The v1.6 roadmap in five tracks: 154 -> 247 registered tools in 28 groups.
Evidence re-measured for the release:
- correctness A/B against v1.5.1: 26/48 -> 48/48, 22 fixed, 0 regressed
(ab-v1.5.1-vs-v1.6.0.json, replacing the -dev report);
- the reference adapter on matrix v7: 21/21 (autocad-mcp-pro.json, which
was the v3 run of 1.5); the published-report test follows it to v7;
- the 100-point rubric re-scored per category, every withheld point named:
93 (the competitor rows are their July snapshots, not re-reviewed);
- the perf lane as medians of five alternating runs, beside a v1.4.0
comparison on one interpreter (perf-v1.4.0-vs-v1.6.0.json).
Two roadmap criteria are not green, and the changelog, the README and the
roadmap say so: criterion 5 is withdrawn (track D, Plant 3D, dropped) and
criterion 7 is not met (creation 1.25x / 1.38x v1.4.0's time, still
1.17x / 1.20x with the timeout off). Carried to 1.7.
README rewritten for 1.6: what is new, the mechanical and architectural
sheets, the rubric and task-matrix charts, the measured limits (including
the four feature tables that ship empty), and the 1.7 roadmap. server.json
describes the 1.6 surface and declares TOOL_PACKS. CLAUDE.md corrects the
ISO 286 letter set and the empty feature tables.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd
The triage over the 43 unaddressed verdicts returned 4 code and 35 doc items.
The code ones shipped in e24b46d; these are the documents.
Twenty-one applied mechanically after a validation pass confirmed each locator
still matched exactly once and each replacement was supported by the tree. Ten
were held back because a re-verify pass found them mutually exclusive -- three
patches rewriting the same benchmarks/README block, three rewriting the same
OCS paragraph -- and applying them in sequence would have corrupted the files.
Those were resolved by hand into one coherent edit each. One was rejected.
The corrections worth naming:
- **The A/B table contradicted itself in three files.** benchmarks/README.md
and CLAUDE.md said 24 checks / 21-24 / 87.5% / "fixed 3" while the suite has
26 and the published JSON says 21/26, 80.8%, five fixed. All three now agree,
and the five are split into the three new-capability `miss -> pass` and the
two repaired-defect `fail -> pass`, which is the distinction the lane exists
to make.
- **"Every number below is produced by a script"** was false twice over:
section 5's nine scores are literals a human assigned, and section 4 now
quotes hand-collected timings. The claim is narrowed to the charts, which
really are all rendered from committed data, and both exceptions are named.
- **The OCS paragraph over-quantified.** TEXT rotation was given as the single
exception; z components and HATCH boundaries are also entity-frame, and the
live backend records `2d_polyline_coordinates_not_verified_against_live_autocad`
in the same reason string. It now points at `ocs_normalized` as the live list.
- **Transactions were shown as bare parity.** The server itself grades them
differently -- `FeatureCapability(True, "undo_mark")` on COM against
`(True, "snapshot")` headless -- and COM rollback sends `_UNDO B` without
confirming it landed. The row says so.
- **"Raw commands: Opt-in"** named a switch that does not exist. Both tools are
registered unconditionally; `DANGEROUS_COMMANDS_ENABLED` only disables the
sanitizer.
- **"Wave A shipped 13 of a planned 39"** hardened an approximation: the only
source writes `~39`.
- **CAD_PROGID** said nothing beyond the connection was tested against
BricsCAD/ZWCAD/GstarCAD, implying the connection was. It was not.
Perf lane reworked against a fresh paired measurement. The old text claimed
"28% slower than v1.4.0, median of three runs" and that
`EZDXF_CALL_TIMEOUT=0` restored v1.4.0's numbers "exactly". Three problems:
`perf_suite` runs each workload once per invocation and cannot produce a
median, "exactly" was refuted by the figures in its own sentence, and the
number moved when the timeout wrapper was rebuilt. It now reports the guard's
cost -- +20.8% and +19.9%, three paired runs, median taken by hand -- says the
figures have no published artifact, and says the wrapper rework recovered about
a third of the v1.5.0 overhead rather than closing it.
perf-ezdxf.json and the perf chart regenerated against the shipped code, and
the README table rebuilt from that artifact so the two cannot drift.
