feat(server): settings pack — 22 environment tools behind TOOL_PACKS, lean grows to 55, three groups, golden set +10

Task 24 (merge sanity) on main after the five track E groups merged.

- TOOL_PACK_NAMES = ("core", "pid", "settings"); PACK_TOOL_NAMES["settings"]
  is exactly the 22 SECTION 20 tools. Styles and page setup stay in core.
- LEAN_TOOL_NAMES += page_setup_apply, batch_plot (53 -> 55).
- _GROUP_TAG_PRIORITY / _GROUP_TAG_LABELS: "style" moved up next to "pid",
  "pagesetup" -> page_setup and "environment" -> environment added; the 28
  SECTION 19/20 decorators gain their section tag, so _tool_groups() reports
  styles 10 / page_setup 6 / environment 22 (23 groups, 204 tools) and
  pack == group == section for the settings pack.
- SECTION headers, backends/base.py composition, contracts/__init__.py
  exports and the layer_states feature entry were already correct after the
  merge (read and confirmed, nothing edited).
- Pack text: config.py, .env.example, README config row; README <sub> line
  204 tools in 23 groups; release snapshot 3259 collected tests (measured).
- docs/tool-inventory.json regenerated; registry snapshot re-derived.
- tests: test_tool_packs.py replaced (11 tests), test_capability_contract.py
  +3 (composition, the agreed capability flags, layer_states honesty),
  golden set +10 tuning cases (81 = 66 tuning + 15 holdout); all ten route
  in the top 3 without alias edits.

Deviation from the plan text: TRACK_E_KEYS carries five keys, not six. The
plan's "mleaderstyle" (ezdxf True / com False) contradicts spec §8.1, which
retracted that key because the ActiveX route exists (§4.2); measured on the
tree: capability_keys() == {boundary_trace, chspace, interactive_prompt,
live_application, preferences, revcloud, wipeout}, and
ComBackend.mleaderstyle_create is a native implementation
(backends/com_backend.py:6752, is_capability_default -> False). The test now
also asserts the key is absent.

No new ActiveX call was written in this task, so no live probe was needed.

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

tests/data/discovery_golden.py

@@ -1,6 +1,6 @@
 """Golden query set for tool discovery: English in, one right tool out.
 
-Seventy-one cases (56 tuning + 15 holdout) split into a **tuning** set and a
+Eighty-one cases (66 tuning + 15 holdout) split into a **tuning** set and a
 **holdout** set. The split is
 the point of the file. Ranking work is measured against
 :data:`TUNING_CASES` only; :data:`HOLDOUT_CASES` were written at the same time,
@@ -155,6 +155,25 @@ TUNING_CASES: tuple[GoldenCase, ...] = (
     ),
     GoldenCase("create a text style that uses the isocp font", "textstyle_create", "paraphrase"),
     GoldenCase("which dimension styles are defined", "dimstyle_list", "counting", risk="read"),
+    # v1.6 track E — styles, page setup, templates, environment. Written
+    # before the alias records were measured against them; a miss here is
+    # fixed by adding vocabulary to the tool's record in discovery/aliases.py,
+    # never by moving the expectation.
+    GoldenCase("PAGESETUP", "page_setup_apply", "command"),
+    GoldenCase("PUBLISH", "batch_plot", "command"),
+    GoldenCase("MLEADERSTYLE", "mleaderstyle_create", "command"),
+    GoldenCase("STYLE", "textstyle_create", "command"),
+    GoldenCase("iso 25 dimension style", "dimstyle_create", "synonym"),
+    GoldenCase("layer snapshot", "layer_state_save", "synonym"),
+    GoldenCase("save this drawing as a template", "drawing_template_save", "paraphrase"),
+    GoldenCase("switch to the other open drawing", "document_activate", "paraphrase"),
+    GoldenCase("ask the operator to click a point on screen", "user_pick_point", "paraphrase"),
+    GoldenCase(
+        "what does the LTSCALE variable mean",
+        "system_variable_describe",
+        "counting",
+        risk="read",
+    ),
 )
 
