uncoded 0.6.0__tar.gz → 0.7.1__tar.gz

{uncoded-0.6.0 → uncoded-0.7.1}/.agents/skills/coherence-review/SKILL.md

@@ -55,8 +55,9 @@ The review proceeds in four sweeps, each building on the previous:
 
 1. **Orient** — load the navigation index and form a mental map.
 2. **Lexical sweep** — read the namespace, look for naming-level inconsistency.
-3. **Promissory sweep** — read stubs, check each symbol's name / signature /
-   docstring for internal disagreement.
+3. **Promissory sweep** — check each symbol's name / signature / docstring
+   for internal disagreement; names and signatures from the stub, docstrings
+   via `find_symbol(include_body=True)`.
 4. **Structural sweep** — combine namespace and imports to find boundary and
    shape symptoms.
 
@@ -66,9 +67,8 @@ until the end.
 ## Step 1: Orient
 
 Read `.uncoded/namespace.yaml` in full. This is the map — directories, files,
-classes, methods, functions — in source order. Do not skim. Every public symbol
-in the codebase is listed here, and the shape of the namespace itself is
-evidence.
+classes, methods, functions. Do not skim. Every public symbol in the codebase
+is listed here, and the shape of the namespace itself is evidence.
 
 While reading, note:
 
@@ -96,8 +96,9 @@ interchangeably. Two classes called `UserRecord` and `AccountProfile` that
 model the same entity.
 
 Detection: scan the namespace for symbol clusters with verb or noun overlap.
-Where suspicion arises, spot-check the relevant stubs to confirm they overlap
-in meaning.
+Where suspicion arises, check the stub signatures and the source docstrings
+(via Serena's `find_symbol` with `include_body=True`) to confirm the
+candidates overlap in meaning.
 
 **Qualifier accretion.** Names carrying modifiers that are fossils of
 iteration: `_new`, `_v2`, `_updated`, `_legacy`, `_real`, `_proper`, `_final`,
@@ -119,16 +120,18 @@ with the rest of the codebase.
 subtly different meanings — visible as different signatures, different docstring
 content, or different domain associations.
 
-Detection: identify name collisions in the namespace, then examine the stubs to
-see whether the uses agree.
+Detection: identify name collisions in the namespace, then compare signatures
+(from the stubs) and docstrings (via Serena's `find_symbol` with
+`include_body=True`) to see whether the uses agree.
 
 ## Step 3: Promissory sweep
 
-Working from stubs, examine each public symbol's name / signature / docstring
-triple for internal disagreement.
-
-For each non-trivial public symbol (skip trivial one-liners and `__init__` with
-no meaningful body):
+Examine each public symbol's name / signature / docstring triple for internal
+disagreement. Load each source file's stub once for the names and signatures
+across that file's symbols. Then, for each non-trivial public symbol (skip
+trivial one-liners and `__init__` with no meaningful body), call Serena's
+`find_symbol` with `include_body=True` to read the docstring — the call also
+returns the body, available if a finding needs it.
 
 **Name–signature mismatch.** Does the name's verb fit the signature's return? A
 function called `validate_*` that returns the validated object rather than
@@ -149,18 +152,15 @@ describe it. "Note: this does not actually X despite the name." "Do not use
 this for Y; use Z instead." These are confessions — someone noticed drift and
 documented it rather than fixing it.
 
-The stub itself is the evidence. Quote the stub excerpt (name, signature,
-first-line docstring) verbatim in the finding.
+Quote evidence verbatim. The stub excerpt is the evidence for name and
+signature findings; the docstring returned by `find_symbol` with
+`include_body=True` is the evidence for any docstring-related finding.
 