1267 passed, 1 xfailed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`add_diameter_dim` / `add_radius_dim` measure to the point they are handed;
`leader_length` is a text placement. The ezdxf backend passed
`centre + (radius + leader)` as that point, so the leader became part of the
measurement: a true 40 mm bore was dimensioned 60 at the default
`leader_length=10`, and 108 at 34. Radius callouts were `leader_length` too
large the same way.
This is the worst class of defect for this project -- a wrong number, at
default settings, on the one part of a drawing that exists to carry numbers --
and it shipped in v1.4.0 and earlier. The size now comes from `radius` and
`location` only places the text.
The live COM backend was never affected: ActiveX's AddDimDiametric takes the
two chord points and the leader length as separate arguments. The two engines
had been disagreeing by 2 x leader_length on the same call.
Found while building the drawing for the new README, whose own bore callout was
wrong. Two checks added to correctness_suite.py; both report fail -> pass
against v1.4.0, not miss -> pass, which is what separates a repaired defect
from new capability in that lane. A/B is now 21/26 -> 26/26, five fixed, zero
regressed.
README rewritten from scratch
- Hero is a real ISO A3 sheet the server drew: m6 z24 involute spur gear, DIN
6885 keyed bore, section A-A, ISO 129 dimensions, ISO 286 H7 fit, ISO 7200
title block, 56 entities, drawing_critique clean. New
scripts/render_readme_showcase.py rebuilds it, and the tool-call sequence
that produces it is printed in the page.
- Star history chart, badge row, capability matrix, five evidence lanes, and a
"what this release is bad at" section carrying the 28% throughput regression.
- Every image and repo link is now an absolute URL: README.md is the PyPI long
description, where relative paths 404.
Corrections this work forced, all measured
- The df=0 example was wrong. FILLET is already in entity_fillet's own text;
the commands that genuinely scored zero are WBLOCK, PEDIT, OVERKILL,
LAYTRANS and MATCHPROP.
- 40,305 tokens is the tools/list wire payload; ~25,143 is the part forwarded
to the model. The old wording said "of schema", which is neither.
- `--tokenizer anthropic` needs a package and credentials this repo does not
lock, and refuses rather than falling back. Said so.
- Discovery ranks over the 149-tool advertised catalog, not the 154 registered.
- Five middlewares, not four; the fifth is what makes refusals machine-readable.
- Five of the 27 capability keys are environment-dependent, so the matrix
carries footnotes instead of bare ticks.
- "Every coordinate is WCS" has one deliberate exception: TEXT rotation.
- render_live_chart's axis still said "fixed 10-task matrix" while our bar came
from v3's fifteen. Each bar now carries its own matrix.
- .env.example said MAX_UNDO_STACK=50 (code: 5) and omitted four real
variables, two of them the remote-HTTP controls.
- server.json described an empty ALLOWED_PATHS as blocking access. It disables
the allowlist.
- server.py called the lean profile ~46 tools; it is 47.
- backends/capability.py cited a test file that does not exist.
1249 passed, 1 xfailed.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
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>
Leaderboard restyled to the LLM-benchmark look: dark background, rank-numbered labels, per-rank color palette, 'NN / 100 - grade' annotations. The live-run, task-matrix and performance charts share the same dark theme for a consistent README identity. Charts regenerated deterministically from their data files.
Co-authored-by: Cursor <cursoragent@cursor.com>
source_review.json extended with three projects reviewed 2026-07-23 (LokmenoWer/best-cad-mcp B/74, daobataotie/CAD-MCP B/35, zh19980811/Easy-MCP-AutoCad C/30) and our entry updated to the released v1.4.0 (95: CI matrix, PyPI packaging, A/B gate, paper space, ISO 286, opt-in solids). Chart regenerated via render_chart; README benchmark section now opens with the leaderboard + 9-row table, source-review lane merged into it.
Co-authored-by: Cursor <cursoragent@cursor.com>