 

tests/test_capability_contract.py

@@ -32,6 +32,9 @@ import pytest
 from backends.base import AutoCADBackend, UnsupportedCapabilityError
 from backends.capability import capability_keys, is_capability_default
 from backends.com_backend import ComBackend
+from backends.contracts.environment import EnvironmentContract
+from backends.contracts.pagesetup import PageSetupContract
+from backends.contracts.styles import StylesContract
 from backends.ezdxf_backend import EzdxfBackend
 
 # Marked per-test rather than module-wide: half of these are plain registry
@@ -116,3 +119,50 @@ async def test_an_implementing_backend_is_unaffected(backend):
     result = await backend.entity_change_space([], "nope")
     assert result["ok"] is False
     assert "error" in result, "the real implementation still does its own validation"
+
+
+# ── track E: composition and the five agreed keys ───────────────────────────
+
+#: spec §8.1 — the flag each engine must report. `mleaderstyle` is deliberately
+#: absent: the spec retracted it (the ActiveX route through the
+#: ``ACAD_MLEADERSTYLE`` dictionary exists, §4.2), and both engines implement
+#: ``mleaderstyle_create`` natively — `capability_keys()` never registers it.
+TRACK_E_KEYS = {
+    "dwgprops": {"ezdxf": False, "com": True},
+    "dwt_write": {"ezdxf": False, "com": True},
+    "live_application": {"ezdxf": False, "com": True},
+    "preferences": {"ezdxf": False, "com": True},
+    "interactive_prompt": {"ezdxf": False, "com": True},
+}
+
+
+def _feature_keys_full(backend) -> dict:
+    return backend.capabilities().to_dict()["features"]
+
+
+def test_track_e_contracts_are_composed_into_the_backend():
+    for contract in (StylesContract, PageSetupContract, EnvironmentContract):
+        assert issubclass(AutoCADBackend, contract), f"{contract.__name__} not in AutoCADBackend"
+        assert issubclass(EzdxfBackend, contract) and issubclass(ComBackend, contract)
+
+
+def test_track_e_capability_keys_carry_the_agreed_flags():
+    assert "mleaderstyle" not in capability_keys(), "spec §8.1: no such engine boundary"
+    ezdxf_features = _feature_keys_full(EzdxfBackend())
+    com_features = _feature_keys_full(ComBackend())
+    for key, flags in TRACK_E_KEYS.items():
+        assert key in ezdxf_features, f"{key} undeclared on ezdxf"
+        assert key in com_features, f"{key} undeclared on com"
+        assert ezdxf_features[key]["supported"] is flags["ezdxf"], key
+        assert com_features[key]["supported"] is flags["com"], key
+
+
+def test_layer_states_are_declared_as_ours_not_autocads():
+    """Spec §7.2 / §8.3: the states live in the file and travel with it, and do
+    not appear in AutoCAD's Layer States Manager. Said in the map, not only in
+    a docstring."""
+    for backend in (EzdxfBackend(), ComBackend()):
+        feature = _feature_keys_full(backend)["layer_states"]
+        assert feature["supported"] is True
+        assert feature["mode"] == "xrecord"
+        assert "layer_states_manager" in (feature["reason"] or ""), backend.name

tests/test_tool_packs.py

@@ -1,4 +1,11 @@
-"""TOOL_PACKS: advertise only the vertical packs a client uses; core is always on."""
+# tests/test_tool_packs.py
+"""TOOL_PACKS: advertise only the vertical packs a client uses; core is always on.
+
+v1.6 track E adds the `settings` pack: the 22 SECTION 20 environment tools.
+Styles (SECTION 18) and page setup / templates (SECTION 19) are drafting
+essentials and stay in `core`, so `TOOL_PACKS=core` hides the environment
+surface and nothing else.
+"""
 
