[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-m0n0x41d--quint-code":3,"tool-m0n0x41d--quint-code":64},[4,17,27,35,48,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",140436,2,"2026-04-05T23:32:43",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,43,44,45,15,46,26,13,47],"数据工具","视频","插件","其他","音频",{"id":49,"name":50,"github_repo":51,"description_zh":52,"stars":53,"difficulty_score":10,"last_commit_at":54,"category_tags":55,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,46],{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},2181,"OpenHands","OpenHands\u002FOpenHands","OpenHands 是一个专注于 AI 驱动开发的开源平台，旨在让智能体（Agent）像人类开发者一样理解、编写和调试代码。它解决了传统编程中重复性劳动多、环境配置复杂以及人机协作效率低等痛点，通过自动化流程显著提升开发速度。\n\n无论是希望提升编码效率的软件工程师、探索智能体技术的研究人员，还是需要快速原型验证的技术团队，都能从中受益。OpenHands 提供了灵活多样的使用方式：既可以通过命令行（CLI）或本地图形界面在个人电脑上轻松上手，体验类似 Devin 的流畅交互；也能利用其强大的 Python SDK 自定义智能体逻辑，甚至在云端大规模部署上千个智能体并行工作。\n\n其核心技术亮点在于模块化的软件智能体 SDK，这不仅构成了平台的引擎，还支持高度可组合的开发模式。此外，OpenHands 在 SWE-bench 基准测试中取得了 77.6% 的优异成绩，证明了其解决真实世界软件工程问题的能力。平台还具备完善的企业级功能，支持与 Slack、Jira 等工具集成，并提供细粒度的权限管理，适合从个人开发者到大型企业的各类用户场景。",70626,"2026-04-05T22:51:36",[26,15,13,45],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":81,"owner_twitter":82,"owner_website":83,"owner_url":84,"languages":85,"stars":94,"forks":95,"last_commit_at":96,"license":97,"difficulty_score":23,"env_os":98,"env_gpu":99,"env_ram":99,"env_deps":100,"category_tags":104,"github_topics":105,"view_count":10,"oss_zip_url":82,"oss_zip_packed_at":82,"status":16,"created_at":126,"updated_at":127,"faqs":128,"releases":158},307,"m0n0x41d\u002Fquint-code","quint-code","Engineering decisions  engine that know when they're stale. Frame, compare, decide — with evidence decay and parity enforcement. For Claude Code, Cursor, Gemini CLI, Codex and more.","quint-code 是一款专为工程决策设计的智能引擎，能无缝连接 Claude Code、Cursor 等主流 AI 编程助手。quint-code 解决了技术方案随时间变旧却无人知晓的问题，确保重要决策始终有据可查。\n\n适合需要频繁做架构选型或技术评估的开发者使用。quint-code 的核心亮点是“证据衰减”功能，决策的可信度会随时间自动降低，过期后自动提醒复查。quint-code 还能将决策整理成标准文档，包含问题背景、约束条件和后果，方便新人快速接手。同时，系统会强制公平对比不同方案，避免被错误指标误导。quint-code 具备跨会话记忆能力，能主动关联过往相关决策。\n\n用户只需输入 `\u002Fq-reason`，quint-code 就能自动分析问题、对比方案并记录决定，也可以手动分步操作。无论是临时的小决定还是长期的架构规划，quint-code 都能帮团队建立起不会过期的工程决策记忆库。","\u003Cimg src=\"assets\u002Fbanner.svg\" alt=\"Quint Code\" width=\"600\">\n\n**Engineering decisions that know when they're stale.**\n\nFrame problems. Compare options fairly. Record decisions as contracts. Know when to revisit.