altium-mcp (coffeenmusic) that mostly hits the electrical engineer.
Failure pattern: agent / failure
Analyst note
From the Grok pain-radar digests, Nov 9-10 2026 (Nov 9 7:41 AM ET through Nov 10 7:41 AM ET digests plus the Haitch editions). Verified live 2026-11-10: issue/PR read; title and state confirmed.
What it costs a team
Modeled annual cost
$30,000
For a team of 5
Prototype delay
3 days
Modeled schedule slip per occurrence
Time lost
240 min
Per engineer per month, about 1 incidents
How the cost is modeled
240 min/month × 5 engineers × 12 months = 240 hours × $125/hour = $30,000
A modeled estimate with medium confidence. Swap in your own team size and loaded rate to size it for your team.
Engineering tools that address it
No engineering tool profile lists this error yet. Browse tools that connect to altium-mcp (coffeenmusic).
Evidence
More recurring errors in altium-mcp (coffeenmusic)
- Adds run_drc so an agent can check its own board edits in Altium. Altium batch DRC call fails for every argument list tried, the scripted DRC opens a modal dialog that waits for a person to press Run, and a cancelled check looks clean. The workaround presses the button from outside Altium with posted mouse messages, confirms the press by the dialog closing, and cross-checks against the .drc report. Commands also now take an explicit board_path instead of acting on whichever board has focus.ECAD · impact 74
- After in-place sheet edits by the agent tools, the check was a compile followed by a hand-written dump of the netlist, compared by eye. Yet that check is the safety net for in-place edits: a label dropped one grid off a wire end silently splits a net. The PR adds a before/after netlist diff with snapshots.ECAD · impact 74
- On a real sheet, a port had auto-sized to a shorter name, so its end no longer touched the wire meeting it. The wire ended inside the port's body, connected to nothing, and compiling did not flag it. The existing checker only checked drafting, and its fixes had to be turned into edit operations by hand.ECAD · impact 74
- A wire a few microns off the pin looks right and connects to nothing. Ports auto-size to their names, so a rename moves the pin end off the wire, and compile did not flag one such case on a real sheet. The PR lets agents wire to 'U20.VCC' by reference and adds a fixed port width.ECAD · impact 66
- Adding parts to an Altium DbLib meant editing its Excel table by hand or script. A stale lock file looked like a real lock, and a script that wrote values as text into a column of numbers left cells that Altium reads as empty, with no warning.ECAD · impact 66
- Editing the .PrjPcb by hand did not work while the project was open because Altium keeps its own copy of the options and writes it back on save. Changing a cross-reference option meant editing the file, closing the project and reopening it, and nothing showed which sheets generate rooms.ECAD · impact 66
Sourced from public community reports. Costs and delays are modeled estimates, not measurements. Know a better source or a fix? Tell us at hello@productflo.io.