 from __future__ import annotations
 
@@ -24,6 +31,63 @@ PID_TOOLS = {
     "pid_from_spec",
 }
 
+STYLE_TOOLS = {
+    "dimstyle_list",
+    "dimstyle_create",
+    "dimstyle_modify",
+    "dimstyle_set_current",
+    "textstyle_list",
+    "textstyle_create",
+    "textstyle_set_current",
+    "mleaderstyle_list",
+    "mleaderstyle_create",
+    "drawing_apply_standard",
+}
+
+PAGE_SETUP_TOOLS = {
+    "page_setup_list",
+    "page_setup_apply",
+    "plot_style_list",
+    "batch_plot",
+    "drawing_template_list",
+    "drawing_template_save",
+}
+
+SETTINGS_TOOLS = {
+    "document_list",
+    "document_activate",
+    "document_close",
+    "layer_state_save",
+    "layer_state_restore",
+    "layer_state_list",
+    "layer_state_delete",
+    "view_named_save",
+    "view_named_restore",
+    "view_named_list",
+    "ucs_list",
+    "ucs_set",
+    "ucs_restore",
+    "system_launch",
+    "system_preferences_get",
+    "system_preferences_set",
+    "drawing_properties_get",
+    "drawing_properties_set",
+    "system_variable_describe",
+    "user_pick_point",
+    "user_select",
+    "system_prompt_message",
+}
+
+LEAN_SETTINGS_ESSENTIALS = {
+    "drawing_apply_standard",
+    "page_setup_apply",
+    "batch_plot",
+    "dimstyle_set_current",
+    "textstyle_set_current",
+}
+
+ALL_PACKS = ["core", "pid", "settings"]
+
 
 @pytest_asyncio.fixture(autouse=True)
 async def _restore(monkeypatch):
@@ -32,29 +96,48 @@ async def _restore(monkeypatch):
     await server._apply_tool_profile("full")
 
 
-def test_pack_registry_names_real_tools_and_only_pid_ones():
-    assert server.TOOL_PACK_NAMES == ("core", "pid")
+def test_pack_registry_names_real_tools_and_only_them():
+    assert server.TOOL_PACK_NAMES == ("core", "pid", "settings")
     assert server.PACK_TOOL_NAMES["pid"] == frozenset(PID_TOOLS)
+    assert server.PACK_TOOL_NAMES["settings"] == frozenset(SETTINGS_TOOLS)
+    assert len(SETTINGS_TOOLS) == 22
+    assert not (server.PACK_TOOL_NAMES["settings"] & server.PACK_TOOL_NAMES["pid"])
 
 
-async def test_core_only_hides_every_pid_tool_but_keeps_block_define(monkeypatch):
+async def test_every_pack_member_is_a_registered_tool():
+    registered = {tool.name for tool in await server._registered_tools()}
+    for pack, names in server.PACK_TOOL_NAMES.items():
+        missing = names - registered
+        assert not missing, f"pack {pack!r} names unregistered tools: {sorted(missing)}"
+
+
+async def test_core_only_hides_pid_and_settings_but_keeps_styles_and_page_setup(monkeypatch):
     monkeypatch.setattr(config.settings, "tool_packs", "core")
     info = await server._apply_tool_profile("full")
     disabled = set(info["disabled_tools"])
     assert PID_TOOLS <= disabled
+    assert SETTINGS_TOOLS <= disabled
+    assert not (STYLE_TOOLS & disabled), "styles are core drafting essentials"
+    assert not (PAGE_SETUP_TOOLS & disabled), "page setup and templates are core"
     assert "block_define" not in disabled and "entity_set_xdata" not in disabled
-    assert info["tool_packs"] == {
-        "enabled": ["core"],
-        "available": ["core", "pid"],
-        "ignored": [],
-    }
+    assert info["tool_packs"] == {"enabled": ["core"], "available": ALL_PACKS, "ignored": []}
 