-**When to read source.** The stub is usually sufficient for discovery — the
-inconsistency IS the mismatch between name, signature, and docstring, all of
-which the stub provides. Read the symbol body when a finding is already
-identified but confidence is genuinely uncertain: an undocumented parameter
-where significance depends on what it controls; a name–behaviour mismatch where
-the stub alone doesn't confirm it; a defensive docstring you want to verify is
-accurate. Use Serena's `find_symbol` with `include_body=True` — targeted to the
-symbol, no offset arithmetic, no risk of over-reading. Never read a whole source
-file during this sweep.
+**When to read further.** Read the body when a finding's confidence needs it:
+a name–behaviour mismatch the docstring alone doesn't settle, or a defensive
+docstring you want to verify against the body. Targeted to the symbol, no
+offset arithmetic, no risk of over-reading. Never read a whole source file
+during this sweep.
 
 ## Step 4: Structural sweep
 
@@ -244,7 +244,8 @@ Regions with two or more findings — examine these first:
 **Confidence:** high | medium | low
 
 **Evidence:**
-> Verbatim quote from namespace.yaml, stub, or import statement.
+> Verbatim quote from namespace.yaml, stub, source docstring (via
+> `find_symbol`), or import statement.
 
 One or two sentences describing the inconsistency. Not a diagnosis. Not a fix.
 
@@ -261,14 +262,14 @@ clear evidence is useful — the human can filter. A dropped finding is not.
 
 **Confidence is part of the finding, not a gate.**
 
-- `high` — the inconsistency is explicit; evidence is directly in the stub or
-  namespace
+- `high` — the inconsistency is explicit; evidence is directly in the
+  namespace, the stub, or the source docstring
 - `medium` — strongly implied but depends on judgement about intent
 - `low` — pattern-based suspicion that needs human interpretation
 
 **Evidence must be verbatim.** Quote the relevant namespace line, stub excerpt,
-or import statement exactly. A finding the human cannot quickly verify is worse
-than no finding.
+source docstring, or import statement exactly. A finding the human cannot
+quickly verify is worse than no finding.
 
 **One finding per inconsistency.** If a single symbol has a name–signature
 mismatch and a docstring–name mismatch, that is two findings on the same

{uncoded-0.6.0 → uncoded-0.7.1}/.claude/skills/coherence-review/SKILL.md

@@ -55,8 +55,9 @@ The review proceeds in four sweeps, each building on the previous:
 
 1. **Orient** — load the navigation index and form a mental map.
 2. **Lexical sweep** — read the namespace, look for naming-level inconsistency.
-3. **Promissory sweep** — read stubs, check each symbol's name / signature /
-   docstring for internal disagreement.
+3. **Promissory sweep** — check each symbol's name / signature / docstring
+   for internal disagreement; names and signatures from the stub, docstrings
+   via `find_symbol(include_body=True)`.
 4. **Structural sweep** — combine namespace and imports to find boundary and
    shape symptoms.
 
@@ -66,9 +67,8 @@ until the end.
 ## Step 1: Orient
 
 Read `.uncoded/namespace.yaml` in full. This is the map — directories, files,
-classes, methods, functions — in source order. Do not skim. Every public symbol
-in the codebase is listed here, and the shape of the namespace itself is
-evidence.
+classes, methods, functions. Do not skim. Every public symbol in the codebase
+is listed here, and the shape of the namespace itself is evidence.
 
 While reading, note:
 
@@ -96,8 +96,9 @@ interchangeably. Two classes called `UserRecord` and `AccountProfile` that
 model the same entity.
 
 Detection: scan the namespace for symbol clusters with verb or noun overlap.
-Where suspicion arises, spot-check the relevant stubs to confirm they overlap
-in meaning.
+Where suspicion arises, check the stub signatures and the source docstrings
+(via Serena's `find_symbol` with `include_body=True`) to confirm the
+candidates overlap in meaning.
 
 **Qualifier accretion.** Names carrying modifiers that are fossils of
 iteration: `_new`, `_v2`, `_updated`, `_legacy`, `_real`, `_proper`, `_final`,
@@ -119,16 +120,18 @@ with the rest of the codebase.
 subtly different meanings — visible as different signatures, different docstring
 content, or different domain associations.
 
-Detection: identify name collisions in the namespace, then examine the stubs to
-see whether the uses agree.
+Detection: identify name collisions in the namespace, then compare signatures
+(from the stubs) and docstrings (via Serena's `find_symbol` with
+`include_body=True`) to see whether the uses agree.
 
