easyeda-mcp-pro (oaslananka) errors

51 recurring problems engineers report in easyeda-mcp-pro (oaslananka), ranked by impact.

#5 by errors trackedECADMCADVerification

Recurring easyeda-mcp-pro (oaslananka) errors, by impact

  1. 1A dual op-amp showed up as two schematic symbols, TL072CP.1 and TL072CP.2, and the read-only BOM bridge counted each one, reporting quantity 2 instead of 1 and risking an incorrect purchasing list. The fix counts a verified physical identity once and fails closed with BOM_IDENTITY_AMBIGUOUS when designators or IDs conflict.ECAD · Agent failure · 1 report74
  2. 2A registered SCH_Event.addPrimitiveEventListener produced no callback during hand-drawn wire creation or wire dragging, and EasyEDA's own docs mark that API beta and not for production. Real schematics also hold repeated vertices and zero-length segments, and the old geometry helper treated every pin in the document as touching them, so a truly unwired part could not move. The fix compares two native readers over two samples 80 ms apart and refuses on any mismatch.ECAD · Agent failure · 1 report74
  3. 3A user building a thin single-sided wearable says the standard sign-off is a look at EasyEDA's 3D view for body collisions, tall parts against the enclosure, and connector and flex exits, and right now an agent cannot help with this at all. The only proxy is a 2D capture of the component-shape layer, which misses parts without shapes and says nothing about height.MCAD · Agent failure · 1 report74
  4. 4EasyEDA Pro 3.2.209 returns matching but stale wire lists from both getAll calls right after a wire is created or deleted, so a component move looks unwired and succeeds even though a connected pin would be disconnected or put on the wrong net. The fix stops trusting the batch list, tracks writes before they start, re-reads each wire by id, and refuses the move if it cannot confirm.ECAD · Agent failure · 1 report74
  5. 5easyeda_schematic_batch_write and place_component report applied, but the part reads back Mirror false, AddIntoBom true, AddIntoPcb true. The bridge only applied rotation after create. Workaround until the fix in PR #805: place the part, then call modify with addIntoBom false. A part meant to stay off the BOM would have gone to the order.ECAD · Agent failure · 1 report74
  6. 6easyeda-mcp-pro found that EasyEDA Pro returns an extensionless native filename, the project UUID, so the CPL and the Gerber zip landed on the same path in one artifact folder. The fix uses format-specific default names and tests both files exist with different contents.ECAD · Agent failure · 1 report74
  7. 7Moving a component with schematic_modify_primitive returns every attached wire in followedWireIds and an empty wireFollowFailures list. On a vertical move the wires stay put and every pin floats, and on a horizontal move the two stubs stretch and merge, one net disappears, and a pin ends up on the other pin's net. ERC was the only thing that caught it.ECAD · Agent failure · 1 report74
  8. 8The agent's PCB planner checked tracks against keepouts by sampling points, ignored track width, and used unrotated bounding boxes for placement, so narrow keepout crossings and rotated parts slipped through. The fix uses exact intersections, but native EasyEDA DRC remains the final check.ECAD · Agent failure · 1 report74
  9. 9The maintainer's answer to issue #931: docs requiring a manual 3D sign-off, because DRC and export were giving a misleading Final Package Ready signal and the 3D API is still beta.Verification · Agent failure · 1 report74
  10. 10The MCP server for EasyEDA read unavailable native board properties as negative findings and counted the full EasyEDA layer catalogue as copper layers. The fix scores all 23 PCB checks as passed, failed, not applicable or unknown, and an unverifiable rule can no longer produce a green "valid" or a passed production review. Blocking review mode now fails closed and names the missing fields. This is the agent-side version of the same problem. A check that can't see the data was reporting a result anyway.ECAD · Agent failure · 1 report74
  11. 11The move guard never ran when an agent changed only a part's rotation or mirror. A regression test showed both requests hit EasyEDA's native modify API without checking that pins and wires would stay connected. The fix fails closed with MOVE_CONNECTIVITY_UNVERIFIED.ECAD · Agent failure · 1 report74
  12. 12The netlist export used an obsolete fallback that the author calls incorrect, and it silently treated an unavailable export as a success. Pick-and-place came back from EasyEDA Pro as UTF-16LE, tab-separated, and the output filenames could overwrite another project's files.ECAD · Agent failure · 1 report74
  13. 13Through the EasyEDA MCP server, createProject in the desktop client returns ok: true with a null result and creates nothing, because EasyEDA local server wants a folder path the extension API cannot pass. A client that goes on writes into whichever project is open, and that happened to the reporter.ECAD · Agent failure · 1 report74
  14. 14Two fresh GND pours, one rebuild call: EasyEDA beta rebuild returns the first pour copper for the second, the bridge readback check fails, and from then on every copper rebuild throws until EasyEDA is restarted. A reconnect does not clear it, and the workaround is to rebuild one pour per call.ECAD · Agent failure · 1 report74