\n\nSupports: Claude Code, Cursor, Gemini CLI, Codex CLI, Codex App, Air\n\n---\n\n## Install\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\nThen in your project, run init **with your tool's flag**:\n\n```bash\n# Claude Code (default if no flag)\nquint-code init\n\n# Cursor\nquint-code init --cursor\n\n# Gemini CLI\nquint-code init --gemini\n\n# Codex CLI \u002F Codex App\nquint-code init --codex\n\n# JetBrains Air\nquint-code init --air\n\n# All tools at once\nquint-code init --all\n```\n\n### What init does per tool\n\nThe binary is the same — only the MCP config location differs:\n\n| Tool | MCP Config | Commands | Skill |\n|------|-----------|----------|-------|\n| Claude Code | `.mcp.json` (project root) | `~\u002F.claude\u002Fcommands\u002F` | `~\u002F.claude\u002Fskills\u002Fq-reason\u002F` |\n| Cursor | `.cursor\u002Fmcp.json` | `~\u002F.cursor\u002Fcommands\u002F` | `~\u002F.cursor\u002Fskills\u002Fq-reason\u002F` |\n| Gemini CLI | `~\u002F.gemini\u002Fsettings.json` | `~\u002F.gemini\u002Fcommands\u002F` | — |\n| Codex CLI | `.codex\u002Fconfig.toml` | `~\u002F.codex\u002Fprompts\u002F` | `~\u002F.agents\u002Fskills\u002Fq-reason\u002F` |\n| Air | `.codex\u002Fconfig.toml` | project `skills\u002F` | project `skills\u002Fq-reason\u002F` |\n\n**Important for Cursor:** After init, open Cursor Settings → MCP → find `quint-code` → enable the toggle. Cursor adds MCP servers as disabled by default.\n\n**Note:** Cursor also picks up Claude Code commands from `~\u002F.claude\u002Fcommands\u002F` — so slash commands work even without `--cursor`. But MCP config (`.cursor\u002Fmcp.json`) must be set up for the tools to connect.\n\nExisting project? Run `\u002Fq-onboard` after init — the agent scans your codebase for existing decisions worth capturing.\n\n**First time?** Ask the agent to explain how it works:\n\n```\n\u002Fq-reason explain how to work with quint-code effectively — what commands exist, when to use each one, and what's the recommended workflow\n```\n\nThe agent has full knowledge of all Quint tools and will walk you through them in context of your project.\n\n---\n\n## How It Works\n\n### One command: `\u002Fq-reason`\n\nDescribe your problem. The agent frames it, generates alternatives, compares them fairly, and records the decision — all in one command. It auto-selects the right depth.\n\n### Or drive each step manually\n\n```\n\u002Fq-frame  → \u002Fq-char  → \u002Fq-explore → \u002Fq-compare → \u002Fq-decide\n  what's      what       genuinely     fair         engineering\n  broken?     matters?   different     comparison   contract\n                         options\n```\n\n### Micro-decisions on the fly\n\nThe agent captures decisions automatically when it notices them in conversation. No rationale — no record. Conflicts with active decisions are flagged. Auto-expires in 90 days.\n\n### When decisions go stale\n\n`\u002Fq-status` shows what's expired and what needs attention. `\u002Fq-refresh` manages the lifecycle of ALL artifact types — waive, reopen, supersede, or deprecate.