# llmstack - 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 llmstack 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

- **AI 研究者或研究型 Agent 构建者**：README 明确围绕研究、实验或论文工作流展开。 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：`README.md` Claim：`clm_0001` supported 0.86

## 怎么开始

- `pip install llmstack` 证据：`README.md` Claim：`clm_0003` supported 0.86

## 继续前判断卡

- **当前建议**：仅建议沙盒试装
- **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**：仅建议沙盒试装
- **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**：真实输出质量不能在安装前相信。
- **继续会触碰**：命令执行、本地环境或项目文件、宿主 AI 上下文

### 现在可以相信

- **适合人群线索：AI 研究者或研究型 Agent 构建者**（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：`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：`README.md` Claim：`clm_0003` supported 0.86

### 现在还不能相信

- **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。
- **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：`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：`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：`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。）
- **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。
- **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_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：`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：`README.md` Claim：`clm_0001` supported 0.86

### 上下文规模

- 文件总数：1156
- 重要文件覆盖：40/1156
- 证据索引条目：57
- 角色 / Skill 条目：15

### 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 llmstack 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 llmstack 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 llmstack 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 15 个角色 / Skill / 项目文档条目。

- **Overview**（project_doc）：LLMStack is a no-code platform for building generative AI agents, workflows and chatbots, connecting them to your data and business processes. Quickstart Documentation Promptly 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`
- **Website**（project_doc）：This website is built using Docusaurus 2 https://docusaurus.io/ , a modern static website generator. 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：`web/README.md`
- **Getting Started with Create React App**（project_doc）：Getting Started with Create React App 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：`llmstack/client/README.md`
- **Index**（project_doc）： 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：`web/docs/index.md`
- **Contributor Covenant Code of Conduct**（project_doc）：Contributor Covenant Code of Conduct 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：`CODE_OF_CONDUCT.md`
- **License**（project_doc）：Licensed under Attribution-ShareAlike 4.0 International https://creativecommons.org/licenses/by-sa/4.0/ 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：`llmstack/client/src/components/apps/lexical/images/emoji/LICENSE.md`
- **License**（project_doc）：Bootstrap Icons https://icons.getbootstrap.com 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：`llmstack/client/src/components/apps/lexical/images/icons/LICENSE.md`
- **License Select2**（project_doc）：Copyright c 2012-2017 Kevin Brown, Igor Vaynberg, and Select2 contributors 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：`llmstack/static/admin/css/vendor/select2/LICENSE-SELECT2.md`
- **License**（project_doc）：Copyright c 2012-2017 Kevin Brown, Igor Vaynberg, and Select2 contributors 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：`llmstack/static/admin/js/vendor/select2/LICENSE.md`
- **Introducing LLMStack**（project_doc）：LLMStack is a low-code AI platform that allows you to build powerful AI applications using pre-trained language models. 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：`web/blog/2023-08-08-introducing-llmstack.md`
- **What is LocalAI?**（project_doc）：Build AI Apps with Open Source LLMs like Llama2 using LocalAI and LLMStack. 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：`web/blog/2023-08-17-run-os-llms-with-localai.md`
- **What is Retrieval Augmented Generation?**（project_doc）：Retrieval Augmented Generation RAG is a simple yet powerful approach that can be used to improve the performance of LLMs on a wide range of 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：`web/blog/2023-09-13-retrieval-augmented-generation.md`
- **2023 09 22 Longer Context With Anthropic**（project_doc）：Build Apps that leverage Anthropic's Long Context Language Model, Claude-2. 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：`web/blog/2023-09-22-longer-context-with-anthropic.md`
- **2023 12 14 Build Ai Apps With Gemini Ai**（project_doc）：Use Gemini Pro to build AI apps that can generate text, answer questions, summarize text, translate text, classify text and more. 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：`web/blog/2023-12-14-build-ai-apps-with-gemini-ai.md`
- **2024 01 18 Realtime Avatars With Heygen**（project_doc）：Introducing HeyGen support in LLMStack / Promptly 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：`web/blog/2024-01-18-realtime-avatars-with-heygen.md`