  15. 15A blank or badly typed expected net used to show up as a sourced WARNING that the actual net differed. Now it fails closed as UNKNOWN and only well-formed mismatches warn.ECAD · Agent failure · 1 report66
  16. 16A board missing fab notes raised a warning and the production review still answered passed: true, with summary.failed=1 in the same response. Second false pass from this server in two hours.ECAD · Agent failure · 1 report66
  17. 17A contributor reproduced EasyEDA Pro 3.2.209 returning an undefined result from rebuildCopperRegion() after pad, via, track or pour edits in the same session. The bridge already refused ambiguous results; now it also tells the caller about the editor's save, close and reopen workaround, and says the project, the pour IDs and the copper geometry have to be checked independently before trying again.ECAD · Agent failure · 1 report66
  18. 18A DRC or copper pour rebuild on a large board takes 30 to 120 seconds, but the bridge's stale sweep kills it after 15 to 45, ignoring the request's own timeout. The workaround is raising the global timeout on every server.ECAD · Agent failure · 1 report66
  19. 19A real EasyEDA Pro GND pour came back as 74 records, but the existing pad-contact screen only looked at the one filled polygon, so 73 stroke records went unchecked and the report never said so. The PR adds a not screened warning. It still does not evaluate them.ECAD · Agent failure · 1 report66
  20. 20An agent cannot place the second half of a dual op-amp in EasyEDA Pro. Placing OPA2333AIDR unit 2 fails with subPartName is not supported by this EasyEDA runtime. The user asks for a way to place both sections under one reference with one package in the BOM, or at least to be told before placement.ECAD · Agent failure · 1 report66
  21. 21An EasyEDA MCP server needs a write-ahead record so an agent does not pour copper twice after a disconnect. The earlier guard stored only the PCB id, so after a reconnect it could not tell whether the pour had already happened. The fix snapshots existing pours, filled copper geometry and requested properties before any native write.ECAD · Agent failure · 1 report66
  22. 22An EasyEDA MCP server reported a copper pour rebuild as verified when EasyEDA returned only unfilled outlines. The native API handed back a 'Poured' object with only fill=false outline records and no copper, and the tool counted it as done.ECAD · Agent failure · 1 report66
  23. 23An EasyEDA Pro extension could report the same version as the MCP server while missing the current copper-zone safety data, and bridge status said nothing about it. The generic API tool also missed some write methods (setState_*, saveProject, deletePrimitive) when deciding what needs explicit confirmation. The fix flags the mismatch and requires confirmation for those writes, while copper-zone writes by the agent are still off pending more testing. Agent write access into ECAD is being fenced in one method name at a time.ECAD · Agent failure · 1 report66
  24. 24api.call results were normalized only to depth 5, so DRC details and nested rule configs came back truncated to the agent.ECAD · Agent failure · 1 report66
  25. 25Asking the EasyEDA MCP server for a schematic PDF calls the PCB export, because the code tries the PCB function first and it always exists, so the schematic never comes back. The reporter says they use the right call daily through a locally patched extension.ECAD · Agent failure · 1 report66
  26. 26EasyEDA native wire edits can silently corrupt or disconnect nets, so the MCP server now refuses component moves that touch a wire.ECAD · Agent failure · 1 report66
  27. 27EasyEDA Pro 3.2.209 rejects subPartName for multi-unit parts, but the dry run still said success, so place_block or connector_breakout could preview an impossible plan and then begin writing. It now fails closed before any read or write, and tells users to place the units in the EasyEDA editor and check designator, package and a single BOM entry by hand.ECAD · Agent failure · 1 report66
  28. 28EasyEDA Pro Web extension sandbox does not reliably expose the editor page location, so the tool guessed it was on desktop, picked desktop defaults, and pcb.addText failed on a Web board. The fix hard-codes the native defaults.ECAD · Agent failure · 1 report66
  29. 29easyeda_pcb_modify_component always failed with 'PCB project identity unavailable' because its readback needed a project ID the tool's schema had no field for.ECAD · Agent failure · 1 report66
  30. 30easyeda_pcb_modify_component read back components without the focused project's UUID, so preview and verification failed before any transform.ECAD · Agent failure · 1 report66
  31. 31Empty create results were counted as success, so createProject and other create calls now need confirmWrite=true, following an earlier case where createProject returned ok with a null result.ECAD · Agent failure · 1 report66
  32. 32Follow-up to last hour's #834. Unavailable PCB measurement counts now stay UNKNOWN instead of becoming zero, missing constraint evidence is UNKNOWN, and missing-rule warnings fire only when a condition is explicitly false. Mounting holes stay UNKNOWN unless the editor exposes them.ECAD · Agent failure · 1 report66
  33. 33MCP writes could hit the wrong PCB because copied boards share primitive IDs, so guarded writes must now name the document UUID and native name.ECAD · Agent failure · 1 report66
