From d44fabb4e4448765a4172624f5a5c18eaa270dec Mon Sep 17 00:00:00 2001 From: Pierre-Marie Charavel Date: Fri, 27 Feb 2026 08:20:55 -0500 Subject: [PATCH] Update CLAUDE.md to reflect MCP server and multi-directory input Co-Authored-By: Claude Sonnet 4.6 --- CLAUDE.md | 41 +++++++++++++++++++++++++++-------------- 1 file changed, 27 insertions(+), 14 deletions(-) diff --git a/CLAUDE.md b/CLAUDE.md index 03eec0a..a14c174 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -4,32 +4,45 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co ## Project Purpose -UnrealDocGenerator is a tool for generating documentation from Unreal Engine C++ header files. The `samples/` directory contains representative Unreal Engine headers used as test inputs: - -- `samples/GeomUtils.h` — Small utility header (~6 KB): geometry/math helpers in `UE::AI` namespace -- `samples/AIController.h` — Medium header (~22 KB): `AAIController` class with movement, behavior tree, perception, and blackboard integration -- `samples/GameplayTagsManager.h` — Large/complex header (~43 KB): `UGameplayTagsManager` singleton with tag trees, INI/DataTable loading, and editor support - -These samples cover a range of complexity (free functions, large classes, deeply nested types, macros, delegates) suitable for stress-testing a parser or doc generator. +UnrealDocGenerator is a tool for generating documentation from Unreal Engine C++ header files and exposing it to Claude via an MCP server for item-granularity lookups. ## Current State -Implementation complete. Three Python files + a Claude Code skill: +Implementation complete. Four Python files + MCP registration: - `ue_parser.py` — Parses UE headers into dataclasses - `ue_markdown.py` — Renders parsed data as Markdown (ultra-compact format, documented items only) - `generate.py` — CLI entry point; two-pass pipeline (parse-all → build type index → render-all) -- `.claude/skills/ue-api/SKILL.md` — Claude Code skill; queries docs via `$UE_DOCS_PATH`, falls back to `$UE_ENGINE_ROOT` source headers +- `ue_mcp_server.py` — MCP server exposing 5 tools for item-granularity doc lookups +- `.mcp.json` — Registers the MCP server with Claude Code (stdio transport) ## Usage ```bash -python generate.py samples/ docs/ # process directory → docs/ -python generate.py samples/AIController.h docs/ # single file +python generate.py [input2 ...] + +python generate.py Runtime/Engine/ Runtime/AIModule/ docs/ # multiple directories +python generate.py Runtime/Engine/Classes/GameFramework/Actor.h docs/ # single file ``` Output: one `.md` per `.h` + `docs/type-index.txt` (compact `TypeName: path/to/File.md` lookup). -Agent lookup pattern: `grep "^AController:" docs/type-index.txt` → instant single-line result. +The last argument is always the output directory. All preceding arguments are inputs (files or directories, processed recursively). + +## MCP Server + +`ue_mcp_server.py` exposes 5 tools Claude can call directly: + +| Tool | Purpose | +|---|---| +| `search_types(pattern)` | Regex search over `type-index.txt` — locate a type before fetching details | +| `get_class_overview(class_name)` | Compact view: description + inherits + property/function name lists | +| `get_member(class_name, member_name)` | Full doc for one function or property (all overloads) | +| `get_file(relative_path)` | Full `.md` file — use when you need delegates, enums, or multiple classes | +| `search_source(pattern, path_hint)` | Grep UE source `.h` files via `$UE_ENGINE_ROOT` | + +Requires `pip install mcp`. Reads `$UE_DOCS_PATH`; `search_source` also requires `$UE_ENGINE_ROOT`. + +Typical query flow: `search_types` → `get_class_overview` → `get_member` (~3 calls, ~15 lines vs ~100 lines for a full file read). ## Architecture Notes @@ -37,14 +50,14 @@ Agent lookup pattern: `grep "^AController:" docs/type-index.txt` → instant sin - `find_matching_close()` extracts balanced `{}`/`()` while skipping `//`, `/* */`, and string literals. - Only items with actual C++ doc comments (`/** */`) are included in output (`_has_doc()` filter). - Two-pass pipeline in `generate.py`: Pass 1 parses all files, Pass 2 renders with cross-reference links. +- Multi-input: each `(header, base)` pair tracks its own base dir for correct relative output paths. - Inline cross-references: base class names linked to their `.md` when in corpus (`_make_type_link()`). -- No "See also" section — inline links in `*Inherits*:` lines cover cross-file navigation. - Deprecated functions excluded from output (`not f.is_deprecated` in visibility filter). - Enums with no value descriptions use compact inline format (`Values: A, B, C`) instead of a table. - Placeholder descriptions (`------//`) filtered by `_PLACEHOLDER_RE` in `_fn_body()`. ## Critical Gotchas -- **Never** use `\s` inside a `[...]` character class with `*?` on C++ source — causes catastrophic regex backtracking (hangs on `AIController.h`). Use line-based scanning instead. +- **Never** use `\s` inside a `[...]` character class with `*?` on C++ source — causes catastrophic regex backtracking. Use line-based scanning instead. - Identifier regex must be `r'\w+'`, **not** `r'[\w:~]+'` — the latter consumes `public:` as one token, breaking access specifier detection (all members become `private`). - Module name casing: use `_caps_to_camel()` (not `.title()`) — `AIMODULE.title()` → `Aimodule` instead of `AIModule`.