 ## Step 3: Promissory sweep
 
-Working from stubs, examine each public symbol's name / signature / docstring
-triple for internal disagreement.
-
-For each non-trivial public symbol (skip trivial one-liners and `__init__` with
-no meaningful body):
+Examine each public symbol's name / signature / docstring triple for internal
+disagreement. Load each source file's stub once for the names and signatures
+across that file's symbols. Then, for each non-trivial public symbol (skip
+trivial one-liners and `__init__` with no meaningful body), call Serena's
+`find_symbol` with `include_body=True` to read the docstring — the call also
+returns the body, available if a finding needs it.
 
 **Name–signature mismatch.** Does the name's verb fit the signature's return? A
 function called `validate_*` that returns the validated object rather than
@@ -149,18 +152,15 @@ describe it. "Note: this does not actually X despite the name." "Do not use
 this for Y; use Z instead." These are confessions — someone noticed drift and
 documented it rather than fixing it.
 
-The stub itself is the evidence. Quote the stub excerpt (name, signature,
-first-line docstring) verbatim in the finding.
+Quote evidence verbatim. The stub excerpt is the evidence for name and
+signature findings; the docstring returned by `find_symbol` with
+`include_body=True` is the evidence for any docstring-related finding.
 
-**When to read source.** The stub is usually sufficient for discovery — the
-inconsistency IS the mismatch between name, signature, and docstring, all of
-which the stub provides. Read the symbol body when a finding is already
-identified but confidence is genuinely uncertain: an undocumented parameter
-where significance depends on what it controls; a name–behaviour mismatch where
-the stub alone doesn't confirm it; a defensive docstring you want to verify is
-accurate. Use Serena's `find_symbol` with `include_body=True` — targeted to the
-symbol, no offset arithmetic, no risk of over-reading. Never read a whole source
-file during this sweep.
+**When to read further.** Read the body when a finding's confidence needs it:
+a name–behaviour mismatch the docstring alone doesn't settle, or a defensive
+docstring you want to verify against the body. Targeted to the symbol, no
+offset arithmetic, no risk of over-reading. Never read a whole source file
+during this sweep.
 
 ## Step 4: Structural sweep
 
@@ -244,7 +244,8 @@ Regions with two or more findings — examine these first:
 **Confidence:** high | medium | low
 
 **Evidence:**
-> Verbatim quote from namespace.yaml, stub, or import statement.
+> Verbatim quote from namespace.yaml, stub, source docstring (via
+> `find_symbol`), or import statement.
 
 One or two sentences describing the inconsistency. Not a diagnosis. Not a fix.
 
@@ -261,14 +262,14 @@ clear evidence is useful — the human can filter. A dropped finding is not.
 
 **Confidence is part of the finding, not a gate.**
 
-- `high` — the inconsistency is explicit; evidence is directly in the stub or
-  namespace
+- `high` — the inconsistency is explicit; evidence is directly in the
+  namespace, the stub, or the source docstring
 - `medium` — strongly implied but depends on judgement about intent
 - `low` — pattern-based suspicion that needs human interpretation
 
 **Evidence must be verbatim.** Quote the relevant namespace line, stub excerpt,
-or import statement exactly. A finding the human cannot quickly verify is worse
-than no finding.
+source docstring, or import statement exactly. A finding the human cannot
+quickly verify is worse than no finding.
 
