一个为 AI 编码代理提供 Godot Engine 文档服务的 MCP 服务器。
与 Unity 和 Unreal 相比，Godot 在 LLM 训练数据中的代表性不足，因此代理经常出现幻觉，编造 GDScript API、发明节点名称，并将 Godot 3.x 的用法套用到 4.x 项目上。该服务器让它们基于真实文档：对手册和教程进行全文搜索，以及基于引擎自身 API 元数据构建的结构化类引用。
基于 MCP 规范修订版 2026-07-28 构建，使用官方的 mcp Python SDK v2，同一服务器也支持所有早期协议修订版。
状态：开发中。 服务器、其工具和资源、命令行以及发布索引的每周工作流均可正常工作，并且已发布 Godot 4.7 的预构建索引，因此
sync无需构建即可安装。尚未发布到 PyPI——请按如下方式从 git 安装。作为 Claude Code 插件（带技能）打包是后续阶段。
文档被离线索引到单个 SQLite 数据库中，因此查找速度快且无需网络连接。
内容 | 来源 | 原因 |
手册、教程、入门指南 |
| 作者编写的散文 |
类引用 |
| 结构化且类型化—— |
docs.godotengine.org 位于 Cloudflare 质询之后，该质询拒绝非浏览器客户端，因此不使用渲染后的站点作为来源。所有内容均来自 GitHub。
Related MCP server: Godot MCP Documentation Server
安装预构建索引。只需下载约 8 MB，每个 Godot 版本只需执行一次。
uvx --from git+https://github.com/mcintalmo/godot-docs-mcp godot-docs-mcp sync
uvx --from git+https://github.com/mcintalmo/godot-docs-mcp godot-docs-mcp status然后让客户端指向它。对于 Claude Code：
claude mcp add godot-docs -- \
uvx --from git+https://github.com/mcintalmo/godot-docs-mcp godot-docs-mcp没有针对您想要的版本发布索引，或者正在处理解析器？请从源代码构建。这需要 git，大约需要一分钟：
uvx --from git+https://github.com/mcintalmo/godot-docs-mcp godot-docs-mcp sync --rebuild问一个模型通常会答错的问题——“move_and_slide 的正确 4.x 签名是什么？”——答案应该从 get_godot_class_member 返回为 bool move_and_slide()，没有参数。在 Godot 3.x 中需要几个参数，而大多数训练数据记住的正是这一点。
命令 | 用途 |
| 通过 stdio 提供服务。这是客户端启动的命令； |
| 下载预构建索引。 |
| 索引了什么、基于什么构建的，以及是否已过时。 |
| 使用工具相同的代码运行查询，并打印排名 |
| 为已安装的索引写入发布资产。用于构建工作流 |
每个设置也是一个环境变量，前缀为 GODOT_DOCS_MCP_——请参阅 config.py 获取完整表格。
文档会随时更改，因此有两件事会独立跟踪它。
一个每周工作流检查上游提交是否已移动，如果已移动，则重新构建索引，查询整个语料库以确认其能回答，并将其发布为发布资产。godot-docs-mcp sync 就是它到达您的方式。
正在运行的服务器每天在后台检查一次是否有更新的索引已发布。它永远不会下载一个——过时的索引仍然能针对其构建的版本正确回答，是否替换由您决定。它所做的只是说明：在 godot://status 中，以及在搜索结果上以一行 notice 的形式说明。如果检查无法运行，则答案为“无法检查”，绝不会是“最新”。
每个工具都是只读的，每个响应都有限制，因此一次调用不会挤占它本应告知的对话。被截断的响应会说明，并指出返回其余部分的调用。
工具 | 用途 |
| 对手册、教程和类引用进行排名搜索。支持用简单语言提问 |
| 获取一个页面或其中的单个部分 |
| 类的继承链和签名映射——没有散文，因此 300 KB 的类条目也能回答 |
| 单个成员的精确签名和完整描述 |
| GDScript 和 C# 代码片段，以及它们来自的页面和标题 |
URI | 内容 |
| 索引版本、计数以及构建它的上游提交 |
| 语料库的地图：包含什么、如何寻址、调用哪个工具 |
| 一个页面或其中的一个部分 |
| 一个类或其中的一个成员 |
515 个页面和 1,078 个类不会在 resources/list 中枚举。模板是每个工具已经返回的 URI，因此搜索命中后，主机无需额外工具调用即可跟进。
需要 uv。
uv sync --dev
uv run pre-commit install命令 | 用途 |
| 单元测试套件（无网络，隔离运行） |
| 针对真实索引的集成测试 |
| 代码检查和格式化 |
| 类型检查 |
| 运行 CI lint 作业的所有内容 |
有关架构和贡献约定，请参阅 CLAUDE.md。
MIT。Godot 文档本身由 Juan Linietsky、Ariel Manzur 和 Godot 社区根据 CC BY 4.0 许可；本项目重新分发其索引形式，且不更改该许可证。
Return one manual or tutorial page as Markdown, or a single section of it. Pass the uri from a search hit exactly as it was given: the #... on the end names the section that matched, and that section is what comes back -- section_title says which. Drop the #... to read the whole page. Pages are cut to fit the response budget: when truncated is set, section_titles lists the titles section accepts, so the rest can be read a section at a time. For a class-reference entry call get_godot_class instead -- this tool serves the prose documentation only.