 
-async def test_all_is_the_default_and_enables_pid(monkeypatch):
+async def test_settings_pack_alone_keeps_pid_hidden(monkeypatch):
+    monkeypatch.setattr(config.settings, "tool_packs", "settings")
+    info = await server._apply_tool_profile("full")
+    disabled = set(info["disabled_tools"])
+    assert PID_TOOLS <= disabled
+    assert not (SETTINGS_TOOLS & disabled)
+    assert info["tool_packs"]["enabled"] == ["core", "settings"]
+
+
+async def test_all_is_the_default_and_enables_every_pack(monkeypatch):
     monkeypatch.setattr(config.settings, "tool_packs", "all")
     info = await server._apply_tool_profile("full")
-    assert not (PID_TOOLS & set(info["disabled_tools"]))
-    assert info["tool_packs"]["enabled"] == ["core", "pid"]
+    disabled = set(info["disabled_tools"])
+    assert not ((PID_TOOLS | SETTINGS_TOOLS) & disabled)
+    assert info["tool_packs"]["enabled"] == ALL_PACKS
 
 
 async def test_unknown_pack_is_ignored_with_a_warning_and_core_cannot_be_dropped(
@@ -65,7 +148,7 @@ async def test_unknown_pack_is_ignored_with_a_warning_and_core_cannot_be_dropped
         info = await server._apply_tool_profile("full")
     assert info["tool_packs"] == {
         "enabled": ["core", "pid"],
-        "available": ["core", "pid"],
+        "available": ALL_PACKS,
         "ignored": ["mech3000"],
     }
     assert "mech3000" in caplog.text
@@ -79,9 +162,34 @@ async def test_lean_intersects_with_packs(monkeypatch):
     assert "entity_create_line" not in info["disabled_tools"]
 
 
+async def test_lean_carries_the_five_settings_essentials_and_nothing_from_the_settings_pack(
+    monkeypatch,
+):
+    """Spec §8.2: lean = 55. The five are core-pack tools, so `TOOL_PACKS=core`
+    leaves them on a lean surface; no environment tool is lean."""
+    assert LEAN_SETTINGS_ESSENTIALS <= server.LEAN_TOOL_NAMES
+    assert not (SETTINGS_TOOLS & server.LEAN_TOOL_NAMES)
+    assert len(server.LEAN_TOOL_NAMES) == 55
+    monkeypatch.setattr(config.settings, "tool_packs", "core")
+    info = await server._apply_tool_profile("lean")
+    disabled = set(info["disabled_tools"])
+    assert not (LEAN_SETTINGS_ESSENTIALS & disabled)
+    assert info["enabled_count"] == 55 - 3, "lean minus the three pid tools"
+
+
+async def test_the_three_track_e_sections_file_under_their_own_groups():
+    """Pack, group and SECTION are the same list, so `system_about` and
+    `docs/tool-inventory.json` describe the surface the packs gate."""
+    groups = await server._tool_groups()
+    assert set(groups["styles"]) == STYLE_TOOLS
+    assert set(groups["page_setup"]) == PAGE_SETUP_TOOLS
+    assert set(groups["environment"]) == SETTINGS_TOOLS
+    assert len(groups) == 23
+
+
 async def test_system_about_reports_packs():
     from fastmcp import Client
 
     async with Client(server.mcp) as client:
         about = (await client.call_tool("system_about", {})).structured_content
-    assert about["tool_packs"]["available"] == ["core", "pid"]
+    assert about["tool_packs"]["available"] == ALL_PACKS

tests/test_tool_registry.py

@@ -495,7 +495,7 @@ async def test_tool_groups_is_byte_identical_to_the_source_declared_grouping():
 
 
 async def test_tool_group_sizes_are_unchanged():
-    """Frozen snapshot of the surface (172 tools, 20 groups).
+    """Frozen snapshot of the surface (204 tools, 23 groups).
 
