docs: drop module docstrings and trim class docstrings to one-liners (#30)

Internal cleanup on the typed-models package — file location and class
name carry the intent; the multi-paragraph rationales were rehashing the
PR description. Function bodies are unchanged.

  - models/__init__.py:        removed 8-line module docstring
  - models/errors.py:          dropped module docstring + SchemaError
                               class docstring trimmed to one line
  - models/conversation.py:    dropped module docstring; Composer,
                               WorkspaceLocalComposer, Bubble docstrings
                               trimmed to one line each
  - models/cli_session.py:     dropped module docstring; CliSessionMeta
                               docstring trimmed
  - models/workspace.py:       dropped module docstring; Workspace
                               docstring trimmed
  - models/export.py:          dropped module docstring; ExportEntry
                               docstring trimmed
  - tests/test_models.py:      dropped module docstring; removed
                               now-unused ``from pathlib import Path``

204 tests pass.

Author: timon0305

models/__init__.py

@@ -1,13 +1,3 @@
-"""Typed domain models for Cursor schema (closes #24).
-
-Cursor's on-disk JSON shapes are not versioned, so silent renames of fields
-like ``composerData`` or ``latestRootBlobId`` would otherwise pass through
-``dict.get(...)`` with a fallback default and produce empty conversations
-with no error raised. The models here add a schema-validation boundary at
-database read sites: ``from_dict`` classmethods raise ``SchemaError`` when
-critical fields are missing, so drift becomes loud instead of silent.
-"""
-
 from models.cli_session import CliSessionMeta
 from models.conversation import Bubble, Composer, WorkspaceLocalComposer
 from models.errors import SchemaError

models/cli_session.py

@@ -1,5 +1,3 @@
-"""CliSessionMeta — typed model for the Cursor CLI ``meta`` blob."""
-
 from __future__ import annotations
 
 from dataclasses import dataclass, field
@@ -10,16 +8,7 @@
 
 @dataclass(frozen=True)
 class CliSessionMeta:
-    """The ``meta`` blob at the head of a Cursor CLI ``store.db`` blob graph.
-
-    ``latestRootBlobId`` is the entry point for the conversation reconstruction
-    BFS in ``utils/cli_chat_reader.traverse_blobs``; without it, the entire
-    conversation is unreachable. ``createdAt`` is documented as part of the
-    meta-blob schema (see ``utils/cli_chat_reader`` module docstring) and is
-    captured here, but it is not gated on — only ``latestRootBlobId`` is the
-    hard requirement, since that is the only field whose absence prevents
-    conversation reconstruction.
-    """
+    """CLI session meta blob; latestRootBlobId is the conversation entry point and the only required field."""
 
     latest_root_blob_id: str
     created_at: Any = None

models/conversation.py

@@ -1,5 +1,3 @@
-"""Composer (conversation) and Bubble (message) typed models."""
-
 from __future__ import annotations
 
 from dataclasses import dataclass, field
@@ -10,19 +8,7 @@
 
 @dataclass(frozen=True)
 class Composer:
-    """A Cursor conversation (a.k.a. "composer") row.
-
-    Required fields per the schema-validation contract (issue #24):
-      - ``fullConversationHeadersOnly`` — without this, a composer cannot be
-        rendered (no message order is recoverable).
-      - ``createdAt`` — Cursor writes this on every composer (verified
-        17/17 against a live workspaceStorage). A missing value is the
-        kind of drift this layer exists to surface.
-
-    The composer ID is intentionally passed in as a constructor argument
-    rather than read from ``raw`` because Cursor stores it in the row key
-    (``composerData:<id>``) rather than in the JSON value.
-    """
+    """Cursor conversation row from globalStorage cursorDiskKV; requires fullConversationHeadersOnly + createdAt."""
 
     composer_id: str
     full_conversation_headers_only: list[dict[str, Any]]
@@ -72,17 +58,7 @@ def from_dict(cls, raw: dict[str, Any], *, composer_id: str) -> "Composer":
 