| Name | Required | Description | Default |
|---|---|---|---|
| uri | Yes | A `godot://doc/...` URI, as returned by search_godot_docs. | |
| section | No | One title from the page's `section_titles`, or a section anchor. Overrides any `#...` on the uri. Omit to get whatever the uri asks for. |
| Name | Required | Description |
|---|---|---|
| uri | Yes | Canonical `godot://doc/...` URI of the page. |
| hint | No | What to call next to see what was omitted, or to narrow the request. Empty when nothing was left out. |
| kind | Yes | Which part of the documentation this belongs to. |
| title | Yes | Page title. |
| markdown | Yes | The page or section content, as Markdown. |
| truncated | No | Whether content was omitted to stay within the budget. |
| breadcrumb | No | Trail of ancestor sections. |
| section_title | No | The single section returned, when one was asked for; empty for a whole page. |
| section_titles | No | Every section on the page, in order. These are the values `section` accepts, so a truncated page can be re-read one section at a time. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Beyond the readOnlyHint annotation, the description discloses important behaviors: pages are truncated to fit the response budget, truncated indicates this, and section_titles lists acceptable section values. It also explains the meaning of the #... suffix and how section overrides it, which is valuable context not visible in the schema.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is dense but every sentence carries necessary operational information. It front-loads the core purpose, then explains uri semantics, truncation handling, and sibling differentiation without wasted words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a documentation-fetching tool with a read-only annotation and an output schema, the description covers all needed calling context: source of the uri, section selection, truncation behavior, and the boundary with get_godot_class. Nothing essential is missing.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Although schema coverage is 100%, the description adds significant meaning beyond the schema: it explains how to pass the uri exactly, what the #... suffix does, how to drop it for the whole page, and how section interacts with the uri. This goes well beyond the baseline for high schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific verb and resource: 'Return one manual or tutorial page as Markdown, or a single section of it.' It also distinguishes itself from get_godot_class by explicitly saying this tool serves prose documentation only, so an agent can tell it apart from siblings.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It gives explicit instructions on when to use the tool: pass the uri from a search hit exactly, drop the #... to read the whole page, and use section to read truncated content section by section. It also names the alternative get_godot_class for class-reference entries, providing clear routing guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Return a map of one engine class: what it inherits, what it is for, and the signature of every member it declares, without their descriptions. This is the cheap first call for any question about a class -- some class entries run past 300 KB, so the map is what makes them answerable. Choose the member needed from members, then call get_godot_class_member for its full text. Inherited members are not listed; request a class named in inheritance_chain to see those.