## 证据索引

- 共索引 57 条证据。

- **Overview**（documentation）：LLMStack is a no-code platform for building generative AI agents, workflows and chatbots, connecting them to your data and business processes. Quickstart Documentation Promptly 证据：`README.md`
- **Website**（documentation）：This website is built using Docusaurus 2 https://docusaurus.io/ , a modern static website generator. 证据：`web/README.md`
- **Getting Started with Create React App**（documentation）：Getting Started with Create React App 证据：`llmstack/client/README.md`
- **Package**（package_manifest）：{ "name": "web", "version": "0.0.0", "private": true, "scripts": { "docusaurus": "docusaurus", "start": "docusaurus start", "build": "docusaurus build", "swizzle": "docusaurus swizzle", "deploy": "docusaurus deploy", "clear": "docusaurus clear", "serve": "docusaurus serve", "write-translations": "docusaurus write-translations", "write-heading-ids": "docusaurus write-heading-ids" }, "dependencies": { "@docusaurus/core": "^3.0.1", "@docusaurus/plugin-google-gtag": "^3.0.1", "@docusaurus/plugin-sitemap": "^3.0.1", "@docusaurus/preset-classic": "^3.0.1", "@easyops-cn/docusaurus-search-local": "^0.38.1", "@mdx-js/react": "^3.0.0", "clsx": "^2.0.0", "prism-react-renderer": "^2.1.0", "react": "^18… 证据：`web/package.json`
- **Package**（package_manifest）：{ "name": "llmstack", "version": "0.2.6", "private": true, "proxy": "http://localhost:9000", "homepage": "https://llmstack.trypromptly.com", "dependencies": { "@fontsource/lato": "^5.0.20", "@glideapps/glide-data-grid": "^6.0.3", "@glideapps/glide-data-grid-cells": "^6.0.3", "@heroicons/react": "^2.1.5", "@jsonforms/material-renderers": "^3.1.0", "@jsonforms/react": "^3.0.0", "@lexical/react": "^0.9.2", "@mui/icons-material": "^5.11.16", "@mui/lab": "^5.0.0-alpha.124", "@mui/x-data-grid": "^6.17.0", "@mui/x-data-grid-generator": "^6.16.2", "@mui/x-date-pickers": "^6.16.1", "@novnc/novnc": "^1.4.0", "@react-pdf-viewer/core": "^3.12.0", "@react-pdf-viewer/default-layout": "^3.12.0", "@react-p… 证据：`llmstack/client/package.json`
- **License**（source_file）：Acceptance By using the software, you agree to all of the terms and conditions below. 证据：`LICENSE`
- **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：`llmstack/client/src/vendor/wavtools/LICENSE`
- **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：`llmstack/static/admin/img/LICENSE`
- **Index**（documentation）：--- id: introduction title: Introduction sidebar label: Introduction sidebar position: 0 --- 证据：`web/docs/index.md`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/app_store/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/apps/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/assets/urls.py`
- **Dumpvectordata**（source_file）：WEAVIATE URL = settings.WEAVIATE URL client = weaviate.Client WEAVIATE URL class WeaviateClassObject BaseModel ⋮---- className: str = Field ..., alias="class" creationTimeUnix: int = Field ..., alias="creationTimeUnix" id: str = Field ..., alias="id" lastUpdateTimeUnix: int = Field ..., alias="lastUpdateTimeUnix" properties: Dict str, Any vectorWeights: Optional Dict str, Any = None vector: Optional List Any = None class WeaviateClass BaseModel ⋮---- description: Optional str = None invertedIndexConfig: Optional Dict = None moduleConfig: Optional Dict = None properties: Optional List Any = None replicationConfig: Optional Dict = None shardingConfig: Optional Dict = None vectorIndexConfig: O… 证据：`llmstack/base/management/commands/dumpvectordata.py`
- **Urls**（source_file）：index module = import settings.INDEX VIEW MODULE, fromlist= "" index view = getattr index module, "index" logout = LLMStackLogoutView.as view urlpatterns = ⋮---- loginViewCls = getattr index module, "PromptlyLoginView", LoginView 证据：`llmstack/base/urls.py`
- **Index**（source_file）：const setInitialRecoilState = 证据：`llmstack/client/src/index.js`
- **Urls**（source_file）：logger = logging.getLogger name class CustomOAuth2LoginView OAuth2LoginView ⋮---- def login self, request, args, kwargs ⋮---- provider = self.adapter.get provider app = provider.app client = self.get client request, app action = request.GET.get "action", AuthAction.AUTHENTICATE auth url = self.adapter.authorize url auth params = provider.get auth params request, action pkce params = provider.get pkce params code verifier = pkce params.pop "code verifier", None ⋮---- class CustomOAuth2CallbackView OAuth2CallbackView ⋮---- def dispatch self, request, args, kwargs ⋮---- auth error = request.GET.get "error", None ⋮---- error = AuthError.CANCELLED ⋮---- error = AuthError.UNKNOWN ⋮---- client = s… 证据：`llmstack/connections/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/data/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/jobs/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/organizations/urls.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/processors/urls.py`
- **Asgi**（source_file）：BASE DIR = dirname dirname abspath file ⋮---- django asgi app = get asgi application application = ProtocolTypeRouter 证据：`llmstack/server/asgi.py`
- **Settings**（source_file）：BASE DIR = Path file .resolve .parent.parent SECRET KEY = os.getenv DEBUG = os.getenv "DEBUG", "False" == "True" LLMSTACK PORT = os.getenv "LLMSTACK PORT", 3000 LLMSTACK HOST = os.getenv "LLMSTACK HOST", "localhost" ALLOWED HOSTS = os.getenv RUNNER HOST = os.getenv "RUNNER HOST", "runner" RUNNER PORT = os.getenv "RUNNER PORT", 50051 RUNNER PLAYWRIGHT PORT = os.getenv "RUNNER PLAYWRIGHT PORT", 50053 PLAYWRIGHT URL = f"ws://{RUNNER HOST}:{RUNNER PLAYWRIGHT PORT}" if RUNNER HOST and RUNNER PLAYWRIGHT PORT else "" CSRF TRUSTED ORIGINS = os.getenv STATIC ROOT = os.path.join BASE DIR, "static" CIPHER KEY SALT = os.getenv "CIPHER KEY SALT", None ADMIN ENABLED = os.getenv "ADMIN ENABLED", True DATA… 证据：`llmstack/server/settings.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/server/urls.py`
- **Wsgi**（source_file）：application = get wsgi application 证据：`llmstack/server/wsgi.py`
- **Urls**（source_file）：urlpatterns = 证据：`llmstack/sheets/urls.py`
- **Index**（source_file）：function initJSONForm element function isDraggingElement element function initializeAllForNode parentElement function init 证据：`llmstack/static/django_jsonform/index.js`
- **Index**（source_file）：function Feature export default function HomepageFeatures 证据：`web/src/components/HomepageFeatures/index.js`
- **Index**（source_file）：function HomepageHeader export default function Home 证据：`web/src/pages/index.js`
- **Contributor Covenant Code of Conduct**（documentation）：Contributor Covenant Code of Conduct 证据：`CODE_OF_CONDUCT.md`
- **License**（documentation）：Licensed under Attribution-ShareAlike 4.0 International https://creativecommons.org/licenses/by-sa/4.0/ 证据：`llmstack/client/src/components/apps/lexical/images/emoji/LICENSE.md`
- **License**（documentation）：Bootstrap Icons https://icons.getbootstrap.com 证据：`llmstack/client/src/components/apps/lexical/images/icons/LICENSE.md`
- **License Select2**（documentation）：Copyright c 2012-2017 Kevin Brown, Igor Vaynberg, and Select2 contributors 证据：`llmstack/static/admin/css/vendor/select2/LICENSE-SELECT2.md`
- **License**（documentation）：Copyright c 2012-2017 Kevin Brown, Igor Vaynberg, and Select2 contributors 证据：`llmstack/static/admin/js/vendor/select2/LICENSE.md`
- **Introducing LLMStack**（documentation）：import ReactPlayer from "react-player"; 证据：`web/blog/2023-08-08-introducing-llmstack.md`
- **What is LocalAI?**（documentation）：Now build AI Apps using Open Source LLMs like Llama2 on LLMStack using LocalAI https://localai.io 证据：`web/blog/2023-08-17-run-os-llms-with-localai.md`
- **What is Retrieval Augmented Generation?**（documentation）：What is Retrieval Augmented Generation? 证据：`web/blog/2023-09-13-retrieval-augmented-generation.md`