 @dataclass(frozen=True)
 class WorkspaceLocalComposer:
-    """A composer entry from ``composer.composerData`` ItemTable rows.
-
-    These are summary records that live in each per-workspace ``state.vscdb``.
-    They share ``composerId`` and ``lastUpdatedAt`` with the global composer
-    schema, but they do **not** carry ``fullConversationHeadersOnly`` or
-    ``createdAt`` — those only exist on the global ``cursorDiskKV`` rows that
-    ``Composer.from_dict`` validates. Treating both shapes through the same
-    model would reject every workspace-local entry, so this slim model
-    exists to keep schema-drift detection at the boundary without conflating
-    the two storage paths.
-    """
+    """Summary composer row from per-workspace state.vscdb ItemTable; only composerId is required."""
 
     composer_id: str
     last_updated_at: Any = None
@@ -112,13 +88,7 @@ def from_dict(cls, raw: dict[str, Any]) -> "WorkspaceLocalComposer":
 
 @dataclass(frozen=True)
 class Bubble:
-    """A single message bubble within a composer.
-
-    The bubble ID lives in the row key (``bubbleId:<composer_id>:<bubble_id>``)
-    rather than the JSON value, so it is passed in explicitly. The raw dict
-    is preserved to keep downstream rendering code (which still walks the
-    untyped shape) working without modification.
-    """
+    """One message in a composer; bubble_id comes from the row key, not the JSON value."""
 
     bubble_id: str
     raw: dict[str, Any] = field(default_factory=dict)

models/errors.py

@@ -1,16 +1,8 @@
-"""Exception types for the typed-model schema-validation layer."""
-
 from __future__ import annotations
 
 
 class SchemaError(ValueError):
-    """Raised when a required Cursor schema field is missing or malformed.
-
-    Inherits from ``ValueError`` so call sites that already catch generic
-    deserialisation errors (e.g. ``json.JSONDecodeError`` is a subclass of
-    ``ValueError``) also catch schema drift without needing a separate
-    ``except`` clause. New code should catch ``SchemaError`` explicitly.
-    """
+    """Raised when a required Cursor schema field is missing or malformed."""
 
     def __init__(self, model: str, field: str, *, hint: str | None = None) -> None:
         self.model = model

models/export.py

@@ -1,5 +1,3 @@
-"""ExportEntry — typed model for an export manifest record (JSONL line)."""
-
 from __future__ import annotations
 
 from dataclasses import dataclass, field
@@ -10,13 +8,7 @@
 
 @dataclass(frozen=True)
 class ExportEntry:
-    """A single record in the export manifest (one line in ``manifest.jsonl``).
-
-    Required fields are the YAML-frontmatter keys that downstream tooling
-    indexes against: a missing ``log_id`` makes the entry unaddressable, and
-    a missing ``title`` produces unreadable output. Timestamps are optional —
-    not every Cursor conversation has both a creation and update time.
-    """
+    """One line of manifest.jsonl; log_id / title / workspace required, timestamps optional."""
 
     log_id: str
     title: str

models/workspace.py

@@ -1,5 +1,3 @@
-"""Workspace — typed model for a single Cursor workspace folder."""
-
 from __future__ import annotations
 
 from dataclasses import dataclass, field
@@ -10,14 +8,7 @@
 
 @dataclass(frozen=True)
 class Workspace:
-    """A Cursor workspace entry.
-
-    The workspace ID is the directory name on disk (Cursor uses random
-    short hashes as workspace IDs) and is passed in explicitly. ``folder``
-    is the absolute path of the project the workspace targets, read from
-    ``workspace.json``; it may legitimately be ``None`` for a CLI-only
-    workspace, so missing-folder is not a schema error.
-    """
+    """A Cursor workspace folder; folder is None for CLI-only workspaces (not a schema error)."""
 
     workspace_id: str
     folder: str | None = None

tests/test_models.py

@@ -1,21 +1,8 @@
-"""Regression tests for issue #24 — typed models + schema validation.
-
-Three fixture-based tests, per the issue acceptance criteria:
-  1. known-good Composer schema parses cleanly
-  2. missing-field Composer schema raises SchemaError
-  3. CliSessionMeta + ``_extract_blob_refs`` together exercise the binary
-     blob path (``0x0a 0x20`` marker)
-
-Run:
-    python -m unittest tests.test_models -v
-"""
-
 from __future__ import annotations
 
 import os
 import sys
 import unittest
-from pathlib import Path
 
 REPO_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 if REPO_ROOT not in sys.path: