# fastmcp - Doramagic AI Context Pack

> 定位：安装前体验与判断资产。它帮助宿主 AI 有一个好的开始，但不代表已经安装、执行或验证目标项目。

## 充分原则

- **充分原则，不是压缩原则**：AI Context Pack 应该充分到让宿主 AI 在开工前理解项目价值、能力边界、使用入口、风险和证据来源；它可以分层组织，但不以最短摘要为目标。
- **压缩策略**：只压缩噪声和重复内容，不压缩会影响判断和开工质量的上下文。

## 给宿主 AI 的使用方式

你正在读取 Doramagic 为 fastmcp 编译的 AI Context Pack。请把它当作开工前上下文：帮助用户理解适合谁、能做什么、如何开始、哪些必须安装后验证、风险在哪里。不要声称你已经安装、运行或执行了目标项目。

## Claim 消费规则

- **事实来源**：Repo Evidence + Claim/Evidence Graph；Human Wiki 只提供显著性、术语和叙事结构。
- **事实最低状态**：`supported`
- `supported`：可以作为项目事实使用，但回答中必须引用 claim_id 和证据路径。
- `weak`：只能作为低置信度线索，必须要求用户继续核实。
- `inferred`：只能用于风险提示或待确认问题，不能包装成项目事实。
- `unverified`：不得作为事实使用，应明确说证据不足。
- `contradicted`：必须展示冲突来源，不得替用户强行选择一个版本。

## 它最适合谁

- **希望把专业流程带进宿主 AI 的用户**：仓库包含 Skill 文档。 证据：`.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md`, `.claude/skills/review-pr/SKILL.md` 等 Claim：`clm_0003` supported 0.86

## 它能做什么

- **AI Skill / Agent 指令资产库**（可做安装前预览）：项目包含可被宿主 AI 读取的 Skill 或 Agent 指令文件，可用于把专业流程带入 Claude、Codex、Cursor 等宿主。 证据：`.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md`, `.claude/skills/review-pr/SKILL.md` 等 Claim：`clm_0001` supported 0.86
- **命令行启动或安装流程**（需要安装后验证）：项目文档中存在可执行命令，真实使用需要在本地或宿主环境中运行这些命令。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md` Claim：`clm_0002` supported 0.86

## 怎么开始

- `pip install fastmcp[anthropic]` 证据：`examples/sampling/README.md` Claim：`clm_0004` supported 0.86
- `curl -X POST \` 证据：`fastmcp_slim/fastmcp/contrib/component_manager/README.md` Claim：`clm_0005` supported 0.86
- `curl -fsSL "https://github.com/cli/cli/releases/download/${VERSION}/${TARBALL}" | tar -xz -C "$TEMP"` 证据：`.claude/hooks/session-init.sh` Claim：`clm_0006` unverified 0.25

## 继续前判断卡

- **当前建议**：先做权限沙盒试用
- **为什么**：项目存在安装命令、宿主配置或本地写入线索，不建议直接进入主力环境，应先在隔离环境试装。

### 30 秒判断

- **现在怎么做**：先做权限沙盒试用
- **最小安全下一步**：先跑 Prompt Preview；若仍要安装，只在隔离环境试装
- **先别相信**：工具权限边界不能在安装前相信。
- **继续会触碰**：命令执行、宿主 AI 配置、本地环境或项目文件

### 现在可以相信

- **适合人群线索：希望把专业流程带进宿主 AI 的用户**（supported）：有 supported claim 或项目证据支撑，但仍不等于真实安装效果。 证据：`.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md`, `.claude/skills/review-pr/SKILL.md` 等 Claim：`clm_0003` supported 0.86
- **能力存在：AI Skill / Agent 指令资产库**（supported）：可以相信项目包含这类能力线索；是否适合你的具体任务仍要试用或安装后验证。 证据：`.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md`, `.claude/skills/review-pr/SKILL.md` 等 Claim：`clm_0001` supported 0.86
- **能力存在：命令行启动或安装流程**（supported）：可以相信项目包含这类能力线索；是否适合你的具体任务仍要试用或安装后验证。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md` Claim：`clm_0002` supported 0.86
- **存在 Quick Start / 安装命令线索**（supported）：可以相信项目文档出现过启动或安装入口；不要因此直接在主力环境运行。 证据：`examples/sampling/README.md` Claim：`clm_0004` supported 0.86

### 现在还不能相信

- **工具权限边界不能在安装前相信。**（unverified）：MCP/tool 类项目通常会触碰文件、网络、浏览器或外部 API，必须真实检查权限和日志。
- **真实输出质量不能在安装前相信。**（unverified）：Prompt Preview 只能展示引导方式，不能证明真实项目中的结果质量。
- **宿主 AI 版本兼容性不能在安装前相信。**（unverified）：Claude、Cursor、Codex、Gemini 等宿主加载规则和版本差异必须在真实环境验证。
- **不会污染现有宿主 AI 行为，不能直接相信。**（inferred）：Skill、plugin、AGENTS/CLAUDE/GEMINI 指令可能改变宿主 AI 的默认行为。 证据：`.claude/hooks/session-init.sh`, `.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md` 等
- **可安全回滚不能默认相信。**（unverified）：除非项目明确提供卸载和恢复说明，否则必须先在隔离环境验证。
- **真实安装后是否与用户当前宿主 AI 版本兼容？**（unverified）：兼容性只能通过实际宿主环境验证。
- **项目输出质量是否满足用户具体任务？**（unverified）：安装前预览只能展示流程和边界，不能替代真实评测。
- **安装命令是否需要网络、权限或全局写入？**（unverified）：这影响企业环境和个人环境的安装风险。 证据：`examples/sampling/README.md`

### 继续会触碰什么

- **命令执行**：包管理器、网络下载、本地插件目录、项目配置或用户主目录。 原因：运行第一条命令就可能产生环境改动；必须先判断是否值得跑。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md`
- **宿主 AI 配置**：Claude/Codex/Cursor/Gemini/OpenCode 等宿主的 plugin、Skill 或规则加载配置。 原因：宿主配置会改变 AI 后续工作方式，可能和用户已有规则冲突。 证据：`.claude/hooks/session-init.sh`, `.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md` 等
- **本地环境或项目文件**：安装结果、插件缓存、项目配置或本地依赖目录。 原因：安装前无法证明写入范围和回滚方式，需要隔离验证。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md`
- **宿主 AI 上下文**：AI Context Pack、Prompt Preview、Skill 路由、风险规则和项目事实。 原因：导入上下文会影响宿主 AI 后续判断，必须避免把未验证项包装成事实。

### 最小安全下一步

- **先跑 Prompt Preview**：用安装前交互式试用判断工作方式是否匹配，不需要授权或改环境。（适用：任何项目都适用，尤其是输出质量未知时。）
- **只在隔离目录或测试账号试装**：避免安装命令污染主力宿主 AI、真实项目或用户主目录。（适用：存在命令执行、插件配置或本地写入线索时。）
- **先备份宿主 AI 配置**：Skill、plugin、规则文件可能改变 Claude/Cursor/Codex 的默认行为。（适用：存在插件 manifest、Skill 或宿主规则入口时。）
- **安装后只验证一个最小任务**：先验证加载、兼容、输出质量和回滚，再决定是否深用。（适用：准备从试用进入真实工作流时。）

### 退出方式

- **保留安装前状态**：记录原始宿主配置和项目状态，后续才能判断是否可恢复。
- **准备移除宿主 plugin / Skill / 规则入口**：如果试装后行为异常，可以把宿主 AI 恢复到试装前状态。
- **记录安装命令和写入路径**：没有明确卸载说明时，至少要知道哪些目录或配置需要手动清理。
- **如果没有回滚路径，不进入主力环境**：不可回滚是继续前阻断项，不应靠信任或运气继续。

## 哪些只能预览

- 解释项目适合谁和能做什么
- 基于项目文档演示典型对话流程
- 帮助用户判断是否值得安装或继续研究

## 哪些必须安装后验证

- 真实安装 Skill、插件或 CLI
- 执行脚本、修改本地文件或访问外部服务
- 验证真实输出质量、性能和兼容性

## 边界与风险判断卡

- **把安装前预览误认为真实运行**：用户可能高估项目已经完成的配置、权限和兼容性验证。 处理方式：明确区分 prompt_preview_can_do 与 runtime_required。 Claim：`clm_0007` inferred 0.45
- **命令执行会修改本地环境**：安装命令可能写入用户主目录、宿主插件目录或项目配置。 处理方式：先在隔离环境或测试账号中运行。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md` Claim：`clm_0008` supported 0.86
- **待确认**：真实安装后是否与用户当前宿主 AI 版本兼容？。原因：兼容性只能通过实际宿主环境验证。
- **待确认**：项目输出质量是否满足用户具体任务？。原因：安装前预览只能展示流程和边界，不能替代真实评测。
- **待确认**：安装命令是否需要网络、权限或全局写入？。原因：这影响企业环境和个人环境的安装风险。

## 开工前工作上下文

### 加载顺序

- 先读取 how_to_use.host_ai_instruction，建立安装前判断资产的边界。
- 读取 claim_graph_summary，确认事实来自 Claim/Evidence Graph，而不是 Human Wiki 叙事。
- 再读取 intended_users、capabilities 和 quick_start_candidates，判断用户是否匹配。
- 需要执行具体任务时，优先查 role_skill_index，再查 evidence_index。
- 遇到真实安装、文件修改、网络访问、性能或兼容性问题时，转入 risk_card 和 boundaries.runtime_required。

### 任务路由

- **AI Skill / Agent 指令资产库**：先基于 role_skill_index / evidence_index 帮用户挑选可用角色、Skill 或工作流。 边界：可做安装前 Prompt 体验。 证据：`.claude/skills/code-review/SKILL.md`, `.claude/skills/python-tests/SKILL.md`, `.claude/skills/review-issue/SKILL.md`, `.claude/skills/review-pr/SKILL.md` 等 Claim：`clm_0001` supported 0.86
- **命令行启动或安装流程**：先说明这是安装后验证能力，再给出安装前检查清单。 边界：必须真实安装或运行后验证。 证据：`.claude/hooks/session-init.sh`, `examples/sampling/README.md`, `fastmcp_slim/fastmcp/contrib/component_manager/README.md` Claim：`clm_0002` supported 0.86

### 上下文规模

- 文件总数：792
- 重要文件覆盖：40/792
- 证据索引条目：74
- 角色 / Skill 条目：7

### 证据不足时的处理

- **missing_evidence**：说明证据不足，要求用户提供目标文件、README 段落或安装后验证记录；不要补全事实。
- **out_of_scope_request**：说明该任务超出当前 AI Context Pack 证据范围，并建议用户先查看 Human Manual 或真实安装后验证。
- **runtime_request**：给出安装前检查清单和命令来源，但不要替用户执行命令或声称已执行。
- **source_conflict**：同时展示冲突来源，标记为待核实，不要强行选择一个版本。

## Prompt Recipes

### 适配判断

- 目标：判断这个项目是否适合用户当前任务。
- 预期输出：适配结论、关键理由、证据引用、安装前可预览内容、必须安装后验证内容、下一步建议。

```text
请基于 fastmcp 的 AI Context Pack，先问我 3 个必要问题，然后判断它是否适合我的任务。回答必须包含：适合谁、能做什么、不能做什么、是否值得安装、证据来自哪里。所有项目事实必须引用 evidence_refs、source_paths 或 claim_id。
```

### 安装前体验

- 目标：让用户在安装前感受核心工作流，同时避免把预览包装成真实能力或营销承诺。
- 预期输出：一段带边界标签的体验剧本、安装后验证清单和谨慎建议；不含真实运行承诺或强营销表述。

```text
请把 fastmcp 当作安装前体验资产，而不是已安装工具或真实运行环境。

请严格输出四段：
1. 先问我 3 个必要问题。
2. 给出一段“体验剧本”：用 [安装前可预览]、[必须安装后验证]、[证据不足] 三种标签展示它可能如何引导工作流。
3. 给出安装后验证清单：列出哪些能力只有真实安装、真实宿主加载、真实项目运行后才能确认。
4. 给出谨慎建议：只能说“值得继续研究/试装”“先补充信息后再判断”或“不建议继续”，不得替项目背书。

硬性边界：
- 不要声称已经安装、运行、执行测试、修改文件或产生真实结果。
- 不要写“自动适配”“确保通过”“完美适配”“强烈建议安装”等承诺性表达。
- 如果描述安装后的工作方式，必须使用“如果安装成功且宿主正确加载 Skill，它可能会……”这种条件句。
- 体验剧本只能写成“示例台词/假设流程”：使用“可能会询问/可能会建议/可能会展示”，不要写“已写入、已生成、已通过、正在运行、正在生成”。
- Prompt Preview 不负责给安装命令；如用户准备试装，只能提示先阅读 Quick Start 和 Risk Card，并在隔离环境验证。
- 所有项目事实必须来自 supported claim、evidence_refs 或 source_paths；inferred/unverified 只能作风险或待确认项。

```

### 角色 / Skill 选择

- 目标：从项目里的角色或 Skill 中挑选最匹配的资产。
- 预期输出：候选角色或 Skill 列表，每项包含适用场景、证据路径、风险边界和是否需要安装后验证。

```text
请读取 role_skill_index，根据我的目标任务推荐 3-5 个最相关的角色或 Skill。每个推荐都要说明适用场景、可能输出、风险边界和 evidence_refs。
```

### 风险预检

- 目标：安装或引入前识别环境、权限、规则冲突和质量风险。
- 预期输出：环境、权限、依赖、许可、宿主冲突、质量风险和未知项的检查清单。

```text
请基于 risk_card、boundaries 和 quick_start_candidates，给我一份安装前风险预检清单。不要替我执行命令，只说明我应该检查什么、为什么检查、失败会有什么影响。
```

### 宿主 AI 开工指令

- 目标：把项目上下文转成一次对话开始前的宿主 AI 指令。
- 预期输出：一段边界明确、证据引用明确、适合复制给宿主 AI 的开工前指令。

```text
请基于 fastmcp 的 AI Context Pack，生成一段我可以粘贴给宿主 AI 的开工前指令。这段指令必须遵守 not_runtime=true，不能声称项目已经安装、运行或产生真实结果。
```

## 角色 / Skill 索引

- 共索引 7 个角色 / Skill / 项目文档条目。

- **reviewing-code**（skill）：Review code for quality, maintainability, and correctness. Use when reviewing pull requests, evaluating code changes, or providing feedback on implementations. Focuses on API design, patterns, and actionable feedback. 激活提示：当用户任务与“reviewing-code”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`.claude/skills/code-review/SKILL.md`
- **testing-python**（skill）：Write and evaluate effective Python tests using pytest. Use when writing tests, reviewing test code, debugging test failures, or improving test coverage. Covers test design, fixtures, parameterization, mocking, and async testing. 激活提示：当用户任务与“testing-python”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`.claude/skills/python-tests/SKILL.md`
- **review-issue**（skill）：Review an incoming external issue and any gated-closed PR behind it and decide whether to assign the contributor or decline. Use when the maintainer says "look at this issue", "review issue N", "should we take this", or asks whether to assign someone. Assigning the author auto-reopens their PR for normal review. This is the entry point for incoming-issue triage — distinct from review-pr, which responds to bot review… 激活提示：当用户任务与“review-issue”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`.claude/skills/review-issue/SKILL.md`
- **review-pr**（skill）：Monitor and respond to automated PR reviews Codex bot . Use when pushing a PR, checking review status, or responding to bot feedback. Handles the full cycle of push - wait for review - evaluate comments - fix - re-push. 激活提示：当用户任务与“review-pr”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`.claude/skills/review-pr/SKILL.md`
- **Code Review Skill**（skill）：Review code for quality, maintainability, and correctness 激活提示：当用户任务与“Code Review Skill”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`examples/skills/sample_skills/code-review/SKILL.md`
- **PDF Processing Skill**（skill）：Extract text from PDFs, fill forms, and merge documents 激活提示：当用户任务与“PDF Processing Skill”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`examples/skills/sample_skills/pdf-processing/SKILL.md`
- **fastmcp-client-cli**（skill）：Query and invoke tools on MCP servers using fastmcp list and fastmcp call. Use when you need to discover what tools a server offers, call tools, or integrate MCP servers into workflows. 激活提示：当用户任务与“fastmcp-client-cli”描述的流程高度相关时，先用它做安装前体验，再决定是否安装。 证据：`skills/fastmcp-client-cli/SKILL.md`

## 证据索引

- 共索引 74 条证据。

- **Community Section**（documentation）：This directory contains community-contributed content and showcases for FastMCP. 证据：`docs/community/README.md`
- **FastMCP 🚀**（documentation）：Made with 💙 by Prefect https://www.prefect.io/ 证据：`README.md`
- **fastmcp-remote**（documentation）：fastmcp-remote is FastMCP's standalone Python stdio bridge for remote MCP servers. It lets MCP clients that launch local stdio processes connect to MCP servers hosted over Streamable HTTP or SSE. 证据：`fastmcp_remote/README.md`
- **FastMCP 🚀**（documentation）：Made with 💙 by Prefect https://www.prefect.io/ 证据：`fastmcp_slim/README.md`
- **QR Code MCP App**（documentation）：An MCP App server that generates QR codes with an interactive viewer UI. Ported from the ext-apps QR server example https://github.com/modelcontextprotocol/ext-apps/tree/main/examples/qr-server to demonstrate FastMCP's MCP Apps support. 证据：`examples/apps/qr_server/README.md`
- **ATProto MCP Server**（documentation）：This example demonstrates a FastMCP server that provides tools and resources for interacting with the AT Protocol Bluesky . 证据：`examples/atproto_mcp/README.md`
- **AuthKit Example**（documentation）：Protects a FastMCP server with WorkOS AuthKit. The server binds the JWT aud claim to its own resource URL automatically — you just paste that same URL into the WorkOS Dashboard as a resource indicator. 证据：`examples/auth/authkit/README.md`
- **AWS Cognito OAuth Example**（documentation）：Demonstrates FastMCP server protection with AWS Cognito OAuth. 证据：`examples/auth/aws_oauth/README.md`
- **Azure Microsoft Entra OAuth Example**（documentation）：Azure Microsoft Entra OAuth Example 证据：`examples/auth/azure_oauth/README.md`
- **Clerk OAuth Example**（documentation）：Demonstrates FastMCP server protection with Clerk OAuth. 证据：`examples/auth/clerk_oauth/README.md`
- **Discord OAuth Example**（documentation）：Demonstrates FastMCP server protection with Discord OAuth. 证据：`examples/auth/discord_oauth/README.md`
- **GitHub OAuth Example**（documentation）：Demonstrates FastMCP server protection with GitHub OAuth. 证据：`examples/auth/github_oauth/README.md`
- **Google OAuth Example**（documentation）：Demonstrates FastMCP server protection with Google OAuth. 证据：`examples/auth/google_oauth/README.md`
- **Keycloak OAuth Example**（documentation）：Demonstrates FastMCP server protection with Keycloak OAuth. 证据：`examples/auth/keycloak_oauth/README.md`
- **Multi-Provider OAuth Example**（documentation）：This example demonstrates mounting multiple OAuth-protected MCP servers in a single application, each with its own OAuth provider. It showcases RFC 8414 path-aware discovery where each server has its own authorization server metadata endpoint. 证据：`examples/auth/mounted/README.md`
- **Oracle OCI IAM Identity Domain OAuth Example**（documentation）：Oracle OCI IAM Identity Domain OAuth Example 证据：`examples/auth/oci_oauth/README.md`
- **PropelAuth OAuth Example**（documentation）：Demonstrates FastMCP server protection with PropelAuth OAuth. 证据：`examples/auth/propelauth_oauth/README.md`
- **Scalekit OAuth Example**（documentation）：Demonstrates FastMCP server protection with Scalekit OAuth. 证据：`examples/auth/scalekit_oauth/README.md`
- **WorkOS OAuth Example**（documentation）：Demonstrates FastMCP server protection with WorkOS OAuth. 证据：`examples/auth/workos_oauth/README.md`
- **Code Mode**（documentation）：CodeMode collapses an entire tool catalog into two meta-tools: search keyword-based discovery and execute run Python scripts that chain tool calls in a sandbox . Instead of burning context tokens on every intermediate result, the LLM writes a script that runs server-side and returns only the final answer. 证据：`examples/code_mode/README.md`
- **FastMCP Configuration Demo**（documentation）：This example demonstrates the recommended way to configure FastMCP servers using fastmcp.json . 证据：`examples/fastmcp_config_demo/README.md`
- **Namespace Activation**（documentation）：Demonstrates session-specific visibility control using tags to organize tools into namespaces that can be activated on demand. 证据：`examples/namespace_activation/README.md`
- **Persistent Session State**（documentation）：This example demonstrates session-scoped state that persists across tool calls within the same MCP session. 证据：`examples/persistent_state/README.md`
- **Dynamic Tools from SQLite**（documentation）：This example demonstrates serving MCP tools from a database. Tools can be added, modified, or disabled by updating the database - no server restart required. 证据：`examples/providers/sqlite/README.md`
- **Sampling Examples**（documentation）：These examples demonstrate FastMCP's sampling API, which allows server tools to request LLM completions from the client. 证据：`examples/sampling/README.md`
- **Search Transforms**（documentation）：When a server exposes many tools, listing them all at once can overwhelm an LLM's context window. Search transforms collapse the full tool catalog behind a search interface — clients see only search tools and call tool , and discover the real tools on demand. 证据：`examples/search/README.md`
- **Skills Provider Example**（documentation）：This example demonstrates how to expose agent skills like Claude Code skills as MCP resources. 证据：`examples/skills/README.md`
- **smart home mcp server**（documentation）：smart home mcp server where .env contains the following: 证据：`examples/smart_home/README.md`
- **FastMCP Tasks Example**（documentation）：Demonstrates background task execution with Docket, including progress tracking, distributed backends, and CLI worker management. 证据：`examples/tasks/README.md`
- **FastMCP Contrib Modules**（documentation）：This directory holds community-contributed modules for FastMCP. These modules extend FastMCP's functionality but are not officially maintained by the core team. 证据：`fastmcp_slim/fastmcp/contrib/README.md`
- **Bulk Tool Caller**（documentation）：This module provides the BulkToolCaller class, which extends the MCPMixin to offer tools for performing multiple tool calls in a single request to a FastMCP server. This can be useful for optimizing interactions with the server by reducing the overhead of individual tool calls. 证据：`fastmcp_slim/fastmcp/contrib/bulk_tool_caller/README.md`
- **Component Manager – Contrib Module for FastMCP**（documentation）：Component Manager – Contrib Module for FastMCP 证据：`fastmcp_slim/fastmcp/contrib/component_manager/README.md`
- **MCP Mixin**（documentation）：from mcp.types import ToolAnnotations 证据：`fastmcp_slim/fastmcp/contrib/mcp_mixin/README.md`
- **OpenAPI Server Implementation New**（documentation）：This directory contains the next-generation FastMCP server implementation for OpenAPI integration, designed to replace the legacy implementation in /server/openapi.py . 证据：`fastmcp_slim/fastmcp/server/providers/openapi/README.md`
- **OpenAPI Utilities**（documentation）：This directory contains the OpenAPI integration utilities for FastMCP. 证据：`fastmcp_slim/fastmcp/utilities/openapi/README.md`
- **FastMCP Testing Demo**（documentation）：A comprehensive example demonstrating FastMCP testing patterns with pytest-asyncio. 证据：`examples/testing_demo/README.md`
- **FastMCP Development Guidelines**（documentation）：Audience : LLM-driven engineering agents and human developers 证据：`CLAUDE.md`
- **Contributing to FastMCP**（documentation）：FastMCP is an actively maintained, high-traffic project. We welcome contributions — but the most impactful way to contribute might not be what you expect. 证据：`CONTRIBUTING.md`
- **Code Review**（skill_instruction）：Code review maintains a healthy codebase while helping contributors succeed. The burden of proof is on the PR to demonstrate it adds value. Your job is to help it get there through actionable feedback. 证据：`.claude/skills/code-review/SKILL.md`
- **Writing Effective Python Tests**（skill_instruction）：Every test should be atomic , self-contained , and test single functionality . A test that tests multiple things is harder to debug and maintain. 证据：`.claude/skills/python-tests/SKILL.md`
- **Triaging contributions under the issue-link gate**（skill_instruction）：Triaging contributions under the issue-link gate 证据：`.claude/skills/review-issue/SKILL.md`
- **PR Review Workflow**（skill_instruction）：This repo has chatgpt-codex-connector bot configured as an automated reviewer. After every push to a PR branch, Codex reviews the diff and either: - Reacts with a thumbs-up on its review body no suggestions — PR is clean - Posts inline comments with suggestions each tagged with a priority badge 证据：`.claude/skills/review-pr/SKILL.md`
- **Code Review Skill**（skill_instruction）：This skill guides you through conducting thorough code reviews. 证据：`examples/skills/sample_skills/code-review/SKILL.md`
- **PDF Processing Skill**（skill_instruction）：This skill helps you work with PDF documents. 证据：`examples/skills/sample_skills/pdf-processing/SKILL.md`
- **FastMCP CLI: List and Call**（skill_instruction）：Use fastmcp list and fastmcp call to interact with any MCP server from the command line. 证据：`skills/fastmcp-client-cli/SKILL.md`
- **License**（source_file）：Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ 证据：`LICENSE`
- **Init**（source_file）：all = "settings" 证据：`examples/atproto_mcp/src/atproto_mcp/__init__.py`
- **Init**（source_file）：all = 证据：`examples/atproto_mcp/src/atproto_mcp/_atproto/__init__.py`
- **Init**（source_file）：all = "settings" 证据：`examples/smart_home/src/smart_home/__init__.py`
- **Init**（source_file）：version = version "fastmcp-remote" ⋮---- version = "0.0.0" all = " version " 证据：`fastmcp_remote/fastmcp_remote/__init__.py`
- **Init**（source_file）：settings = Settings ⋮---- version = version "fastmcp-slim" ⋮---- version = version "fastmcp" ⋮---- def getattr name: str - object all = 证据：`fastmcp_slim/fastmcp/__init__.py`
- **Init**（source_file）：all = ⋮---- def getattr name: str - object 证据：`fastmcp_slim/fastmcp/apps/__init__.py`
- **Init**（source_file）：install app = cyclopts.App 证据：`fastmcp_slim/fastmcp/cli/install/__init__.py`
- **Apply runtime settings from deployment config**（source_file）：logger = get logger "cli.run" TransportType = Literal "stdio", "http", "sse", "streamable-http" LogLevelType = Literal "DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL" WATCHED EXTENSIONS: set str = { def is url path: str - bool ⋮---- url pattern = re.compile r"^https?://" ⋮---- def create client server url: str - Any ⋮---- """Create a FastMCP server from a client URL. Args: url: The URL to connect to Returns: A FastMCP server instance """ ⋮---- client = fastmcp.Client url server = create proxy client ⋮---- def create mcp config server mcp config path: Path - FastMCP None ⋮---- """Create a FastMCP server from a MCPConfig.""" ⋮---- mcp config = json.load src server = create proxy mcp config ⋮-… 证据：`fastmcp_slim/fastmcp/cli/run.py`
- **Init**（source_file）：all = 证据：`fastmcp_slim/fastmcp/client/__init__.py`
- **Init**（source_file）：all = "BearerAuth", "OAuth" 证据：`fastmcp_slim/fastmcp/client/auth/__init__.py`
- **Log with appropriate level and data**（source_file）：logger: Logger = get logger name= name from server logger: Logger = get logger name="fastmcp.client.from server" LogMessage: TypeAlias = LoggingMessageNotificationParams LogHandler: TypeAlias = Callable LogMessage , Awaitable None async def default log handler message: LogMessage - None ⋮---- data = message.data level map = { log fn = level map.get message.level.lower , logger.info msg prefix: str = f"Received {message.level.upper } from server" ⋮---- Log with appropriate level and data ⋮---- def create log callback handler: LogHandler None = None - LoggingFnT ⋮---- handler = default log handler async def log callback params: LoggingMessageNotificationParams - None 证据：`fastmcp_slim/fastmcp/client/logging.py`
- **Init**（source_file）：all = 证据：`fastmcp_slim/fastmcp/client/mixins/__init__.py`
- **Inject trace context into meta for propagation to server**（source_file）：logger = get logger name AUTO PAGINATION MAX PAGES = 250 ResourceTaskResponseUnion = RootModel class ClientResourcesMixin ⋮---- result = await self. await with session monitoring ⋮---- all resources: list mcp.types.Resource = cursor: str None = None seen cursors: set str = set ⋮---- result = await self.list resources mcp cursor=cursor ⋮---- cursor = result.nextCursor ⋮---- """Send a resources/listResourceTemplates request and return the complete MCP protocol result. Args: cursor: Optional pagination cursor from a previous request's nextCursor. Returns: mcp.types.ListResourceTemplatesResult: The complete response object from the protocol, containing the list of resource templates and any add… 证据：`fastmcp_slim/fastmcp/client/mixins/resources.py`
- **--- Call Tool ---**（source_file）：logger = get logger name AUTO PAGINATION MAX PAGES = 250 ToolTaskResponseUnion = RootModel mcp.types.CreateTaskResult mcp.types.CallToolResult class ClientToolsMixin ⋮---- result = await self. await with session monitoring ⋮---- all tools: list mcp.types.Tool = cursor: str None = None seen cursors: set str = set ⋮---- result = await self.list tools mcp cursor=cursor ⋮---- cursor = result.nextCursor ⋮---- --- Call Tool --- ⋮---- """Send a tools/call request and return the complete MCP protocol result. This method returns the raw CallToolResult object, which includes an isError flag and other metadata. It does not raise an exception if the tool call results in an error. Args: name str : The n… 证据：`fastmcp_slim/fastmcp/client/mixins/tools.py`
- 其余 14 条证据见 `AI_CONTEXT_PACK.json` 或 `EVIDENCE_INDEX.json`。

## 宿主 AI 必须遵守的规则

- **把本资产当作开工前上下文，而不是运行环境。**：AI Context Pack 只包含证据化项目理解，不包含目标项目的可执行状态。 证据：`docs/community/README.md`, `README.md`, `fastmcp_remote/README.md`
- **回答用户时区分可预览内容与必须安装后才能验证的内容。**：安装前体验的消费者价值来自降低误装和误判，而不是伪装成真实运行。 证据：`docs/community/README.md`, `README.md`, `fastmcp_remote/README.md`

## 用户开工前应该回答的问题

- 你准备在哪个宿主 AI 或本地环境中使用它？
- 你只是想先体验工作流，还是准备真实安装？
- 你最在意的是安装成本、输出质量、还是和现有规则的冲突？

## 验收标准

- 所有能力声明都能回指到 evidence_refs 中的文件路径。
- AI_CONTEXT_PACK.md 没有把预览包装成真实运行。
- 用户能在 3 分钟内看懂适合谁、能做什么、如何开始和风险边界。

---

## Doramagic Context Augmentation

下面内容用于强化 Repomix/AI Context Pack 主体。Human Manual 只提供阅读骨架；踩坑日志会被转成宿主 AI 必须遵守的工作约束。

## Human Manual 骨架

使用规则：这里只是项目阅读路线和显著性信号，不是事实权威。具体事实仍必须回到 repo evidence / Claim Graph。

宿主 AI 硬性规则：
- 不得把页标题、章节顺序、摘要或 importance 当作项目事实证据。
- 解释 Human Manual 骨架时，必须明确说它只是阅读路线/显著性信号。
- 能力、安装、兼容性、运行状态和风险判断必须引用 repo evidence、source path 或 Claim Graph。

- **Framework Overview & Provider Architecture**：importance `high`
  - source_paths: README.md, fastmcp_slim/fastmcp/server/providers/__init__.py, fastmcp_slim/fastmcp/server/providers/base.py, fastmcp_slim/fastmcp/server/providers/local_provider/local_provider.py, fastmcp_slim/fastmcp/server/providers/aggregate.py
- **Authentication, OAuth Providers & Authorization**：importance `high`
  - source_paths: fastmcp_slim/fastmcp/server/auth/__init__.py, fastmcp_slim/fastmcp/server/auth/auth.py, fastmcp_slim/fastmcp/server/auth/oauth_proxy/proxy.py, fastmcp_slim/fastmcp/server/auth/oauth_proxy/models.py, fastmcp_slim/fastmcp/server/auth/oauth_proxy/consent.py
- **Middleware, Tools, Tasks & Request Lifecycle**：importance `high`
  - source_paths: fastmcp_slim/fastmcp/server/middleware/__init__.py, fastmcp_slim/fastmcp/server/middleware/middleware.py, fastmcp_slim/fastmcp/server/middleware/caching.py, fastmcp_slim/fastmcp/server/middleware/rate_limiting.py, fastmcp_slim/fastmcp/server/middleware/logging.py
- **Transports, Proxy Clients, Schemas, Apps & Transforms**：importance `high`
  - source_paths: fastmcp_slim/fastmcp/client/__init__.py, fastmcp_slim/fastmcp/client/client.py, fastmcp_slim/fastmcp/client/transports/__init__.py, fastmcp_slim/fastmcp/client/transports/base.py, fastmcp_slim/fastmcp/client/transports/stdio.py

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `3b8538e2422a1c43fdb69661c610de7985b785f2`
- inspected_files: `README.md`, `pyproject.toml`, `uv.lock`, `docs/apps/architecture.mdx`, `docs/apps/demos/bar-chart.py`, `docs/apps/demos/contacts.py`, `docs/apps/demos/dashboard.py`, `docs/apps/demos/data-table.py`, `docs/apps/demos/hitchhikers.py`, `docs/apps/demos/pie-chart.py`, `docs/apps/demos/reactive.py`, `docs/apps/demos/team-directory-reactive.py`, `docs/apps/demos/team-directory.py`, `docs/apps/development.mdx`, `docs/apps/examples.mdx`, `docs/apps/fastmcp-app.mdx`, `docs/apps/generative.mdx`, `docs/apps/low-level.mdx`, `docs/apps/overview.mdx`, `docs/apps/prefab.mdx`

宿主 AI 硬性规则：
- 没有 repo_clone_verified=true 时，不得声称已经读过源码。
- 没有 repo_inspection_verified=true 时，不得把 README/docs/package 文件判断写成事实。
- 没有 quick_start_verified=true 时，不得声称 Quick Start 已跑通。

## Doramagic Pitfall Constraints / 踩坑约束

这些规则来自 Doramagic 发现、验证或编译过程中的项目专属坑点。宿主 AI 必须把它们当作工作约束，而不是普通说明文字。

### Constraint 1: 来源证据：RecursionError in json_schema_to_type on self-referential $ref in tool outputSchema

- Trigger: GitHub 社区证据显示该项目存在一个安装相关的待验证问题：RecursionError in json_schema_to_type on self-referential $ref in tool outputSchema
- Why it matters: 可能增加新用户试用和生产接入成本。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4306 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 2: 来源证据：Upgrade checks failing on main branch

- Trigger: GitHub 社区证据显示该项目存在一个安装相关的待验证问题：Upgrade checks failing on main branch
- Why it matters: 可能阻塞安装或首次运行。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4241 | 来源类型 github_issue 暴露的待验证使用条件。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 3: 来源证据：caching middleware broken

- Trigger: GitHub 社区证据显示该项目存在一个安装相关的待验证问题：caching middleware broken
- Why it matters: 可能阻塞安装或首次运行。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4300 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 4: 来源证据：FastMCP StatefulProxyClient.clear() can cause KeyError during proxy session teardown

- Trigger: GitHub 社区证据显示该项目存在一个配置相关的待验证问题：FastMCP StatefulProxyClient.clear() can cause KeyError during proxy session teardown
- Why it matters: 可能阻塞安装或首次运行。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4321 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 5: 来源证据：feat: hook or factory for customizing tool timeout error messages and types

- Trigger: GitHub 社区证据显示该项目存在一个配置相关的待验证问题：feat: hook or factory for customizing tool timeout error messages and types
- Why it matters: 可能增加新用户试用和生产接入成本。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4305 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 6: 来源证据：expand_uri_template and match_uri_template do not round-trip path values containing reserved characters

- Trigger: GitHub 社区证据显示该项目存在一个维护/版本相关的待验证问题：expand_uri_template and match_uri_template do not round-trip path values containing reserved characters
- Why it matters: 可能影响升级、迁移或版本选择。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4326 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 7: 失败模式：security_permissions: No guardrails against destructive tool capabilities (shell exec, file deletion, env access)

- Trigger: Developers should check this security_permissions risk before relying on the project: No guardrails against destructive tool capabilities (shell exec, file deletion, env access)
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: No guardrails against destructive tool capabilities (shell exec, file deletion, env access). Context: Observed when using python
- Why it matters: Developers may expose sensitive permissions or credentials: No guardrails against destructive tool capabilities (shell exec, file deletion, env access)
- Evidence: failure_mode_cluster:github_issue | https://github.com/PrefectHQ/fastmcp/issues/4318 | No guardrails against destructive tool capabilities (shell exec, file deletion, env access)
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 8: 失败模式：security_permissions: on_message middleware not called for unauthenticated requests

- Trigger: Developers should check this security_permissions risk before relying on the project: on_message middleware not called for unauthenticated requests
- Host AI rule: Before packaging this project, run the relevant install/config/quickstart check for: on_message middleware not called for unauthenticated requests. Context: Observed when using python, linux
- Why it matters: Developers may expose sensitive permissions or credentials: on_message middleware not called for unauthenticated requests
- Evidence: failure_mode_cluster:github_issue | https://github.com/PrefectHQ/fastmcp/issues/4309 | on_message middleware not called for unauthenticated requests
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 9: 来源证据：No guardrails against destructive tool capabilities (shell exec, file deletion, env access)

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：No guardrails against destructive tool capabilities (shell exec, file deletion, env access)
- Why it matters: 可能影响升级、迁移或版本选择。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4318 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。

### Constraint 10: 来源证据：RateLimitingMiddleware.get_client_id passing an async callable silently disables rate limiting

- Trigger: GitHub 社区证据显示该项目存在一个安全/权限相关的待验证问题：RateLimitingMiddleware.get_client_id passing an async callable silently disables rate limiting
- Why it matters: 可能影响授权、密钥配置或安全边界。
- Evidence: community_evidence:github | https://github.com/PrefectHQ/fastmcp/issues/4320 | 来源讨论提到 python 相关条件，需在安装/试用前复核。
- Hard boundary: 不要把这个坑点包装成已解决、已验证或可忽略，除非后续验证证据明确证明它已经关闭。