- **2023 09 22 Longer Context With Anthropic**（documentation）：If you have built content generation apps like blog post generators or any other app that generates or processes text, you know that the biggest challenge is handling large amounts of text. The current state of the art models are limited by the context windows making it hard to generate or handle large text. Anthropic's launch of Claude-2 https://www.anthropic.com/index/claude-2 , a 100K context window language model, changes that. At Promptly, we are excited to announce that you can now build apps that leverage Claude-2 via our LLMStack https://llmstack.ai framework and Promptly https://trypromptly.com platform. 证据：`web/blog/2023-09-22-longer-context-with-anthropic.md`
- **2023 12 14 Build Ai Apps With Gemini Ai**（documentation）：Google announced https://blog.google/technology/ai/google-gemini-ai their much awaited set of Generative AI APIs, Gemini last week. They have made available their Gemini Pro available via API with the much powerful Gemini Ultra coming later next year. At Promptly, we are excited to announce that we are making Gemini Pro available to our users via our LLMStack https://llmstack.ai framework and Promptly https://trypromplty.com platform. 证据：`web/blog/2023-12-14-build-ai-apps-with-gemini-ai.md`
- **2024 01 18 Realtime Avatars With Heygen**（documentation）：import realtimeAvatarChatAppImage from '@site/static/img/realtime-avatar-chat-screenshot.png'; 证据：`web/blog/2024-01-18-realtime-avatars-with-heygen.md`
- **Manifest**（structured_config）：{ "short name": "Promptly", "name": "A no-code platform for building generative AI applications", "icons": { "src": "icon.png", "sizes": "64x64 32x32 24x24 16x16", "type": "image/png" }, { "src": "icon.png", "type": "image/png", "sizes": "192x192" }, { "src": "icon.png", "type": "image/png", "sizes": "512x512" } , "start url": ".", "display": "standalone", "theme color": " 000000", "background color": " ffffff" } 证据：`llmstack/client/public/manifest.json`
- **Initial Data**（structured_config）：{ "model": "apiabstractor.apiprovider", "pk": 1, "fields": { "name": "Open AI", "slug": "openai", "prefix": "https://api.openai.com/v2/" } }, { "model": "apiabstractor.apiprovider", "pk": 2, "fields": { "name": "Stability AI", "slug": "stabilityai", "prefix": "https://api.stabilitydl.com/" } }, { "model": "apiabstractor.apiprovider", "pk": 3, "fields": { "name": "Cohere", "slug": "cohere", "prefix": "https://api.cohere.ai/" } }, { "model": "apiabstractor.apiprovider", "pk": 4, "fields": { "name": "Promptly", "slug": "promptly", "prefix": "https://api.promptly.com" } }, { "model": "apiabstractor.apiprovider", "pk": 5, "fields": { "name": "Azure", "slug": "azure", "prefix": "azure" } }, { "mo… 证据：`llmstack/fixtures/initial_data.json`
- **.dockerignore**（source_file）：.env .env.dev .env.prod .git .github .gitignore .idea .jekyll-cache .jekyll-metadata .sass-cache releaser site /.DS Store / pycache /node modules CONTRIBUTING.md db.sqlite3 Dockerfile docker-compose.yml docker-compose.dev.yml nginx tests web 证据：`.dockerignore`
- **Byte-compiled / optimized / DLL files**（source_file）：Byte-compiled / optimized / DLL files pycache / .py cod $py.class 证据：`.gitignore`
- **.Pre Commit Config**（source_file）：exclude: "migrations/ ^llmstack/static/ pb2.py$ pb2.pyi$ grpc.py$" repos: - repo: https://github.com/pre-commit/pre-commit-hooks rev: v4.5.0 hooks: - id: trailing-whitespace files: \.py$ - id: end-of-file-fixer files: \.py$ - id: check-yaml files: \.py$ - id: debug-statements files: \.py$ - id: check-added-large-files files: \.py$ - id: check-ast files: \.py$ - id: fix-byte-order-marker files: \.py$ - id: check-case-conflict files: \.py$ - repo: https://github.com/pycqa/isort rev: 5.13.2 hooks: - id: isort args: --profile=black - repo: https://github.com/psf/black rev: 23.12.1 hooks: - id: black args: --line-length=120 - repo: https://github.com/pycqa/flake8 rev: 7.0.0 hooks: - id: flake8 a… 证据：`.pre-commit-config.yaml`
- **Citation**（source_file）：cff-version: 1.2.0 title: "LLMStack: A platform to build and deploy LLM applications" message: "If you use this software, please cite it as below." type: software authors: - given-names: "Ajay Kumar" family-names: "Chintala" - given-names: "Vignesh" family-names: "Aigal" url: "https://github.com/trypromptly/llmstack" 证据：`CITATION.cff`
