# graphrag - 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 graphrag 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 - **想在安装前理解开源项目价值和边界的用户**：当前证据主要来自项目文档。 证据：`README.md` Claim：`clm_0002` supported 0.86 ## 它能做什么 - **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：`packages/graphrag-input/README.md` Claim：`clm_0001` supported 0.86 ## 怎么开始 - `pip install 'markitdown[pdf]' # required dependency for pdf processing` 证据：`packages/graphrag-input/README.md` Claim：`clm_0003` supported 0.86 ## 继续前判断卡 - **当前建议**：需要管理员/安全审批 - **为什么**：继续前可能涉及密钥、账号、外部服务或敏感上下文，建议先经过管理员或安全审批。 ### 30 秒判断 - **现在怎么做**：需要管理员/安全审批 - **最小安全下一步**：先跑 Prompt Preview；若涉及凭证或企业环境，先审批再试装 - **先别相信**：角色质量和任务匹配不能直接相信。 - **继续会触碰**：角色选择偏差、命令执行、本地环境或项目文件 ### 现在可以相信 - **适合人群线索：想在安装前理解开源项目价值和边界的用户**（supported）：有 supported claim 或项目证据支撑，但仍不等于真实安装效果。 证据：`README.md` Claim：`clm_0002` 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：`packages/graphrag-input/README.md` Claim：`clm_0001` 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：`packages/graphrag-input/README.md` Claim：`clm_0003` supported 0.86 ### 现在还不能相信 - **角色质量和任务匹配不能直接相信。**（unverified）：角色库证明有很多角色，不证明每个角色都适合你的具体任务，也不证明角色能产生高质量结果。 - **不能把角色文案当成真实执行能力。**（unverified）：安装前只能判断角色描述和任务画像是否匹配，不能证明它能在宿主 AI 里完成任务。 - **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。 - **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。 ### 继续会触碰什么 - **角色选择偏差**：用户对任务应该由哪个专家角色处理的判断。 原因：选错角色会让 AI 从错误专业视角回答，浪费时间或误导决策。 - **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：`packages/graphrag-input/README.md` - **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：`packages/graphrag-input/README.md` - **环境变量 / API Key**：项目入口文档明确出现 API key、token、secret 或账号凭证配置。 原因：如果真实安装需要凭证，应先使用测试凭证并经过权限/合规判断。 证据：`docs/config/models.md`, `docs/config/yaml.md`, `docs/get_started.md`, `packages/graphrag-llm/graphrag_llm/README.md` 等 - **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**：先用交互式试用验证任务画像和角色匹配，不要先导入整套角色库。（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。） - **不要使用真实生产凭证**：环境变量/API key 一旦进入宿主或工具链，可能产生账号和合规风险。（适用：出现 API、TOKEN、KEY、SECRET 等环境线索时。） - **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。 - **保留原始角色选择记录**：如果输出偏题，可以回到任务画像阶段重新选择角色，而不是继续沿着错误角色推进。 - **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。 - **准备撤销测试 API key 或 token**：测试凭证泄露或误用时，可以快速止损。 - **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_0004` 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：`packages/graphrag-input/README.md` Claim：`clm_0005` 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 - **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：`packages/graphrag-input/README.md` Claim：`clm_0001` supported 0.86 ### 上下文规模 - 文件总数：619 - 重要文件覆盖：40/619 - 证据索引条目：76 - 角色 / Skill 条目：43 ### 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 graphrag 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 graphrag 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 graphrag 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 43 个角色 / Skill / 项目文档条目。 - **GraphRAG**（project_doc）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 Read the docs https://microsoft.github.io/graphrag 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`README.md` - **Unified Search**（project_doc）：Unified Search Unified demo for GraphRAG search comparisons. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`unified-search-app/README.md` - **GraphRAG Cache**（project_doc）：This package contains a collection of utilities to handle GraphRAG caching implementation. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-cache/README.md` - **GraphRAG Chunking**（project_doc）：This package contains a collection of text chunkers, a core config model, and a factory for acquiring instances. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-chunking/README.md` - **GraphRAG Common**（project_doc）：This package provides utility modules for GraphRAG, including a flexible factory system for dependency injection and service registration, and a comprehensive configuration loading system with Pydantic model support, environment variable substitution, and automatic file discovery. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-common/README.md` - **GraphRAG Inputs**（project_doc）：This package provides input document loading utilities for GraphRAG, supporting multiple file formats including CSV, JSON, JSON Lines, and plain text. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-input/README.md` - **GraphRAG LLM**（project_doc）：This example demonstrates basic usage of the LLM library to interact with Azure OpenAI. It loads environment variables for API configuration, creates a ModelConfig for Azure OpenAI, and sends a simple question to the model. The code handles both streaming and non-streaming responses streaming responses are printed chunk by chunk in real-time, while non-streaming responses are printed all at once . It also shows how… 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-llm/README.md` - **GraphRAG LLM**（project_doc）：View the notebooks notebooks for detailed examples. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-llm/graphrag_llm/README.md` - **API Key and version are optional**（project_doc）：To run the notebooks you need to add a .env file to the notebooks directory with the following information 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-llm/notebooks/README.md` - **GraphRAG Storage**（project_doc）：This package provides a unified storage abstraction layer with support for multiple backends including file system, Azure Blob, Azure Cosmos, and memory storage. It features a factory-based creation system with configuration-driven setup and extensible architecture for implementing custom storage providers. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-storage/README.md` - **GraphRAG Vectors**（project_doc）：This package provides vector store implementations for GraphRAG with support for multiple backends including LanceDB, Azure AI Search, and Azure Cosmos DB. It offers both a convenient configuration-driven API and direct factory access for creating and managing vector stores with flexible index schema definitions. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag-vectors/README.md` - **GraphRAG**（project_doc）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 Read the docs https://microsoft.github.io/graphrag 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`packages/graphrag/README.md` - **Contributing to GraphRAG**（project_doc）：Thank you for your interest in contributing to GraphRAG! We welcome contributions from the community to help improve the project. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`CONTRIBUTING.md` - **Blog Posts**（project_doc）：- :octicons-arrow-right-24: GraphRAG: Unlocking LLM discovery on narrative private data https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/blog_posts.md` - **CLI Reference**（project_doc）：This page documents the command-line interface of the graphrag library. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/cli.md` - **Development Guide**（project_doc）：Name Installation Purpose ------------------- ------------------------------------------------------------ ----------------------------------------------------------------------------------- Python 3.10-3.12 Download https://www.python.org/downloads/ The library is Python-based. uv Instructions https://docs.astral.sh/uv/ uv is used for package management and virtualenv management in Python codebases 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/developing.md` - **Getting Started**（project_doc）：⚠️ GraphRAG can consume a lot of LLM resources! We strongly recommend starting with the tutorial dataset here until you understand how the system works, and consider experimenting with fast/inexpensive models first before committing to a big indexing job. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/get_started.md` - **Welcome to GraphRAG**（project_doc）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index.md` - **Visualizing and Debugging Your Knowledge Graph**（project_doc）：Visualizing and Debugging Your Knowledge Graph 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/visualization_guide.md` - **Configuring GraphRAG Indexing**（project_doc）：To start using GraphRAG, you must generate a configuration file. The init command is the easiest way to get started. It will create a .env and settings.yaml files in the specified directory with the necessary configuration settings. It will also output the default LLM prompts used by GraphRAG. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/config/init.md` - **Language Model Selection and Overriding**（project_doc）：Language Model Selection and Overriding 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/config/models.md` - **Configuring GraphRAG Indexing**（project_doc）：The GraphRAG system is highly configurable. This page provides an overview of the configuration options available for the GraphRAG indexing engine. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/config/overview.md` - **Default Configuration Mode using YAML/JSON**（project_doc）：Default Configuration Mode using YAML/JSON 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/config/yaml.md` - **About**（project_doc）：This document Operation Dulce is an AI-generated science fiction novella, included here for the purposes of integration testing. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/data/operation_dulce/ABOUT.md` - **Operation: Dulce**（project_doc）：The thrumming of monitors cast a stark contrast to the rigid silence enveloping the group. Agent Alex Mercer, unfailingly determined on paper, seemed dwarfed by the enormity of the sterile briefing room where Paranormal Military Squad's elite convened. With dulled eyes, he scanned the projectors outlining their impending odyssey into Operation: Dulce. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/data/operation_dulce/Operation Dulce v2 1 1.md` - **About**（project_doc）：This document Operation Dulce is an AI-generated science fiction novella, included here for the purposes of providing a starting point for notebook experimentation. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/examples_notebooks/inputs/operation dulce/ABOUT.md` - **Operation: Dulce**（project_doc）：The thrumming of monitors cast a stark contrast to the rigid silence enveloping the group. Agent Alex Mercer, unfailingly determined on paper, seemed dwarfed by the enormity of the sterile briefing room where Paranormal Military Squad's elite convened. With dulled eyes, he scanned the projectors outlining their impending odyssey into Operation: Dulce. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/examples_notebooks/inputs/operation dulce/Operation Dulce v2 1 1.md` - **Indexing Architecture**（project_doc）：In order to support the GraphRAG system, the outputs of the indexing engine in the Default Configuration Mode are aligned to a knowledge model we call the GraphRAG Knowledge Model . This model is designed to be an abstraction over the underlying data storage technology, and to provide a common interface for the GraphRAG system to interact with. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/architecture.md` - **Bring Your Own Graph**（project_doc）：Several users have asked if they can bring their own existing graph and have it summarized for query with GraphRAG. There are many possible ways to do this, but here we'll describe a simple method that aligns with the existing GraphRAG workflows quite easily. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/byog.md` - **Indexing Dataflow**（project_doc）：The knowledge model is a specification for data outputs that conform to our data-model definition. You can find these definitions in the python/graphrag/graphrag/model folder within the GraphRAG repository. The following entity types are provided. The fields here represent the fields that are text-embedded by default. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/default_dataflow.md` - **Inputs**（project_doc）：GraphRAG supports several input formats to simplify ingesting your data. The mechanics and features available for input files and text chunking are discussed here. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/inputs.md` - **Indexing Methods**（project_doc）：GraphRAG is a platform for our research into RAG indexing methods that produce optimal context window content for language models. We have a standard indexing pipeline that uses a language model to extract the graph that our memory model is based upon. We may introduce additional indexing methods from time to time. This page documents those options. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/methods.md` - **Outputs**（project_doc）：The default pipeline produces a series of output tables that align with the conceptual knowledge model ../index/default dataflow.md . This page describes the detailed output table schemas. By default we write these tables out as parquet files on disk. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/outputs.md` - **GraphRAG Indexing 🤖**（project_doc）：The GraphRAG indexing package is a data pipeline and transformation suite that is designed to extract meaningful, structured data from unstructured text using LLMs. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/index/overview.md` - **Auto Prompt Tuning ⚙️**（project_doc）：GraphRAG provides the ability to create domain-adapted prompts for the generation of the knowledge graph. This step is optional, though it is highly encouraged to run it as it will yield better results when executing an Index Run. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/prompt_tuning/auto_prompt_tuning.md` - **Manual Prompt Tuning ⚙️**（project_doc）：The GraphRAG indexer, by default, will run with a handful of prompts that are designed to work well in the broad context of knowledge discovery. However, it is quite common to want to tune the prompts to better suit your specific use case. We provide a means for you to do this by allowing you to specify a custom prompt file, which will each use a series of token-replacements internally. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/prompt_tuning/manual_prompt_tuning.md` - **Prompt Tuning ⚙️**（project_doc）：This page provides an overview of the prompt tuning options available for the GraphRAG indexing engine. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/prompt_tuning/overview.md` - **DRIFT Search 🔎**（project_doc）：GraphRAG is a technique that uses large language models LLMs to create knowledge graphs and summaries from unstructured text documents and leverages them to improve retrieval-augmented generation RAG operations on private datasets. It offers comprehensive global overviews of large, private troves of unstructured text documents while also enabling exploration of detailed, localized information. By using LLMs to creat… 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/drift_search.md` - **Global Search 🔎**（project_doc）：Baseline RAG struggles with queries that require aggregation of information across the dataset to compose an answer. Queries such as “What are the top 5 themes in the data?” perform terribly because baseline RAG relies on a vector search of semantically similar text content within the dataset. There is nothing in the query to direct it to the correct information. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/global_search.md` - **Local Search 🔎**（project_doc）：The local search https://github.com/microsoft/graphrag/blob/main/packages/graphrag/graphrag/query/structured search/local search/ method combines structured data from the knowledge graph with unstructured data from the input documents to augment the LLM context with relevant entity information at query time. It is well-suited for answering questions that require an understanding of specific entities mentioned in the… 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/local_search.md` - **API Notebooks**（project_doc）：- API Overview Notebook ../../examples notebooks/api overview.ipynb - Bring-Your-Own Vector Store ../../examples notebooks/custom vector store.ipynb 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/notebooks/overview.md` - **Query Engine 🔎**（project_doc）：The Query Engine is the retrieval module of the GraphRAG library, and operates over completed indexes ../index/overview.md . It is responsible for the following tasks: 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/overview.md` - **Question Generation ❔**（project_doc）：The question generation https://github.com/microsoft/graphrag/blob/main/packages/graphrag/graphrag/query/question gen/ method combines structured data from the knowledge graph with unstructured data from the input documents to generate candidate questions related to specific entities. 激活提示：当用户需要理解项目结构、安装方式或边界时参考。 证据：`docs/query/question_generation.md` ## 证据索引 - 共索引 76 条证据。 - **GraphRAG**（documentation）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 Read the docs https://microsoft.github.io/graphrag 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 证据：`README.md` - **Unified Search**（documentation）：Unified Search Unified demo for GraphRAG search comparisons. 证据：`unified-search-app/README.md` - **GraphRAG Cache**（documentation）：This package contains a collection of utilities to handle GraphRAG caching implementation. 证据：`packages/graphrag-cache/README.md` - **GraphRAG Chunking**（documentation）：This package contains a collection of text chunkers, a core config model, and a factory for acquiring instances. 证据：`packages/graphrag-chunking/README.md` - **GraphRAG Common**（documentation）：This package provides utility modules for GraphRAG, including a flexible factory system for dependency injection and service registration, and a comprehensive configuration loading system with Pydantic model support, environment variable substitution, and automatic file discovery. 证据：`packages/graphrag-common/README.md` - **GraphRAG Inputs**（documentation）：This package provides input document loading utilities for GraphRAG, supporting multiple file formats including CSV, JSON, JSON Lines, and plain text. 证据：`packages/graphrag-input/README.md` - **GraphRAG LLM**（documentation）：This example demonstrates basic usage of the LLM library to interact with Azure OpenAI. It loads environment variables for API configuration, creates a ModelConfig for Azure OpenAI, and sends a simple question to the model. The code handles both streaming and non-streaming responses streaming responses are printed chunk by chunk in real-time, while non-streaming responses are printed all at once . It also shows how to use the gather completion response utility function as a simpler alternative that automatically handles both response types and returns the complete text. 证据：`packages/graphrag-llm/README.md` - **GraphRAG LLM**（documentation）：View the notebooks notebooks for detailed examples. 证据：`packages/graphrag-llm/graphrag_llm/README.md` - **API Key and version are optional**（documentation）：To run the notebooks you need to add a .env file to the notebooks directory with the following information 证据：`packages/graphrag-llm/notebooks/README.md` - **GraphRAG Storage**（documentation）：This package provides a unified storage abstraction layer with support for multiple backends including file system, Azure Blob, Azure Cosmos, and memory storage. It features a factory-based creation system with configuration-driven setup and extensible architecture for implementing custom storage providers. 