| Name | Required | Description | Default |
|---|---|---|---|
| include | No | List only members of this kind. Omit for every kind. Worth setting when a full map came back `truncated`. | |
| class_name | Yes | The class name, exactly as the engine spells it, e.g. 'Node2D'. |
| Name | Required | Description |
|---|---|---|
| uri | Yes | Canonical `godot://class/...` URI. |
| hint | No | What to call next to see what was omitted, or to narrow the request. Empty when nothing was left out. |
| name | Yes | Class name. |
| members | No | Declared members, in source order. Inherited members are not listed; request a class from `inheritance_chain` to see those. |
| brief_md | No | One-line summary. |
| inherits | No | Immediate base class, empty for roots. |
| truncated | No | Whether content was omitted to stay within the budget. |
| tutorials | No | `(title, url)` pairs of linked tutorials. |
| deprecated | No | Why this class is deprecated; empty if not. |
| experimental | No | Why this class is experimental; empty if not. |
| description_md | No | Full class description, as Markdown. |
| inheritance_chain | No | Base classes from the immediate parent up to the root. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description discloses behavior such as returning a map, potential truncation, and omission of inherited members. Combined with readOnlyHint=true annotation, the read-only nature is clear. No contradiction with annotations; description adds valuable context beyond the hints.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise and well-structured, using three sentences to convey purpose, usage guidance, and edge cases (truncation, inherited members). No redundant wording; each sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the output schema exists, the description adequately explains the return type (map) and how to interact with the results (select a member, follow inheritance_chain). It also covers the truncation scenario. The tool is fully understandable within its context.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers all parameters (class_name and include) with clear descriptions. The tool description does not add additional meaning beyond what is already in the schema, so a baseline score of 3 is appropriate given 100% schema coverage.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool returns a map of one engine class, including inheritance, purpose, and member signatures without descriptions. It distinguishes itself as the 'cheap first call' for class questions, differentiating from siblings like get_godot_class_member and search_godot_docs.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly advises when to use this tool (first call for any class question) and when to use alternatives (get_godot_class_member for member details, request inherited classes from inheritance_chain). It also explains the truncation behavior and how to follow up with include parameter.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Return the exact signature and full description of one member of one class. Use this to confirm a signature before writing a call: Godot's API changed substantially between 3.x and 4.x, and a remembered signature is often the 3.x one. Several members can share a name -- Godot overloads constructors and operators -- so members may hold more than one. If the name is wrong, the error names the closest members the class does declare, which is usually where a 3.x name is corrected to its 4.x form.
| Name | Required | Description | Default |
|---|---|---|---|
| class_name | Yes | The class name, exactly as the engine spells it, e.g. 'Node2D'. | |
| member_kind | No | Restrict to one kind, for the few names a class uses for members of more than one kind. | |
| member_name | Yes | The member name, e.g. 'move_and_slide'. |
| Name | Required | Description |
|---|---|---|
| hint | No | What to call next to see what was omitted, or to narrow the request. Empty when nothing was left out. |
| members | No | Every member of that name, in declaration order. |
| truncated | No | Whether content was omitted to stay within the budget. |
| class_name | Yes | The owning class. |
| member_name | Yes | The member name that was requested. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Beyond the readOnlyHint annotation, the description reveals meaningful behavior: overloaded constructors/operators can lead to multiple matches, and an incorrect name returns the closest declared members, often exposing the 3.x-to-4.x rename. This extra context is valuable for interpreting results and handling ambiguous cases.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Three sentences with every sentence earning its place: it states the primary behavior, gives the concrete use-case with version mismatch, and warns about overloads and error behavior. It is concise and front-loaded.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With an output schema present, the return values are already documented. The description adds the only important context missing from structured fields: the version-confliction motivation, overload ambiguity, and the diagnostic behavior when a name is wrong. This makes the description functionally complete for the tool's simple lookup purpose.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, and each parameter already has a clear description with examples. The tool description adds no parameter-specific semantics beyond what the schema provides, so the baseline 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with the specific verb-object pair 'Return the exact signature and full description of one member of one class,' making the tool's purpose immediately clear. The phrase 'one member of one class' implicitly distinguishes it from class-level lookups like get_godot_class, so an agent can separate it from siblings.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly tells when to use the tool: 'confirm a signature before writing a call,' specifically because Godot's 3.x and 4.x APIs differ. It does not name the alternative tools for 'when not to use it,' but the when-to-use guidance is concrete and actionable.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search runnable code examples extracted from the documentation. Each sample carries its language tag, the URI of the page or class it came from, and the heading it appeared under. Use this when working code is wanted; use search_godot_docs when an explanation is wanted.
| Name | Required | Description | Default |
|---|---|---|---|
| limit | No | Maximum results to return. Values above 50 are reduced to 50 rather than rejected. Fewer may come back: the response is also capped by total size. | |
| query | Yes | What to look for, in plain words. A question is fine. | |
| language | No | Restrict to one language tag, e.g. 'gdscript' or 'csharp'. Omit for every language. |
| Name | Required | Description |
|---|---|---|
| hint | No | What to call next to see what was omitted, or to narrow the request. Empty when nothing was left out. |
| query | Yes | The query as it was actually searched. |
| samples | No | Examples, best first. |
| truncated | No | Whether content was omitted to stay within the budget. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations indicate read-only, non-destructive behaviorni. The description adds that results carry language tag, URI, and heading, which gives the agent context about what to expect in results (though the output schema likely also documents this). It doesn't mention any side effects or limitations, but for a search tool that's fine. The annotation covers safety, so the description adds marginal value about result content.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, front-loaded with purpose, then usage guidance. Efficient and clear.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The description covers purpose, what results contain, and when to use this tool vs the sibling. With output schema present and full schema coverage, this is complete for the agent to decide and invoke correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, meaning all parameters (limit and query) have descriptions. The description itself doesn't add any parameter-specific semantics beyond what the schema provides)Skip. So baseline 3.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb ('Search') and a precise resource ('runnable code examples extracted from the documentation'), and immediately differentiates from the sibling search tool by stating the intended use case (working code vs. explanations). This makes the tool's purpose unmistakable.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly states when to use this tool ('when working code is wanted') and when to use the alternative ('use search_godot_docs when an explanation is wanted'). This clearly guides the agent to choose between siblings.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the Godot documentation -- manual, tutorials, and class reference -- and return ranked matches. Start here whenever the exact class or page is not already known. Natural-language questions work as well as keywords: framing words are stripped before searching, so 'how do I make a character jump' finds the tutorial rather than every page titled 'How do I'. Follow a hit with its own fields: uri is what fetch_godot_doc takes, and a hit from the class reference also carries class_name and member_name, which are what get_godot_class and get_godot_class_member take. At most one hit is returned per page or class; also_matched counts the further matches on that same source that were folded into it.
| Name | Required | Description | Default |
|---|---|---|---|
| kind | No | Restrict results to one part of the documentation. | |
| limit | No | Maximum results to return. Values above 50 are reduced to 50 rather than rejected. Fewer may come back: the response is also capped by total size. | |
| query | Yes | What to look for, in plain words. A question is fine. |
| Name | Required | Description |
|---|---|---|
| hint | No | What to call next to see what was omitted, or to narrow the request. Empty when nothing was left out. |
| hits | No | Matches, best first, at most one per page or class. |
| query | Yes | The query as it was actually searched, which may differ from what was asked when question phrasing was stripped out. |
| notice | No | An advisory about the index itself -- that a newer one has been published, for instance. It says nothing about the hits, which are as correct for the indexed version as they ever were. Empty when there is nothing to report. |
| truncated | No | Whether content was omitted to stay within the budget. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
ReadOnlyHint already covers the read-only nature. The description adds valuable behavior details: output fields (uri, class_name, member_name) and the handling of limit reductions. Slightly more detail could be given about ranking, but overall transparent.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Every sentence contributes: search scope, usage direction, query example, output fields, and limit caveats are all packed without redundancy. The description is long but efficient.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's search nature and relationship to sibling tools, the description fully covers what results look like, how they map to sibling tools, and edge cases (limit reduction). No gaps for correct invocation.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema descriptions are thorough for all three parameters, including nuances like plain words for query, restriction by kind, and limit behavior (reduction to 50, size cap). The description reinforces these but does not introduce entirely new meaning beyond the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb ('Search') and resource ('Godot documentation'), and explicitly mentions it is for when the exact class or page is not known, distinguishing it from sibling tools that fetch specific docs.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides explicit direction: 'Start here whenever the exact class or page is not already known.' Includes example of natural-language query and notes that framing words are stripped, guiding the agent on appropriate queries.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.0fetch_godot_docget_godot_classget_godot_class_membersearch_godot_code_examplessearch_godot_docsScored across 5 tools
Each tool targets a distinct retrieval need: prose pages, document/code search, class overviews, and individual member details. Cross-references are explicit, such as fetch_godot_doc directing class-reference lookups to get_godot_class, so an agent is unlikely to confuse them.
All tool names use a lowercase snake_case verb_noun pattern: fetch/search/get followed by the object. The singular doc in fetch_godot_doc is meaningful because it fetches one page, while search_godot_docs searches the corpus, and get_godot_class_member extends get_godot_class naturally.
Five tools cover the full Godot docs workflow without bloat: search prose/classes, fetch prose, get class summaries, get member details, and search code examples. Each tool has a distinct role and none is redundant.
The surface covers the main documentation workflows: discovering sources via search, retrieving prose pages section-by-section, extracting cheap class overviews, drilling into individual members, and finding runnable examples. Search hits carry the exact fields needed by the retrieval tools, so there are no dead ends.
Serves your design system and coding standards to coding agents, so they stop guessing.
Versioned documentation registry and semantic search for AI tools and coding assistants.
Provide your AI coding tools with token-efficient access to up-to-date technical documentation for…
The documentation, as a tool your agent can call: 950+ AI-dev guides. Search + fetch tools.