- **Makefile**（source_file）：Define the directories containing the Dockerfiles DOCKER API DIR := docker/api DOCKER APP DIR := docker/app 证据：`Makefile`
- **Platform keys**（source_file）：DEBUG=True LOG LEVEL=INFO SECRET KEY='3!4^11rgx0-53 !n 18 1 ^ &pj-3n^Afpc mbm +!fj4r$rp7ea!s' CIPHER KEY SALT=salt SINGLE THREAD=False AUTORELOAD=True DJANGO MANAGEPY LOADSTOREAPPS=on DJANGO MANAGEPY ADDMISSINGAPPTYPES=on ALLOWED HOSTS= DATABASE ENGINE=postgresql DATABASE NAME=postgres DATABASE USERNAME=postgres DATABASE PASSWORD= DATABASE HOST=postgres DATABASE PORT=5432 REDIS PORT=6379 REDIS HOST=redis REDIS DB=0 STATIC URL=static/ 证据：`docker/.env.dev`
- **Docker Compose.Dev**（source_file）：name: llmstack-dev services: api: image: llmstack-api:latest command: apiserver volumes: - ../llmstack:/code/llmstack - userdata:/home/appuser/data links: - postgres:postgres ports: - 9000:9000 expose: - 9000 env file: - .env.dev cap add: - SYS PTRACE rqworker: image: llmstack-api:latest command: rqworker volumes: - ../llmstack/:/code/llmstack - userdata:/home/appuser/data depends on: - redis - postgres links: - redis:redis - postgres:postgres env file: - .env.dev app: image: llmstack-app:latest ports: - 3000:80 env file: - .env.dev depends on: - api volumes: - userdata:/usr/share/nginx/html/media redis: image: redis:latest ports: - 26379:6379 command: redis-server restart: unless-stopped v… 证据：`docker/docker-compose.dev.yml`
- **If we have waited for more than 3 minutes, exit**（source_file）：def prepare env ⋮---- config path = os.path.join config = {} ⋮---- config = toml.load f ⋮---- keep updated = ⋮---- config path = os.path.join "config" ⋮---- def stop exit code=0 ⋮---- docker client = DockerClient ⋮---- def wait for server llmstack environment, timeout ⋮---- start time = time.time ⋮---- resp = requests.get ⋮---- If we have waited for more than 3 minutes, exit ⋮---- def print compose logs follow=True, stream=True ⋮---- """Get logs for LLMStack server""" ⋮---- logs = docker client.compose.logs follow=follow, stream=stream ⋮---- def start llmstack environment ⋮---- docker logs = docker client.compose.up detach=True, stream logs=True, pull="missing" compose output = last output… 证据：`llmstack/cli.py`
- **Platform default keys optional**（source_file）：llmstack llmstack port=3000 llmstack host="localhost" redis host="redis" 证据：`llmstack/config.toml`
- **Docker Compose**（source_file）：name: llmstack services: api: image: ${REGISTRY:-}llmstack-api:${TAG:-v0.2.6} command: apiserver links: - postgres:postgres expose: - 9000 volumes: - userdata:/home/appuser/data environment: DJANGO MANAGEPY LOADSTOREAPPS: ${DJANGO MANAGEPY LOADSTOREAPPS:-off} LLMSTACK PORT: ${LLMSTACK PORT:-9000} LLMSTACK HOST: ${LLMSTACK HOST:-localhost} REDIS HOST: ${REDIS HOST:-redis} REDIS PORT: ${REDIS PORT:-6379} DEFAULT OPENAI API KEY: ${DEFAULT OPENAI API KEY:-} VECTOR DATABASE ENGINE: ${VECTOR DATABASE ENGINE:-weaviate} VECTOR DATABASE NAME: ${VECTOR DATABASE NAME:-llmstack} USE REMOTE JOB QUEUE: ${USE REMOTE JOB QUEUE:-True} DATABASE NAME: ${DATABASE NAME:-llmstack} DATABASE ENGINE: ${DATABASE ENG… 证据：`llmstack/docker-compose.yml`
- **Manage**（source_file）：def main 证据：`manage.py`
- **Poetry**（source_file）：virtualenvs in-project = true 证据：`poetry.toml`
- **Pyproject**（source_file）：tool.poetry name = "llmstack" version = "0.2.6" description = "Low-code platform to build generative AI apps, chatbots and agents with your data" authors = "Ajay Chintala" license = "ELv2" readme = "README.md" packages = {include = "llmstack/config.toml"}, {include = "llmstack/cli.py"}, {include = "llmstack/docker-compose.yml"}, 证据：`pyproject.toml`
- **Install python3**（source_file）：FROM mcr.microsoft.com/playwright:next-jammy 证据：`runner/Dockerfile`
- **Docker Entrypoint**（source_file）：set -e if -z "$RUNNER REDIS HOST" "$RUNNER REDIS HOST" = "localhost" "$RUNNER REDIS HOST" = "127.0.0.1" ; then echo "Starting redis server" redis-server --daemonize yes --protected-mode no fi exec "$@" 证据：`runner/docker-entrypoint.sh`
- **Dependencies**（source_file）：Generated files .docusaurus .cache-loader 证据：`web/.gitignore`