     Taken before @cad_tool landed at 131 tools; `batch` moved 2 -> 3 when
     v1.5.0's `cad_batch` joined `entity_batch_create`/`entity_batch_modify`, and
@@ -517,35 +517,18 @@ async def test_tool_group_sizes_are_unchanged():
     when `pid_tag_parse` (the ISA-5.1 tag grammar as a read-only tool)
     joined; the `pid`
     tag is ranked first in `_GROUP_TAG_PRIORITY` so those tools file under
-    `pid` rather than under their secondary `query` / `create` tags. `styles`
-    appeared (0 -> 9) when track E's SECTION 18 opened with the dimension,
-    text and multileader style tools, then 9 -> 10 when
-    `drawing_apply_standard` (ISO or ANSI styles, units and layers in one
-    call) joined; the `style` tag sits ahead of `layer` and `drawing` in
-    `_GROUP_TAG_PRIORITY` so they never file under their secondary
-    `query` / `create` / `modify` / `drawing` tags. `system`
-    moved 7 -> 8 when v1.6's `system_variable_describe` (the sysvar catalogue
-    as a read-only tool) opened SECTION 20. `drawing` moved 11 -> 13 when
-    v1.6's `drawing_properties_get` / `drawing_properties_set` (DWGPROPS on
-    both engines) joined SECTION 20 — tagged `drawing` and `settings`, and
-    `settings` is not a group tag, so they file under drawing.
-    `layouts` moved 12 -> 16 when v1.6's SECTION 19 opened with
-    `page_setup_list` / `page_setup_apply` / `plot_style_list` / `batch_plot`
-    — all tagged `layout`, which `_GROUP_TAG_PRIORITY` ranks above
-    `query`/`export`, so they file with the sheets they set up. `templates`
-    moved 2 -> 3 when v1.6's `drawing_template_list` joined SECTION 19
-    (tagged `template`, which ranks above `drawing`), then 3 -> 4 when
-    `drawing_template_save` joined. `drawing`
-    moved 13 -> 16 when v1.6's `document_list` / `document_activate` /
-    `document_close` (multi-document on both engines) opened SECTION 20;
-    `layers` moved 14 -> 18 with the four `layer_state_*` tools and `view`
-    4 -> 10 with `view_named_*` and `ucs_*` (UCS files under `view`, where
-    AutoCAD's own ribbon keeps it) when v1.6's SECTION 20 grew. `system`
-    moved 8 -> 14 with v1.6's live-only environment tools (`system_launch`,
-    `system_preferences_get/set`, `user_pick_point`, `user_select`,
-    `system_prompt_message`), which refuse headlessly with declared
-    capability keys. Every other number here has been unchanged since the
-    snapshot was taken.
+    `pid` rather than under their secondary `query` / `create` tags.
+    Three groups appeared with v1.6's track E — `styles` (10, SECTION 18),
+    `page_setup` (6, SECTION 19) and `environment` (22, SECTION 20) — each
+    tagged with its section name, ranked directly after `pid` in
+    `_GROUP_TAG_PRIORITY` so a style tool that also carries `dimension` or an
+    environment tool that also carries `layer` files under its section; the
+    `environment` group is exactly `PACK_TOOL_NAMES["settings"]`. (Until the
+    merge task tagged them, the four groups had filed those 28 tools under
+    `layouts` / `templates` / `drawing` / `layers` / `view` / `system` by
+    their secondary tags; those counts returned to their pre-track-E values
+    when the section tags landed.) Every other number here has been unchanged
+    since the snapshot was taken.
     """
     sizes = {label: len(names) for label, names in (await server._tool_groups()).items()}
     assert sizes == {
@@ -554,21 +537,23 @@ async def test_tool_group_sizes_are_unchanged():
         "blocks": 9,
         "corner_ops": 4,
         "dimensions": 5,
-        "drawing": 16,
+        "drawing": 11,
         "engineering": 10,
         "entity_creation": 18,
         "entity_modification": 16,
         "entity_query": 9,
-        "layers": 18,
-        "layouts": 16,
+        "environment": 22,
+        "layers": 14,
+        "layouts": 12,
+        "page_setup": 6,
         "pid": 9,
         "premium": 12,
         "solids": 5,
         "styles": 10,
-        "system": 14,
-        "templates": 4,
+        "system": 7,
+        "templates": 2,
         "transactions": 3,
         "validation": 1,
-        "view": 10,
+        "view": 4,
     }
     assert sum(sizes.values()) == 204