autocad-mcp-pro

226 Commits 2 Branches 5 Tags

3 Commits

Include renames

Author	SHA1	Message	Date
Umut CANandClaude Opus 5	e57ae8b3ac	fix(task-12): restore the caller's tab from the layout listing; source the projection symbol enter_layout read the caller's space from `backend._current_space`, which only EzdxfBackend has. On ComBackend the getattr fell through to "Model", so `sheet_frame` / `titleblock_apply` with a `layout=` dropped a live caller onto the Model tab instead of putting them back -- the opposite of what the docstring and both tool docs promise, and a dual-engine parity break in new shared infrastructure. Measured on the live seat (AutoCAD 2026, scratch document from drawing_new(), closed with discard): scratch document: Drawing24.dwg layouts: ['Layout1', 'Layout2', 'Model'] B) current BEFORE: Layout2 B) the value the OLD code read (backend._current_space): <<absent -> getattr default 'Model'>> `layout_list()` already reported the right value two lines earlier on both engines (ezdxf from $TILEMODE / the active layout, COM from `doc.ActiveLayout.Name` -- read live as "Layout2" above), so `previous` now comes from `listing["current"]`. The projection-angle symbol was justified by an in-code derivation that cannot decide the question, and the test restated the implementation. The derivation was also applied inconsistently: the third-angle symbol mirrored the cone as well as the end view, leaving the trapezoid's short side pointing AWAY from the circles -- which is the first-angle arrangement. Transcribed instead from two independent reproductions of the ISO 5456-2 figure, both downloaded and measured (ISO's own figure is paywalled and was NOT read; the module says so): * FreeCAD's ISO 5457 sheet template A3_Landscape_ISO5457_advanced.svg, `first_angle_trapezoid` = "m 389,222 -10,2.5 v 5 l 10,2.5 z" -> 5-unit side at x=379, 10-unit side at x=389; circles r2.5/r5 at cx=396. * Wikimedia Commons "Convention placement vues dessin technique.svg", both symbols side by side, labelled FR and US: FR 35.856 @ x=13.343 / 66.135 @ x=81.869, circles x=148.005; US 35.856 @ x=380.275 / 66.135 @ x=448.801, circles x=314.140. Both draw the trapezoid the same way -- short side left -- and move only the end view: first angle puts it right of the cone, third angle left. Equivalently the short side points away from the circles in first angle and towards them in third, which is how every source states it in words. `third` now follows. Tests: the reference figures are transcribed into the test as data, and the symbol is asserted to land its three landmarks in the order each figure does, so the test cannot pass by agreeing with the code. The layout restore is pinned through a proxy that hides `_current_space` from the real backend rather than a fake that owns members ComBackend does not have. Both new tests fail on the previous code (measured: "assert 16.0 < 8.0" for the mirrored cone; "- SheetB + Model" for the tab). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd	2026-09-22 19:49:28 +03:00
Umut CANandClaude Opus 5	b03a7615b1	release: v1.5.0 — discovery layer, honesty debt closed, Wave A Tool count 131 -> 154; 474 -> 1242 collected tests. The release exists for two measured reasons. A 149-tool catalog costs a client ~40,300 tokens before it has asked for anything, and a drafter searching the way drafters think -- FILLET, BPOLY, QSELECT, WBLOCK -- scored df=0 against the stock index, because those words were simply not in the data. Discovery and batching - DISCOVERY_MODE=search replaces the catalog with search_tools + call_tool, ranked over an AutoCAD command and synonym corpus (discovery/aliases.py) on top of fastmcp's BM25 index. Measured: 40,305 -> 356 tokens advertised. - cad_batch runs a step list in one round trip; fields= projects result-heavy responses. The `core` profile is gone -- discovery replaced it. Honesty debt (Tier 0) - OCS/ECS: `extrusion` appeared nowhere in this codebase, so mirrored and rotated geometry reported coordinates that were quietly wrong. backends/ocs.py translates at the backend boundary; every tool coordinate is WCS on both engines, and a tilted plane refuses rather than answering with a number that looks fine. - drawing_save_as(fmt="dwg") wrote DXF bytes to a .dwg path. drawing_audit never repaired. analysis_measure_area shoelaced points the caller typed in. - ezdxf calls gained a per-call timeout, so one hung call can no longer wedge a server whose document lock is a single asyncio.Lock. Architecture - The 1854-line backend ABC is 19 contract modules; @capability(key, reason=) makes optional methods declarative and a gate fails on undeclared holes. - 8 layout/viewport tools, CHSPACE, CAD_PROGID escape hatch. - Wave A: selection filters, boundary tracing, hatch depth, annotation objects, measurement -- 13 tools of a planned 39. The REGION and 2D-boolean family was cut on evidence: add_region() yields zero ACIS bytes and the greiner_hormann substitute loses 28.2% of the area on a square with one semicircular edge. Verified against a live AutoCAD 2026, which found four COM defects no amount of reading would have caught -- AcadPViewport has no ViewCenter, and the line that set it had been swallowed by a bare except since v1.4. Benchmark v3 (15 tasks) and what it caught - v2 scored this server 10/10, which carried no information: every task in it exercised something the server was built around. Five tasks were added because they can fail, and three did while being written -- a hatch could not be measured though the refusal listed HATCH as measurable, the boundary tools and the measure tool disagreed about the same shape (139.2177 vs 139.2699), and a circle stored as two bulged vertices measured 0.0. - Competitor reports are NOT re-scored. They are pinned v1.4 runs against v2 and were never asked these five questions; the chart shows those rows `not run`. - Correctness A/B: 21/24 -> 24/24, zero regressions. Performance, stated plainly The published v1.4 perf report was recorded on CPython 3.14 and the new one on 3.11.15, so the apparent 2-6x speedup is the interpreter, not this release. Re-measured on one interpreter, v1.5.0 is 28% slower creating entities and 22% slower on the 10k roundtrip. Fully attributed: EZDXF_CALL_TIMEOUT=0 restores v1.4.0's exact numbers. That is the deadlock fix being paid for, the knob is documented, and winning it back is on the 1.6 roadmap. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>	2026-08-06 09:45:20 +03:00
U-C4N	7b121ab416	Initial commit	2026-04-27 12:20:57 +03:00