证据：`packages/graphrag-storage/README.md` - **GraphRAG Vectors**（documentation）：This package provides vector store implementations for GraphRAG with support for multiple backends including LanceDB, Azure AI Search, and Azure Cosmos DB. It offers both a convenient configuration-driven API and direct factory access for creating and managing vector stores with flexible index schema definitions. 证据：`packages/graphrag-vectors/README.md` - **GraphRAG**（documentation）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 Read the docs https://microsoft.github.io/graphrag 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 证据：`packages/graphrag/README.md` - **Contributing to GraphRAG**（documentation）：Thank you for your interest in contributing to GraphRAG! We welcome contributions from the community to help improve the project. 证据：`CONTRIBUTING.md` - **License**（source_file）：Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files the "Software" , to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: 证据：`LICENSE` - **Blog Posts**（documentation）：- :octicons-arrow-right-24: GraphRAG: Unlocking LLM discovery on narrative private data https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 证据：`docs/blog_posts.md` - **CLI Reference**（documentation）：This page documents the command-line interface of the graphrag library. 证据：`docs/cli.md` - **Development Guide**（documentation）：Name Installation Purpose ------------------- ------------------------------------------------------------ ----------------------------------------------------------------------------------- Python 3.10-3.12 Download https://www.python.org/downloads/ The library is Python-based. uv Instructions https://docs.astral.sh/uv/ uv is used for package management and virtualenv management in Python codebases 证据：`docs/developing.md` - **Getting Started**（documentation）：⚠️ GraphRAG can consume a lot of LLM resources! We strongly recommend starting with the tutorial dataset here until you understand how the system works, and consider experimenting with fast/inexpensive models first before committing to a big indexing job. 证据：`docs/get_started.md` - **Welcome to GraphRAG**（documentation）：👉 Microsoft Research Blog Post https://www.microsoft.com/en-us/research/blog/graphrag-unlocking-llm-discovery-on-narrative-private-data/ 👉 GraphRAG Arxiv https://arxiv.org/pdf/2404.16130 证据：`docs/index.md` - **Visualizing and Debugging Your Knowledge Graph**（documentation）：Visualizing and Debugging Your Knowledge Graph 证据：`docs/visualization_guide.md` - **Configuring GraphRAG Indexing**（documentation）：To start using GraphRAG, you must generate a configuration file. The init command is the easiest way to get started. It will create a .env and settings.yaml files in the specified directory with the necessary configuration settings. It will also output the default LLM prompts used by GraphRAG. 证据：`docs/config/init.md` - **Language Model Selection and Overriding**（documentation）：Language Model Selection and Overriding 证据：`docs/config/models.md` - **Configuring GraphRAG Indexing**（documentation）：The GraphRAG system is highly configurable. This page provides an overview of the configuration options available for the GraphRAG indexing engine. 证据：`docs/config/overview.md` - **Default Configuration Mode using YAML/JSON**（documentation）：Default Configuration Mode using YAML/JSON 证据：`docs/config/yaml.md` - **About**（documentation）：This document Operation Dulce is an AI-generated science fiction novella, included here for the purposes of integration testing. 证据：`docs/data/operation_dulce/ABOUT.md` - **Operation: Dulce**（documentation）：The thrumming of monitors cast a stark contrast to the rigid silence enveloping the group. Agent Alex Mercer, unfailingly determined on paper, seemed dwarfed by the enormity of the sterile briefing room where Paranormal Military Squad's elite convened. With dulled eyes, he scanned the projectors outlining their impending odyssey into Operation: Dulce. 证据：`docs/data/operation_dulce/Operation Dulce v2 1 1.md` - **About**（documentation）：This document Operation Dulce is an AI-generated science fiction novella, included here for the purposes of providing a starting point for notebook experimentation. 证据：`docs/examples_notebooks/inputs/operation dulce/ABOUT.md` - **Operation: Dulce**（documentation）：The thrumming of monitors cast a stark contrast to the rigid silence enveloping the group. Agent Alex Mercer, unfailingly determined on paper, seemed dwarfed by the enormity of the sterile briefing room where Paranormal Military Squad's elite convened. With dulled eyes, he scanned the projectors outlining their impending odyssey into Operation: Dulce. 证据：`docs/examples_notebooks/inputs/operation dulce/Operation Dulce v2 1 1.md` - **Indexing Architecture**（documentation）：In order to support the GraphRAG system, the outputs of the indexing engine in the Default Configuration Mode are aligned to a knowledge model we call the GraphRAG Knowledge Model . This model is designed to be an abstraction over the underlying data storage technology, and to provide a common interface for the GraphRAG system to interact with. 证据：`docs/index/architecture.md` - **Bring Your Own Graph**（documentation）：Several users have asked if they can bring their own existing graph and have it summarized for query with GraphRAG. There are many possible ways to do this, but here we'll describe a simple method that aligns with the existing GraphRAG workflows quite easily. 证据：`docs/index/byog.md` - **Indexing Dataflow**（documentation）：The knowledge model is a specification for data outputs that conform to our data-model definition. You can find these definitions in the python/graphrag/graphrag/model folder within the GraphRAG repository. The following entity types are provided. The fields here represent the fields that are text-embedded by default. 证据：`docs/index/default_dataflow.md` - **Inputs**（documentation）：GraphRAG supports several input formats to simplify ingesting your data. The mechanics and features available for input files and text chunking are discussed here. 证据：`docs/index/inputs.md` - **Indexing Methods**（documentation）：GraphRAG is a platform for our research into RAG indexing methods that produce optimal context window content for language models. We have a standard indexing pipeline that uses a language model to extract the graph that our memory model is based upon. We may introduce additional indexing methods from time to time. This page documents those options. 证据：`docs/index/methods.md` - **Outputs**（documentation）：The default pipeline produces a series of output tables that align with the conceptual knowledge model ../index/default dataflow.md . This page describes the detailed output table schemas. By default we write these tables out as parquet files on disk. 证据：`docs/index/outputs.md` - **GraphRAG Indexing 🤖**（documentation）：The GraphRAG indexing package is a data pipeline and transformation suite that is designed to extract meaningful, structured data from unstructured text using LLMs. 证据：`docs/index/overview.md` - **Auto Prompt Tuning ⚙️**（documentation）：GraphRAG provides the ability to create domain-adapted prompts for the generation of the knowledge graph. This step is optional, though it is highly encouraged to run it as it will yield better results when executing an Index Run. 证据：`docs/prompt_tuning/auto_prompt_tuning.md` - **Manual Prompt Tuning ⚙️**（documentation）：The GraphRAG indexer, by default, will run with a handful of prompts that are designed to work well in the broad context of knowledge discovery. However, it is quite common to want to tune the prompts to better suit your specific use case. We provide a means for you to do this by allowing you to specify a custom prompt file, which will each use a series of token-replacements internally. 证据：`docs/prompt_tuning/manual_prompt_tuning.md` - **Prompt Tuning ⚙️**（documentation）：This page provides an overview of the prompt tuning options available for the GraphRAG indexing engine. 证据：`docs/prompt_tuning/overview.md` - **DRIFT Search 🔎**（documentation）：GraphRAG is a technique that uses large language models LLMs to create knowledge graphs and summaries from unstructured text documents and leverages them to improve retrieval-augmented generation RAG operations on private datasets. It offers comprehensive global overviews of large, private troves of unstructured text documents while also enabling exploration of detailed, localized information. By using LLMs to create comprehensive knowledge graphs that connect and describe entities and relationships contained in those documents, GraphRAG leverages semantic structuring of the data to generate responses to a wide variety of complex user queries. 证据：`docs/query/drift_search.md` - **Global Search 🔎**（documentation）：Baseline RAG struggles with queries that require aggregation of information across the dataset to compose an answer. Queries such as “What are the top 5 themes in the data?” perform terribly because baseline RAG relies on a vector search of semantically similar text content within the dataset. There is nothing in the query to direct it to the correct information. 证据：`docs/query/global_search.md` - **Local Search 🔎**（documentation）：The local search https://github.com/microsoft/graphrag/blob/main/packages/graphrag/graphrag/query/structured search/local search/ method combines structured data from the knowledge graph with unstructured data from the input documents to augment the LLM context with relevant entity information at query time. It is well-suited for answering questions that require an understanding of specific entities mentioned in the input documents e.g., “What are the healing properties of chamomile?” . 证据：`docs/query/local_search.md` - **API Notebooks**（documentation）：- API Overview Notebook ../../examples notebooks/api overview.ipynb - Bring-Your-Own Vector Store ../../examples notebooks/custom vector store.ipynb 证据：`docs/query/notebooks/overview.md` - **Query Engine 🔎**（documentation）：The Query Engine is the retrieval module of the GraphRAG library, and operates over completed indexes ../index/overview.md . It is responsible for the following tasks: 证据：`docs/query/overview.md` - **Question Generation ❔**（documentation）：The question generation https://github.com/microsoft/graphrag/blob/main/packages/graphrag/graphrag/query/question gen/ method combines structured data from the knowledge graph with unstructured data from the input documents to generate candidate questions related to specific entities. 证据：`docs/query/question_generation.md` - **Init**（source_file）：all = 证据：`packages/graphrag-cache/graphrag_cache/__init__.py` - **Cache Factory**（source_file）：class CacheFactory Factory "Cache" cache factory = CacheFactory ⋮---- config = config or CacheConfig config model = config.model dump cache strategy = config.type ⋮---- storage = create storage config.storage ⋮---- msg = f"CacheConfig.type '{cache strategy}' is not registered in the CacheFactory. Registered types: {', '.join cache factory.keys }." 证据：`packages/graphrag-cache/graphrag_cache/cache_factory.py` - **Chunker Factory**（source_file）：class ChunkerFactory Factory Chunker chunker factory = ChunkerFactory ⋮---- config model = config.model dump ⋮---- chunker strategy = config.type ⋮---- msg = f"ChunkingConfig.strategy '{chunker strategy}' is not registered in the ChunkerFactory. Registered types: {', '.join chunker factory.keys }." 证据：`packages/graphrag-chunking/graphrag_chunking/chunker_factory.py` - **Init**（source_file）：all = "ConfigParsingError", "load config" 证据：`packages/graphrag-common/graphrag_common/config/__init__.py` - **Init**（source_file）：all = "Factory", "ServiceScope" 证据：`packages/graphrag-common/graphrag_common/factory/__init__.py` - **Delete entries with value None**（source_file）：T = TypeVar "T", covariant=True ServiceScope = Literal "singleton", "transient" ⋮---- @dataclass class ServiceDescriptor Generic T ⋮---- scope: ServiceScope initializer: Callable ..., T class Factory ABC, Generic T ⋮---- instance: ClassVar "Factory None" = None def new cls, args: Any, kwargs: Any - "Factory T " def init self def contains self, strategy: str - bool def keys self - list str ⋮---- def create self, strategy: str, init args: dict str, Any None = None - T ⋮---- msg = f"Strategy '{strategy}' is not registered. Registered strategies are: {', '.join list self. service initializers.keys }" ⋮---- Delete entries with value None That way services can have default values init args = {k:… 证据：`packages/graphrag-common/graphrag_common/factory/factory.py` - **Init**（source_file）：all = 证据：`packages/graphrag-common/graphrag_common/hasher/__init__.py` - **Init**（source_file）：all = 证据：`packages/graphrag-input/graphrag_input/__init__.py` - **Input Reader Factory**（source_file）：logger = logging.getLogger name class InputReaderFactory Factory InputReader input reader factory = InputReaderFactory ⋮---- def create input reader config: InputConfig, storage: Storage - InputReader ⋮---- config model = config.model dump input strategy = config.type ⋮---- msg = f"InputConfig.type '{input strategy}' is not registered in the InputReaderFactory. Registered types: {', '.join input reader factory.keys }." 证据：`packages/graphrag-input/graphrag_input/input_reader_factory.py` - **Init**（source_file）：all = 证据：`packages/graphrag-llm/graphrag_llm/cache/__init__.py` - **Init**（source_file）：all = 证据：`packages/graphrag-llm/graphrag_llm/completion/__init__.py` - **Completion Factory**（source_file）：class CompletionFactory Factory "LLMCompletion" completion factory = CompletionFactory ⋮---- cache key creator = cache key creator or create cache key model id = f"{model config.model provider}/{model config.model}" strategy = model config.type extra: dict str, Any = model config.model extra or {} ⋮---- msg = f"ModelConfig.type '{strategy}' is not registered in the CompletionFactory. Registered strategies: {', '.join completion factory.keys }" ⋮---- tokenizer = tokenizer or create tokenizer TokenizerConfig model id=model id rate limiter: RateLimiter None = None ⋮---- rate limiter = create rate limiter rate limit config=model config.rate limit retrier: Retry None = None ⋮---- retrier = creat… 证据：`packages/graphrag-llm/graphrag_llm/completion/completion_factory.py` - **Init**（source_file）：all = 证据：`packages/graphrag-llm/graphrag_llm/config/__init__.py` - **Init**（source_file）：all = 证据：`packages/graphrag-llm/graphrag_llm/embedding/__init__.py` - **Embedding Factory**（source_file）：class EmbeddingFactory Factory "LLMEmbedding" embedding factory = EmbeddingFactory ⋮---- cache key creator = cache key creator or create cache key model id = f"{model config.model provider}/{model config.model}" strategy = model config.type extra: dict str, Any = model config.model extra or {} ⋮---- msg = f"ModelConfig.type '{strategy}' is not registered in the CompletionFactory. Registered strategies: {', '.join embedding factory.keys }" ⋮---- tokenizer = tokenizer or create tokenizer TokenizerConfig model id=model id rate limiter: RateLimiter None = None ⋮---- rate limiter = create rate limiter rate limit config=model config.rate limit retrier: Retry None = None ⋮---- retrier = create ret… 证据：`packages/graphrag-llm/graphrag_llm/embedding/embedding_factory.py` - **Init**（source_file）：all = 证据：`packages/graphrag-llm/graphrag_llm/metrics/__init__.py` - 其余 16 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：`README.md`, `unified-search-app/README.md`, `packages/graphrag-cache/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：`README.md`, `unified-search-app/README.md`, `packages/graphrag-cache/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。 - **GraphRAG Overview and System Architecture**：importance `high` - source_paths: README.md, pyproject.toml, packages/graphrag/README.md, packages/graphrag/graphrag/__init__.py, unified-search-app/README.md - **Indexing Pipeline, Workflows, and Data Flow**：importance `high` - source_paths: packages/graphrag/graphrag/index/run/run_pipeline.py, packages/graphrag/graphrag/index/workflows/factory.py, packages/graphrag/graphrag/index/workflows/load_input_documents.py, packages/graphrag/graphrag/index/workflows/extract_graph.py, packages/graphrag/graphrag/index/workflows/extract_graph_nlp.py - **Query Engine and Search Methods**：importance `high` - source_paths: packages/graphrag/graphrag/api/query.py, packages/graphrag/graphrag/query/factory.py, packages/graphrag/graphrag/query/structured_search/local_search/search.py, packages/graphrag/graphrag/query/structured_search/global_search/search.py, packages/graphrag/graphrag/query/structured_search/drift_search/search.py - **Configuration, LLM Providers, Storage, and Extensibility**：importance `high` - source_paths: packages/graphrag/graphrag/config/models/graph_rag_config.py, packages/graphrag/graphrag/config/load_config.py, packages/graphrag/graphrag/config/models/local_search_config.py, packages/graphrag/graphrag/config/models/global_search_config.py, packages/graphrag/graphrag/config/models/drift_search_config.py ## Repo Inspection Evidence / 源码检查证据 - repo_clone_verified: true - repo_inspection_verified: true - repo_commit: `6d02c2355c3fed4c49007572fbe951d73258a37f` - inspected_files: `README.md`, `pyproject.toml`, `uv.lock`, `docs/blog_posts.md`, `docs/cli.md`, `docs/config/init.md`, `docs/config/models.md`, `docs/config/overview.md`, `docs/config/yaml.md`, `docs/data/operation_dulce/ABOUT.md`, `docs/data/operation_dulce/Operation Dulce v2 1 1.md`, `docs/developing.md`, `docs/examples_notebooks/inputs/operation dulce/ABOUT.md`, `docs/examples_notebooks/inputs/operation dulce/Operation Dulce v2 1 1.md`, `docs/get_started.md`, `docs/index/architecture.md`, `docs/index/byog.md`, `docs/index/default_dataflow.md`, `docs/index/inputs.md`, `docs/index/methods.md` 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: 能力判断依赖假设 - Trigger: README/documentation is current enough for a first validation pass. - Host AI rule: 将假设转成下游验证清单。 - Why it matters: 假设不成立时，用户拿不到承诺的能力。 - Evidence: capability.assumptions | https://github.com/microsoft/graphrag | README/documentation is current enough for a first validation pass. - 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: 维护活跃度未知 - Trigger: 未记录 last_activity_observed。 - Host AI rule: 补 GitHub 最近 commit、release、issue/PR 响应信号。 - Why it matters: 新项目、停更项目和活跃项目会被混在一起，推荐信任度下降。 - Evidence: evidence.maintainer_signals | https://github.com/microsoft/graphrag | last_activity_observed missing - 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。 - Trigger: no_demo - Evidence: downstream_validation.risk_items | https://github.com/microsoft/graphrag | no_demo; severity=medium - 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: 存在评分风险 - Trigger: no_demo - Why it matters: 风险会影响是否适合普通用户安装。 - Evidence: risks.scoring_risks | https://github.com/microsoft/graphrag | no_demo; severity=medium - 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: issue/PR 响应质量未知 - Trigger: issue_or_pr_quality=unknown。 - Host AI rule: 抽样最近 issue/PR，判断是否长期无人处理。 - Why it matters: 用户无法判断遇到问题后是否有人维护。 - Evidence: evidence.maintainer_signals | https://github.com/microsoft/graphrag | issue_or_pr_quality=unknown - 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: 发布节奏不明确 - Trigger: release_recency=unknown。 - Host AI rule: 确认最近 release/tag 和 README 安装命令是否一致。 - Why it matters: 安装命令和文档可能落后于代码，用户踩坑概率升高。 - Evidence: evidence.maintainer_signals | https://github.com/microsoft/graphrag | release_recency=unknown - 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。