 **One finding per inconsistency.** If a single symbol has a name–signature
 mismatch and a docstring–name mismatch, that is two findings on the same

{uncoded-0.6.0 → uncoded-0.7.1}/.serena/project.yml

@@ -85,6 +85,7 @@ excluded_tools:
 - edit_memory
 - delete_memory
 - rename_memory
+- activate_project
 - onboarding
 - check_onboarding_performed
 - open_dashboard

{uncoded-0.6.0 → uncoded-0.7.1}/.uncoded/namespace.yaml

@@ -2,8 +2,7 @@
 # Generated by uncoded — do not edit; regeneration overwrites.
 #
 # Pure key hierarchy (no lists, no values); indent to zoom in.
-# Directory keys end with "/". Entries within a file or class appear
-# in source order.
+# Directory keys end with "/".
 #
 # For signatures and types, see stubs:
 #   src/foo/bar.py  →  .uncoded/stubs/src/foo/bar.pyi
@@ -13,11 +12,11 @@ src/:
     __init__.py:
       __version__:
     cli.py:
-      DEFAULT_MAP_OUTPUT:
       _sync:
       main:
     config.py:
       find_pyproject_toml:
+      read_project_name:
       read_source_roots:
       read_instruction_files:
     extract.py:
@@ -30,18 +29,17 @@ src/:
         constants:
         classes:
         functions:
-      _property_kind:
+      property_kind:
       _assign_target_name:
       extract_module:
       iter_source_files:
-      walk_source:
+      extract_modules:
     instruction_files.py:
       MARKER_START:
       MARKER_END:
       DEFAULT_INSTRUCTION_FILES:
       _SECTION_BODY:
       SECTION:
-      generate_section:
       _replace_or_append:
       sync_instruction_file:
     namespace_map.py:
@@ -51,16 +49,16 @@ src/:
       build_map:
       render_map:
     serena_setup.py:
+      _Status:
       SERENA_VERSION:
       MCP_SERVER_SERENA:
-      SERENA_PROJECT_YML:
+      SERENA_PROJECT_FIELDS:
       SERENA_ALLOWED_TOOLS:
       _STATUS_VERB:
-      read_project_name:
       _sync_mcp_json:
-      _sync_serena_project:
+      _write_serena_project_if_absent:
       _sync_claude_settings:
-      setup_serena:
+      setup:
     skill.py:
       SKILL_OUTPUTS:
       LEGACY_SKILL_OUTPUTS:
@@ -68,7 +66,6 @@ src/:
       sync_skill:
     stubs.py:
       VALUE_WIDTH_CAP:
-      DEFAULT_STUBS_OUTPUT:
       StubParam:
         name:
         annotation:
@@ -76,17 +73,15 @@ src/:
         name:
         params:
         return_annotation:
-        docstring_excerpt:
         is_async:
       StubAssignment:
         name:
         annotation:
         value_source:
-        is_type_alias:
+        is_pep695_alias:
       StubClass:
         name:
         bases:
-        docstring_excerpt:
         attributes:
         methods:
       StubModule:
@@ -95,7 +90,6 @@ src/:
         constants:
         classes:
         functions:
-      _first_sentence:
       _extract_params:
       _render_value:
       _extract_assignment:
@@ -109,6 +103,7 @@ src/:
       _render_assignment:
       render_stub:
       _generate_stubs:
+      _write_stubs:
       build_stubs:
     sync.py:
       sync_file:
@@ -118,31 +113,42 @@ tests/:
     TestSyncApplyMode:
       test_writes_namespace_map_stubs_and_instruction_file:
       test_idempotent_second_run:
+      test_dedupes_when_claude_md_is_symlink_to_agents_md:
+      test_instruction_file_outside_project_uses_absolute_path:
       test_error_when_no_pyproject_toml:
       test_error_when_source_root_missing:
+      test_error_when_uncoded_section_missing:
+      test_skip_warning_emitted_once_per_broken_file:
+      test_anchors_reads_and_writes_at_project_root_when_cwd_is_subdir:
+      test_artefacts_match_when_run_from_subdir_vs_project_root:
     TestSyncCheckMode:
       test_returns_one_and_does_not_write_on_empty_repo:
       test_returns_zero_when_index_is_up_to_date:
       test_returns_one_when_source_changes_after_sync:
       test_returns_one_when_source_file_deleted:
       test_returns_one_when_instruction_file_drifts:
+      test_dedupes_when_claude_md_is_symlink_to_agents_md:
       test_error_still_returns_one:
     TestMainDispatch:
       test_sync_subcommand_runs_in_apply_mode:
       test_check_subcommand_runs_in_check_mode:
       test_check_subcommand_returns_zero_on_fresh_index:
-      test_setup_serena_subcommand:
+      test_setup_subcommand:
       test_no_subcommand_is_an_error:
       test_unknown_subcommand_is_an_error:
     _init_repo:
   test_config.py:
     TestFindPyprojectToml:
-      test_finds_in_cwd:
-      test_finds_in_parent:
+      test_finds_at_start:
+      test_finds_in_parent_of_start:
       test_returns_none_if_not_found:
+    TestReadProjectName:
+      test_reads_name_from_pyproject_toml:
+      test_falls_back_to_start_name_when_no_pyproject:
+      test_falls_back_to_start_name_when_no_project_section:
+      test_falls_back_to_start_name_when_name_missing:
     TestReadSourceRoots:
       test_reads_source_roots:
-      test_raises_if_no_pyproject_toml:
       test_raises_if_no_uncoded_section:
     TestReadInstructionFiles:
       test_returns_default_when_no_pyproject_toml:
@@ -150,7 +156,7 @@ tests/:
       test_reads_configured_list:
       test_empty_list_is_respected:
   test_extract.py:
-    TestExtractModule:
+    TestExtractModuleFromSource:
       test_classes_and_functions:
       test_async_functions_and_methods:
       test_empty_module:
@@ -158,20 +164,27 @@ tests/:
       test_type_alias_classic:
       test_type_alias_pep695:
       test_tuple_unpacking_skipped:
+      test_chained_assignment_skipped:
       test_unannotated_class_variable:
-      test_module_with_only_constants_is_kept:
+      test_class_tuple_unpacking_skipped:
       test_annotated_attributes:
       test_property_classified_as_attribute:
       test_property_setter_and_deleter_suppressed:
+      test_non_property_decorator_classified_as_method:
       test_preserves_source_order:
-    TestWalkSource:
+    TestIterAndExtract:
       test_basic_walk:
       test_nested_subpackage:
       test_includes_init_with_symbols:
       test_skips_empty_init:
       test_skips_syntax_errors:
+    TestExtractModulesFromFiles:
+      test_returns_module_info_per_parseable_file:
+      test_preserves_source_order:
+      test_module_with_only_constants_is_kept:
+      test_skips_files_with_no_symbols:
   test_instruction_files.py:
-    TestGenerateSection:
+    TestSection:
       test_contains_markers:
       test_markers_in_order:
       test_ends_with_newline:
@@ -187,6 +200,9 @@ tests/:
       test_does_not_create_file:
       test_does_not_update_existing_file:
       test_reports_no_change_when_clean:
+    TestSyncInstructionFileProjectRootAnchor:
+      test_project_root_anchors_create_independent_of_cwd:
+      test_project_root_anchors_update_of_existing_file:
   test_namespace_map.py:
     TestBuildMap:
       test_single_file:
@@ -195,9 +211,7 @@ tests/:
       test_class_with_attributes_and_methods:
       test_function_is_none:
       test_class_with_no_members:
-      test_source_order_preserved:
       test_module_level_constants:
-      test_constants_precede_classes_and_functions:
     TestRenderMap:
       test_roundtrips_through_yaml:
       test_preserves_insertion_order:
@@ -208,16 +222,13 @@ tests/:
   test_serena_setup.py:
     REPO_ROOT:
     EXPECTED_EXCLUDED_TOOLS:
-    TestReadProjectName:
-      test_reads_name_from_pyproject_toml:
-      test_falls_back_to_cwd_name_when_no_pyproject:
-      test_falls_back_to_cwd_name_when_no_project_section:
-      test_falls_back_to_cwd_name_when_name_missing:
-    TestSetupSerena:
+    EXPECTED_MCP_ARGS:
+    TestSetup:
       _run:
       test_creates_all_three_files:
       test_mcp_json_is_valid_and_pins_version:
       test_serena_project_yml_uses_ty_and_ignores_uncoded:
+      test_serena_project_yml_escapes_yaml_special_chars_in_name:
       test_claude_settings_enables_serena_and_allowlists_tools:
       test_idempotent:
       test_merges_into_existing_mcp_json:
@@ -225,9 +236,11 @@ tests/:
       test_merges_into_existing_claude_settings:
       test_does_not_overwrite_existing_serena_project_yml:
       test_does_not_duplicate_on_second_merge:
-      test_falls_back_to_cwd_name_when_no_pyproject:
+      test_appends_missing_allowed_tool_to_existing_settings:
+      test_setup_uses_root_name_when_no_pyproject:
+      test_setup_reads_name_from_pyproject:
     TestRepoDogfooding:
-      test_repo_mcp_json_pins_same_serena_version:
+      test_repo_mcp_json_matches_template_contract:
       test_repo_claude_settings_allowlists_every_serena_tool:
       test_repo_serena_project_yml_matches_template_contract:
   test_skill.py:
@@ -243,6 +256,9 @@ tests/:
       test_check_mode_reports_no_change_when_in_sync:
       test_removes_legacy_skill_files:
       test_check_mode_reports_legacy_skill_files_without_removing:
+    TestSyncSkillProjectRootAnchor:
+      test_project_root_anchors_writes_independent_of_cwd:
+      test_project_root_anchors_legacy_removal_independent_of_cwd:
   test_stubs.py:
     TestExtractStub:
       test_simple_function:
@@ -252,9 +268,7 @@ tests/:
       test_class_with_attributes_and_methods:
       test_class_with_bases:
       test_class_no_bases:
-      test_docstring_first_sentence_only:
-      test_no_docstring:
-      test_kwargs_and_varargs:
+      test_extract_params_covers_input_kind:
       test_imports_collected:
       test_syntax_error_raises:
       test_source_order_preserved:
@@ -266,6 +280,8 @@ tests/:
       test_type_alias_classic:
       test_type_alias_pep695:
       test_tuple_unpacking_skipped:
+      test_ann_assign_non_name_target_skipped:
+      test_class_tuple_unpacking_skipped:
       test_class_with_unannotated_attribute:
       test_property_rendered_as_attribute:
       test_property_without_return_annotation:
@@ -275,10 +291,10 @@ tests/:
       test_imports_rendered:
       test_rendered_stub_has_no_line_range_comments:
       test_async_function_prefix:
-      test_function_with_annotations:
-      test_docstring_excerpt_rendered:
+      test_render_param_covers_input_kind:
+      test_return_annotation_rendered:
       test_class_with_bases:
-      test_class_no_bases:
+      test_class_with_no_members_renders_body:
       test_attribute_with_annotation:
       test_method_indented:
       test_ends_with_newline:
@@ -289,8 +305,10 @@ tests/:
       test_constant_bare_annotation_rendered:
       test_type_alias_pep695_rendered:
       test_unannotated_class_attribute_rendered:
+      test_renders_valid_python_for_representative_source:
     TestBuildStubs:
       _setup:
+      _build:
       test_writes_expected_stubs:
       test_removes_orphan_stub_when_source_deleted:
       test_removes_orphan_stub_when_source_renamed:
@@ -299,12 +317,21 @@ tests/:
       test_no_op_when_clean:
       test_reports_count_on_first_build:
       test_reports_zero_when_clean:
+      test_skips_module_with_no_symbols:
     TestBuildStubsCheckMode:
       _setup:
+      _build:
       test_does_not_write_stub_in_check_mode:
       test_zero_changes_when_clean:
       test_detects_stale_stub_content:
       test_detects_orphan_stub_without_removing_it:
+    TestWriteStubs:
+      test_writes_stubs:
+      test_check_mode_does_not_write:
+      test_prunes_orphan_stubs:
+      test_project_root_anchors_writes_independent_of_cwd:
+      test_project_root_anchors_orphan_pruning_independent_of_cwd:
+      test_source_root_outside_project_root_skips_cleanup:
   test_sync.py:
     TestSyncFile:
       test_creates_missing_file:
@@ -320,6 +347,12 @@ tests/:
       test_noop_when_absent:
       test_check_mode_does_not_remove:
       test_check_mode_reports_noop_when_absent:
+    TestSyncFileProjectRootAnchor:
+      test_project_root_anchors_write_independent_of_cwd:
+      test_project_root_preserves_relative_path_in_message:
+      test_absolute_path_makes_project_root_a_no_op:
+    TestRemoveFileProjectRootAnchor:
+      test_project_root_anchors_removal_independent_of_cwd:
   test_uncoded.py:
     test_import:
     test_version_is_exposed:

uncoded-0.7.1/.uncoded/stubs/src/uncoded/cli.pyi

@@ -0,0 +1,19 @@
+# src/uncoded/cli.py
+
+import argparse
+import sys
+from pathlib import Path
+from uncoded.config import find_pyproject_toml, read_instruction_files, read_source_roots
+from uncoded.extract import extract_modules, iter_source_files
+from uncoded.instruction_files import sync_instruction_file
+from uncoded.namespace_map import build_map, render_map
+from uncoded.serena_setup import setup
+from uncoded.skill import sync_skill
+from uncoded.stubs import build_stubs
+from uncoded.sync import sync_file
+
+def _sync(*, start: Path | None, check: bool) -> int:
+    ...
+
+def main() -> int:
+    ...

uncoded-0.7.1/.uncoded/stubs/src/uncoded/config.pyi

@@ -0,0 +1,17 @@
+# src/uncoded/config.py
+
+import tomllib
+from pathlib import Path
+from uncoded.instruction_files import DEFAULT_INSTRUCTION_FILES
+
+def find_pyproject_toml(start: Path) -> Path | None:
+    ...
+
+def read_project_name(start: Path) -> str:
+    ...
+
+def read_source_roots(pyproject_path: Path) -> list[Path]:
+    ...
+
+def read_instruction_files(start: Path) -> list[Path]:
+    ...

uncoded-0.7.1/.uncoded/stubs/src/uncoded/extract.pyi

@@ -0,0 +1,33 @@
+# src/uncoded/extract.py
+
+import ast
+import sys
+from collections.abc import Iterable, Iterator
+from dataclasses import dataclass, field
+from pathlib import Path
+
+def property_kind(node: ast.FunctionDef | ast.AsyncFunctionDef) -> str | None:
+    ...
+
+def _assign_target_name(node: ast.Assign | ast.AnnAssign) -> str | None:
+    ...
+
+def extract_module(source: str, rel_path: str) -> ModuleInfo:
+    ...
+
+def iter_source_files(source_root: Path, project_root: Path) -> Iterator[tuple[str, str]]:
+    ...
+
+def extract_modules(files: Iterable[tuple[str, str]]) -> list[ModuleInfo]:
+    ...
+
+class ClassInfo:
+    name: str
+    attributes: list[str] = field(default_factory=list)
+    methods: list[str] = field(default_factory=list)
+
+class ModuleInfo:
+    rel_path: str
+    constants: list[str] = field(default_factory=list)
+    classes: list[ClassInfo] = field(default_factory=list)
+    functions: list[str] = field(default_factory=list)

{uncoded-0.6.0 → uncoded-0.7.1}/.uncoded/stubs/src/uncoded/instruction_files.pyi

@@ -9,14 +9,8 @@ DEFAULT_INSTRUCTION_FILES = [Path('CLAUDE.md'), Path('AGENTS.md')]
 _SECTION_BODY = ...
 SECTION = f'{MARKER_START}\n{_SECTION_BODY}\n{MARKER_END}\n'
 
-def generate_section() -> str:
-    """Return the full delimited uncoded section for an instruction file."""
-    ...
-
 def _replace_or_append(existing: str, section: str) -> str:
-    """Replace the delimited section in existing text, or append it if absent."""
     ...
 
-def sync_instruction_file(path: Path, *, check: bool) -> bool:
-    """Write or update the uncoded navigation section in an instruction file."""
+def sync_instruction_file(path: Path, *, project_root: Path, check: bool) -> bool:
     ...