\n\n---\n\n## What Makes It Different\n\n- **Decisions are live** — they have computed trust scores (R_eff) that degrade as evidence ages. An expired benchmark drops the whole score.\n- **Comparison is honest** — parity enforced, dimensions cross-checked, asymmetric scoring warned. Anti-Goodhart: tag dimensions as \"observation\" to prevent optimizing the wrong metric.\n- **Memory across sessions** — when you frame a problem, the tool surfaces related past decisions. When you explore, it checks for similar variants.\n- **The loop closes** — failed measurements suggest reopening. Evidence decay triggers review. Periodic refresh prompts ensure nothing goes stale silently.\n- **Decisions are contracts** — FPF E.9 format: Problem Frame, Decision (invariants + DO\u002FDON'T), Rationale, Consequences. A new engineer reads it 6 months later and gets everything.\n\n---\n\n## 6 Tools\n\n| Tool | What it does |\n|------|-------------|\n| `quint_note` | Micro-decisions with validation + auto-expiry |\n| `quint_problem` | Frame problems, define comparison dimensions with roles |\n| `quint_solution` | Explore variants with diversity check, compare with parity |\n| `quint_decision` | FPF E.9 decision contract, impact measurement, evidence |\n| `quint_refresh` | Lifecycle management for all artifacts |\n| `quint_query` | Search, status dashboard, file-to-decision lookup |\n\n---\n\n## Built on First Principles Framework\n\n[FPF](https:\u002F\u002Fgithub.com\u002Failev\u002FFPF) by [Anatoly Levenchuk](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Failev\u002F) — a rigorous, transdisciplinary architecture for thinking.\n\n`\u002Fq-reason` gives your AI agent an FPF-native operating system for engineering decisions: problem framing before solutions, characterization before comparison, parity enforcement, evidence with congruence penalties, weakest-link assurance, and the lemniscate cycle that closes itself when evidence ages or measurements fail.\n\n`quint-code fpf search` gives you access to 4243 indexed sections from the FPF specification — the agent can look up any concept on demand.\n\n---\n\n## Learn More\n\nSee the [documentation](https:\u002F\u002Fquint.codes\u002Flearn) for detailed guides on decision modes, the DRR format, computed features, and lifecycle management.\n\n## Requirements\n\n- Go 1.24+ (for building from source)\n- Any MCP-capable AI tool\n\n## License\n\nMIT\n","\u003Cimg src=\"assets\u002Fbanner.svg\" alt=\"Quint Code\" width=\"600\">\n\n**知道何时过时的工程决策。**\n\n框定问题。公平比较选项。将决策记录为契约。知道何时需要重新审视。