Author

SHA1

Message

Date

Umut CANandClaude Opus 5

e57ae8b3ac

fix(task-12): restore the caller's tab from the layout listing; source the projection symbol

enter_layout read the caller's space from `backend._current_space`, which only
EzdxfBackend has. On ComBackend the getattr fell through to "Model", so
`sheet_frame` / `titleblock_apply` with a `layout=` dropped a live caller onto
the Model tab instead of putting them back -- the opposite of what the
docstring and both tool docs promise, and a dual-engine parity break in new
shared infrastructure. Measured on the live seat (AutoCAD 2026, scratch
document from drawing_new(), closed with discard):

    scratch document: Drawing24.dwg
    layouts: ['Layout1', 'Layout2', 'Model']
    B) current BEFORE: Layout2
    B) the value the OLD code read (backend._current_space):
       <<absent -> getattr default 'Model'>>

`layout_list()` already reported the right value two lines earlier on both
engines (ezdxf from $TILEMODE / the active layout, COM from
`doc.ActiveLayout.Name` -- read live as "Layout2" above), so `previous` now
comes from `listing["current"]`.

The projection-angle symbol was justified by an in-code derivation that cannot
decide the question, and the test restated the implementation. The derivation
was also applied inconsistently: the third-angle symbol mirrored the cone as
well as the end view, leaving the trapezoid's short side pointing AWAY from the
circles -- which is the first-angle arrangement. Transcribed instead from two
independent reproductions of the ISO 5456-2 figure, both downloaded and
measured (ISO's own figure is paywalled and was NOT read; the module says so):

  * FreeCAD's ISO 5457 sheet template A3_Landscape_ISO5457_advanced.svg,
    `first_angle_trapezoid` = "m 389,222 -10,2.5 v 5 l 10,2.5 z" -> 5-unit side
    at x=379, 10-unit side at x=389; circles r2.5/r5 at cx=396.
  * Wikimedia Commons "Convention placement vues dessin technique.svg", both
    symbols side by side, labelled FR and US: FR 35.856 @ x=13.343 / 66.135 @
    x=81.869, circles x=148.005; US 35.856 @ x=380.275 / 66.135 @ x=448.801,
    circles x=314.140.