  34. 34New easyeda_pcb_place_batch moves up to 64 parts, each with readback and per-item rollback, but stops at the first failure and reports partial application rather than rolling back the whole batch, so batch PCB placement for agents is explicitly not atomic.ECAD · Agent failure · 1 report66
  35. 35Real EasyEDA Pro schematics can hold two records for one designator, like two U3 entries. That unrelated duplicate made the datasheet ERC review report UNKNOWN for every other component, hiding real net contradictions.ECAD · Agent failure · 1 report66
  36. 36Schematic PDF export was routed to the PCB PDF API, and the author says there is no proof of a live export while the right API is beta and deprecated from EDA v4.1.ECAD · Agent failure · 1 report66
  37. 37The 73 unscreened GND strokes are now flagged as boundary candidates, but the result stays overall_status UNKNOWN with engineering_sign_off false because EasyEDA Pro's beta PCB_PrimitivePoured.getAll does not expose whether current can actually reach. The author says plainly this is source geometry, not proof of copper contact.ECAD · Agent failure · 1 report66
  38. 38The agent's DFM review reported width 0 and height 0 as within the declared max board size, so a zero-size board passed the factory limits check.ECAD · Agent failure · 1 report66
  39. 39The datasheet-aware ERC needed users to hand-assemble every net and pin entry even with the schematic open, and the auto version silently ignored datasheet constraints passed to it. The fix reads live nets, refuses incomplete inventories instead of returning a false electrical result, and keeps engineering_sign_off=false on the output.ECAD · Agent failure · 1 report66
  40. 40The EasyEDA MCP pick-and-place export with format csv writes an xlsx file named .csv, because the project id gets passed as the file name and no file type is sent. The workaround is to read the xlsx.ECAD · Agent failure · 1 report66
  41. 41The EasyEDA MCP placement bridge silently dropped requested mirror, addIntoBom and addIntoPcb, so a part could be placed and still be missing from the BOM. The fix applies them in a second documented modify call and will not report flags it could not read back.ECAD · Agent failure · 1 report66
  42. 42The EasyEDA Pro beta API can return stale objects or fail for copper pours an agent creates, so a pour outline can look like finished copper when no fill exists. The fix cross-checks every pour against a separate listing of filled regions by ID and geometry signature and fails closed on any mismatch. Agent writes into an EDA tool need a second, independent read to confirm they happened.ECAD · Agent failure · 1 report66
  43. 43The easyeda_bom_validate tool could return validated=true and count LCSC parts as valid with no supplier connected, and could also call a part invalid when a keyless lookup simply missed or the network failed. The fix adds an unverified_lcsc list and returns validated=false whenever the evidence is incomplete.ECAD · Agent failure · 1 report66
  44. 44The easyeda_schematic_verify_write tool accepts expectedComponentCountDelta and silently ignores it, because it expects an array from the same bridge call another tool in the same file reads as an object. An agent that writes to the schematic and calls this tool to confirm the write gets no confirmation and no failure, only a warning phrased as a transport problem. Filed in August, still active today.ECAD · Agent failure · 1 report66
  45. 45The generic API tool cuts every result at depth 5, so an agent asking for DRC results sees the category names and none of the findings, positions or distances. Issue #798 adds that there is also no ODB++ export tool, which matters because some assemblers like AISLER will not take Gerbers for assembly. Both are worked around today by patching the built extension after every release.ECAD · Agent failure · 1 report66
  46. 46The MCP sync tool reported success even when EasyEDA returned false or never acknowledged opening its import dialog. It now fails closed. The author is blunt that the real forward transfer and exactly one physical footprint on a disposable PCB remain not verified, and the 1.14.0 release stays on hold.ECAD · Agent failure · 1 report66
  47. 47The pad-to-net review accepted two different component pads reporting the same native padPrimitiveId, and marked both native_reported. Fix withholds both as unavailable, and electrical continuity stays UNKNOWN.ECAD · Agent failure · 1 report66
  48. 48The same EasyEDA MCP server's zone check trusted a PCB handle even when the focused tab was a different document or project. It also could not tell a rectangle touching the board edge from one fully inside it.ECAD · Agent failure · 1 report66
  49. 49Two fallback paths could report success without checking schematic connectivity, so the server now refuses those writes.ECAD · Agent failure · 1 report66
  50. 50Two native pad getter results could share a primitive ID, so the placement-pad reviewer attributed measurements to conflicting sources. It now returns UNKNOWN with no distance or net attribution.ECAD · Agent failure · 1 report66
  51. 51With a projectId given, the tool could open the PCB import for whatever schematic tab had focus, even one from a different project. It now fails closed. Companion fix to #1002, same sync tool.ECAD · Agent failure · 1 report66

How it fails

  1. Agent failure51

Engineering tools that address these errors

Sourced from public community reports. Costs and delays on individual entries are modeled estimates, not measurements.