\n\n支持：Claude Code, Cursor, Gemini CLI, Codex CLI, Codex App, Air\n\n---\n\n## 安装\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\n然后在您的项目中，运行 init **并带上您所用工具的标志（flag）**：\n\n```bash\n# Claude Code (default if no flag)\nquint-code init\n\n# Cursor\nquint-code init --cursor\n\n# Gemini CLI\nquint-code init --gemini\n\n# Codex CLI \u002F Codex App\nquint-code init --codex\n\n# JetBrains Air\nquint-code init --air\n\n# All tools at once\nquint-code init --all\n```\n\n### 各工具的初始化操作\n\n二进制文件相同 —— 仅 MCP（Model Context Protocol）配置位置不同：\n\n| 工具 | MCP 配置 | 命令 | 技能 |\n|------|-----------|----------|-------|\n| Claude Code | `.mcp.json` (project root) | `~\u002F.claude\u002Fcommands\u002F` | `~\u002F.claude\u002Fskills\u002Fq-reason\u002F` |\n| Cursor | `.cursor\u002Fmcp.json` | `~\u002F.cursor\u002Fcommands\u002F` | `~\u002F.cursor\u002Fskills\u002Fq-reason\u002F` |\n| Gemini CLI | `~\u002F.gemini\u002Fsettings.json` | `~\u002F.gemini\u002Fcommands\u002F` | — |\n| Codex CLI | `.codex\u002Fconfig.toml` | `~\u002F.codex\u002Fprompts\u002F` | `~\u002F.agents\u002Fskills\u002Fq-reason\u002F` |\n| Air | `.codex\u002Fconfig.toml` | project `skills\u002F` | project `skills\u002Fq-reason\u002F` |\n\n**Cursor 重要提示：** 初始化后，打开 Cursor 设置 → MCP → 找到 `quint-code` → 启用开关。Cursor 默认会禁用 MCP 服务器。\n\n**注意：** Cursor 也会从 `~\u002F.claude\u002Fcommands\u002F` 读取 Claude Code 命令 —— 因此即使没有 `--cursor`，斜杠命令也能工作。但必须设置 MCP 配置 (`.cursor\u002Fmcp.json`) 才能让工具连接。\n\n已有项目？在初始化后运行 `\u002Fq-onboard` —— 代理（Agent）会扫描您的代码库以捕获值得记录的现有决策。\n\n**第一次使用？** 让代理解释其工作原理：\n\n```\n\u002Fq-reason explain how to work with quint-code effectively — what commands exist, when to use each one, and what's the recommended workflow\n```\n\n该代理完全了解所有 Quint 工具，并将结合您的项目上下文引导您使用它们。\n\n---\n\n## 工作原理\n\n### 一个命令：`\u002Fq-reason`\n\n描述您的问题。代理会框定它、生成替代方案、公平比较它们并记录决策 —— 所有这些都在一个命令中完成。它会自动选择合适的深度。\n\n### 或者手动驱动每一步\n\n```\n\u002Fq-frame  → \u002Fq-char  → \u002Fq-explore → \u002Fq-compare → \u002Fq-decide\n  what's      what       genuinely     fair         engineering\n  broken?     matters?   different     comparison   contract\n                         options\n```\n\n### 即时微决策\n\n当代理在对话中注意到决策时，会自动捕获它们。没有理由 —— 就没有记录。与活跃决策的冲突会被标记。90 天后自动过期。\n\n### 当决策过时\n\n`\u002Fq-status` 显示哪些已过期以及需要注意的内容。`\u002Fq-refresh` 管理所有工件（Artifact）类型的生命周期 —— 豁免、重新打开、取代或弃用。\n\n---\n\n## 有何不同\n\n- **决策是实时的** —— 它们拥有计算得出的信任分数（R_eff），会随着证据老化而降低。过期的基准测试会降低整体分数。\n- **比较是诚实的** —— 强制执行对等性，交叉检查维度，警告不对称评分。反古德哈特效应（Anti-Goodhart）：将维度标记为“观察”，以防止优化错误的指标。\n- **跨会话记忆** —— 当您框定问题时，工具会浮现相关的过往决策。当您探索时，它会检查是否存在类似的变体。\n- **闭环** —— 失败的测量建议重新打开。证据衰减触发审查。定期刷新提示确保没有任何内容静默过时。\n- **决策即契约** —— FPF E.9 格式：问题框架、决策（不变量 + DO\u002FDON'T）、理由、后果。新工程师六个月后阅读它也能完全理解。