Last hour's #171 gave agents real pin-to-net links from a bare .kicad_sch. Importing the matching .kicad_pro broke them again without any error. The project importer prefixed net keys by sheet but left the pin and net IDs unprefixed. On real boards that was 128 broken pin-to-net references (ESP32-S3), 99 (SparkFun Qwiic Navigation) and 229 (Mitayi Pico D1). Fourth zaptrace PR on this tonight.
Recurring agent failure issue in zaptrace (oaslananka)
What engineers report
A recurring agent failure problem in zaptrace (oaslananka) that mostly hits the electrical engineer.
Failure pattern: agent / silent failure
Analyst note
From the Grok pain-radar digests, Nov 11-12 2026 (Nov 11 8:41 AM ET through Nov 12 5:41 AM ET digests plus the Haitch editions). Verified live 2026-11-12: 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 zaptrace (oaslananka).
Evidence
More recurring errors in zaptrace (oaslananka)
- A KiCad importer checked against KiCad's own netlists matches 0 of 64 net names on one real board. zaptrace added a gate comparing its imported graph with 'kicad-cli sch export netlist' output: SparkFun Qwiic Navigation has 32 native nets and 47 imported with 11 exact shared names and 5 pin mismatches, while Mitayi Pico D1 has 64 native and 95 imported with 0 exact shared names. Both are marked BLOCKED.ECAD · impact 74
- A KiCad schematic importer silently merged USB_D+ and USB_D- into one net. zaptrace normalized net names into internal keys, and both became 'usb_d_'. On a real reference board (Mitayi Pico D1) the import came out with 94 nets instead of 95, and the quality score read 1.0106 on a scale that tops out at 1.ECAD · impact 74
- A KiCad review tool adds a step that writes an assistant request bound to one selected DRC finding, the exact PCB hashes and the review state, because copying a free-form chat prompt loses the selected rule, exact PCB state and non-approval boundaries. Reviewers previously saw only DRC counts and could not tell which rule fired.ECAD · impact 66
- After last hour's net fix, the importer still exposed empty pin lists on a real ESP32-S3 board, even for the MCU. An AI design agent could see 47 nets but could not ask which pin is EN, what type it is, or which net it sits on. The PR fills 128 pin records from KiCad's embedded symbol definitions and links each one to an existing net. That is the third zaptrace PR tonight on the same board.ECAD · impact 66
- Importing a real ESP32-S3 dev board schematic silently dropped 3 LED anode nets that KiCad's own netlist had: 44 nets and 126 pin nodes vs KiCad's 47 and 132. KiCad allows two parts' pins to touch at the same point with no wire, and the importer only indexed wire endpoints. The tool's own verdict across its 20-board corpus is still 0 verified, 16 blocked, 4 not run.ECAD · impact 66
- More netlist-parity work against KiCad 10.0.8. This one makes the importer match KiCad's handling of intentionally unconnected pins (6 on a SparkFun board, 3 on a Mitayi board) and keep power and ERC flag symbols out of the physical pin list. It fails closed when a no-connect marker is ambiguous rather than inventing a net. Fifth zaptrace PR today chasing the same goal, a netlist that matches what KiCad itself exports.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.