Both draw the trapezoid the same way -- short side left -- and move only the
end view: first angle puts it right of the cone, third angle left. Equivalently
the short side points away from the circles in first angle and towards them in
third, which is how every source states it in words. `third` now follows.

Tests: the reference figures are transcribed into the test as data, and the
symbol is asserted to land its three landmarks in the order each figure does,
so the test cannot pass by agreeing with the code. The layout restore is pinned
through a proxy that hides `_current_space` from the real backend rather than a
fake that owns members ComBackend does not have. Both new tests fail on the
previous code (measured: "assert 16.0 < 8.0" for the mirrored cone; "- SheetB
+ Model" for the tab).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd

2026-09-22 19:49:28 +03:00

Umut CANandClaude Opus 5

b03a7615b1

release: v1.5.0 — discovery layer, honesty debt closed, Wave A

Tool count 131 -> 154; 474 -> 1242 collected tests.

The release exists for two measured reasons. A 149-tool catalog costs a client
~40,300 tokens before it has asked for anything, and a drafter searching the way
drafters think -- FILLET, BPOLY, QSELECT, WBLOCK -- scored df=0 against the stock
index, because those words were simply not in the data.

Discovery and batching
- DISCOVERY_MODE=search replaces the catalog with search_tools + call_tool,
  ranked over an AutoCAD command and synonym corpus (discovery/aliases.py) on
  top of fastmcp's BM25 index. Measured: 40,305 -> 356 tokens advertised.
- cad_batch runs a step list in one round trip; fields= projects result-heavy
  responses. The `core` profile is gone -- discovery replaced it.

Honesty debt (Tier 0)
- OCS/ECS: `extrusion` appeared nowhere in this codebase, so mirrored and
  rotated geometry reported coordinates that were quietly wrong. backends/ocs.py
  translates at the backend boundary; every tool coordinate is WCS on both
  engines, and a tilted plane refuses rather than answering with a number that
  looks fine.
- drawing_save_as(fmt="dwg") wrote DXF bytes to a .dwg path. drawing_audit never
  repaired. analysis_measure_area shoelaced points the caller typed in.
- ezdxf calls gained a per-call timeout, so one hung call can no longer wedge a
  server whose document lock is a single asyncio.Lock.

Architecture
- The 1854-line backend ABC is 19 contract modules; @capability(key, reason=)
  makes optional methods declarative and a gate fails on undeclared holes.
- 8 layout/viewport tools, CHSPACE, CAD_PROGID escape hatch.
- Wave A: selection filters, boundary tracing, hatch depth, annotation objects,
  measurement -- 13 tools of a planned 39. The REGION and 2D-boolean family was
  cut on evidence: add_region() yields zero ACIS bytes and the greiner_hormann
  substitute loses 28.2% of the area on a square with one semicircular edge.

Verified against a live AutoCAD 2026, which found four COM defects no amount of
reading would have caught -- AcadPViewport has no ViewCenter, and the line that
set it had been swallowed by a bare except since v1.4.

Benchmark v3 (15 tasks) and what it caught
- v2 scored this server 10/10, which carried no information: every task in it
  exercised something the server was built around. Five tasks were added because
  they can fail, and three did while being written -- a hatch could not be
  measured though the refusal listed HATCH as measurable, the boundary tools and
  the measure tool disagreed about the same shape (139.2177 vs 139.2699), and a
  circle stored as two bulged vertices measured 0.0.
- Competitor reports are NOT re-scored. They are pinned v1.4 runs against v2 and
  were never asked these five questions; the chart shows those rows `not run`.
- Correctness A/B: 21/24 -> 24/24, zero regressions.

Performance, stated plainly
The published v1.4 perf report was recorded on CPython 3.14 and the new one on
3.11.15, so the apparent 2-6x speedup is the interpreter, not this release.
Re-measured on one interpreter, v1.5.0 is 28% slower creating entities and 22%
slower on the 10k roundtrip. Fully attributed: EZDXF_CALL_TIMEOUT=0 restores
v1.4.0's exact numbers. That is the deadlock fix being paid for, the knob is
documented, and winning it back is on the 1.6 roadmap.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

2026-08-06 09:45:20 +03:00

U-C4N

7b121ab416

Initial commit

2026-04-27 12:20:57 +03:00