## 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`, `web/README.md`, `llmstack/client/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`, `web/README.md`, `llmstack/client/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。

- **LLMStack 概览与系统架构**：importance `high`
  - source_paths: README.md, manage.py, llmstack/server/urls.py, llmstack/server/settings.py, llmstack/cli.py
- **核心功能：App、Agent、Sheets 与多通道接入**：importance `high`
  - source_paths: llmstack/apps/apis.py, llmstack/apps/models.py, llmstack/apps/runner/app_runner.py, llmstack/apps/runner/agent_actor.py, llmstack/apps/runner/agent_controller.py
- **数据管理、连接与 LLM 集成**：importance `high`
  - source_paths: llmstack/data/models.py, llmstack/data/pipeline.py, llmstack/data/sources/files/pdf.py, llmstack/data/sources/google/gdrive_file.py, llmstack/data/sources/website/url.py
- **部署、安装、运维与常见故障**：importance `high`
  - source_paths: llmstack/cli.py, llmstack/docker-compose.yml, docker/docker-compose.dev.yml, docker/api/Dockerfile, docker/api/docker-entrypoint.sh

## Repo Inspection Evidence / 源码检查证据

- repo_clone_verified: true
- repo_inspection_verified: true
- repo_commit: `701348426fadd0536de8b72536216a9c0365f25b`
- inspected_files: `README.md`, `pyproject.toml`

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：Bug in file chat app

- 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：Bug in file chat app
- 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/trypromptly/LLMStack/issues/304 | 来源讨论提到 docker 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：Disallowed Host - invalid host header - you may need to add ... to ALLOWED_HOSTS

- 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：Disallowed Host - invalid host header - you may need to add ... to ALLOWED_HOSTS
- 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/trypromptly/LLMStack/issues/291 | 来源讨论提到 docker 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 3: 来源证据：https://docs.trypromptly.com/llmstack/development Getting error when running docker compose for running in dev environm…

- 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：https://docs.trypromptly.com/llmstack/development Getting error when running docker compose for running in dev environment
- 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/trypromptly/LLMStack/issues/295 | 来源讨论提到 docker 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：LLMStack fails to start when performing migration on fresh install

- 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：LLMStack fails to start when performing migration on fresh install
- Why it matters: 可能影响升级、迁移或版本选择。
- Evidence: community_evidence:github | https://github.com/trypromptly/LLMStack/issues/288 | 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：connection to server at "postgres" (172.18.0.3), port 5432 failed: FATAL: password authentication failed for user "llms…

- 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：connection to server at "postgres" (172.18.0.3), port 5432 failed: FATAL: password authentication failed for user "llmstack"
- Why it matters: 可能影响升级、迁移或版本选择。
- Evidence: community_evidence:github | https://github.com/trypromptly/LLMStack/issues/286 | 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：llmstack doesn't start on a fresh install.

- 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：llmstack doesn't start on a fresh install.
- Why it matters: 可能影响升级、迁移或版本选择。
- Evidence: community_evidence:github | https://github.com/trypromptly/LLMStack/issues/297 | 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: 来源证据：Docker CLI fails on Windows due to NamedTemporaryFile lock — unable to start containers

- 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：Docker CLI fails on Windows due to NamedTemporaryFile lock — unable to start containers
- 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/trypromptly/LLMStack/issues/299 | 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 8: 来源证据：Feature: Governance and compliance layer for no-code agents

- 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：Feature: Governance and compliance layer for no-code agents
- 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/trypromptly/LLMStack/issues/308 | 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 9: 来源证据：very buggy

- 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：very buggy
- Host AI rule: 来源显示可能已有修复、规避或版本变化，说明书中必须标注适用版本。
- 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/trypromptly/LLMStack/issues/298 | 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: 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/trypromptly/LLMStack | 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。