\n\n---\n\n## 6 个工具\n\n| 工具 | 功能 |\n|------|-------------|\n| `quint_note` | 带验证和自动过期的微决策 |\n| `quint_problem` | 框定问题，定义带有角色的比较维度 |\n| `quint_solution` | 探索变体并进行多样性检查，进行对等比较 |\n| `quint_decision` | FPF E.9 决策契约，影响测量，证据 |\n| `quint_refresh` | 所有工件的生命周期管理 |\n| `quint_query` | 搜索，状态仪表板，文件到决策查找 |\n\n---\n\n## 基于第一性原理框架\n\n[FPF（第一性原理框架）](https:\u002F\u002Fgithub.com\u002Failev\u002FFPF) by [Anatoly Levenchuk](https:\u002F\u002Fwww.linkedin.com\u002Fin\u002Failev\u002F) —— 一种严谨的、跨学科的思维架构。\n\n`\u002Fq-reason` 为您的 AI 代理提供了一个用于工程决策的 FPF 原生操作系统：解决方案前的问题框定，比较前的特征化，对等性执行，带有一致性惩罚的证据，最弱环节保证，以及当证据老化或测量失败时自我闭合的双纽线循环（lemniscate cycle）。\n\n`quint-code fpf search` 让您访问 FPF 规范中的 4243 个索引章节 —— 代理可按需查找任何概念。\n\n---\n\n## 了解更多\n\n查看 [文档](https:\u002F\u002Fquint.codes\u002Flearn) 以获取有关决策模式、DRR 格式、计算功能和生命周期管理的详细指南。\n\n## 要求\n\n- Go 1.24+（用于从源代码构建）\n- 任何支持 MCP 的 AI 工具\n\n## 许可证\n\nMIT","# quint-code 快速上手指南\n\n**quint-code** 是一个基于第一性原理框架（FPF）的 AI 工程决策助手。它能帮助你在开发过程中记录决策契约、对比选项，并自动管理决策的生命周期（如过期提醒）。它通过 MCP 协议集成到各类 AI 编程工具中。\n\n## 环境准备\n\n在使用之前，请确保满足以下条件：\n\n- **AI 工具支持**：你需要一个支持 MCP（Model Context Protocol）的 AI 编程工具。官方支持的包括：\n  - Claude Code\n  - Cursor\n  - Gemini CLI\n  - Codex CLI \u002F Codex App\n  - JetBrains Air\n- **网络环境**：安装脚本依赖 GitHub Raw 资源，请确保网络可访问 `raw.githubusercontent.com`。\n- **系统要求**：Linux 或 macOS 终端环境（Windows 用户建议使用 WSL）。\n- **Go 语言**：仅当你需要从源码编译时必需（Go 1.24+），预编译二进制文件无需此依赖。\n\n## 安装步骤\n\n### 1. 安装二进制文件\n\n在终端中运行以下命令进行安装：\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\n### 2. 初始化项目\n\n进入你的项目目录，根据你使用的 AI 工具运行对应的初始化命令。\n\n**通用初始化（默认适配 Claude Code）：**\n```bash\nquint-code init\n```\n\n**指定工具初始化：**\n```bash\n# Cursor\nquint-code init --cursor\n\n# Gemini CLI\nquint-code init --gemini\n\n# Codex CLI \u002F Codex App\nquint-code init --codex\n\n# JetBrains Air\nquint-code init --air\n\n# 同时为所有工具初始化\nquint-code init --all\n```\n\n> **⚠️ 重要提示（针对 Cursor 用户）**：\n> 初始化后，打开 Cursor 设置 → MCP → 找到 `quint-code` → **启用开关**。Cursor 默认会禁用新添加的 MCP 服务器。\n\n### 3. 现有项目接入（可选）\n\n如果是已有项目，初始化后可运行以下命令让 Agent 扫描代码库中值得记录的既有决策：\n\n```bash\n\u002Fq-onboard\n```\n\n## 基本使用\n\n### 核心指令：一键推理\n\n最简便的使用方式是直接调用 `\u002Fq-reason`。Agent 会自动完成问题构建、方案生成、公平对比及决策记录。\n\n**示例：**\n```text\n\u002Fq-reason 如何在微服务架构中处理分布式事务？\n```\n\n### 分步手动控制\n\n如果你希望更精细地控制流程，可以按顺序执行以下步骤：\n\n| 指令 | 作用 |\n| :--- | :--- |\n| `\u002Fq-frame` | 定义问题框架 |\n| `\u002Fq-char` | 明确关键要素 |\n| `\u002Fq-explore` | 探索不同选项 |\n| `\u002Fq-compare` | 公平对比方案 |\n| `\u002Fq-decide` | 记录最终决策契约 |\n\n### 查看状态与生命周期\n\n- **查看决策状态**（检查是否有过期或需关注的决策）：\n  ```text\n  \u002Fq-status\n  ```\n- **管理生命周期**（更新、作废或重新打开决策）：\n  ```text\n  \u002Fq-refresh\n  ```\n\n### 新手引导\n\n如果你是第一次使用，可以让 Agent 为你解释工作流程：\n\n```text\n\u002Fq-reason explain how to work with quint-code effectively — what commands exist, when to use each one, and what's the recommended workflow\n```","某电商后端团队在重构用户中心时，面临高并发场景下选择 Redis 分布式缓存还是本地内存缓存的关键架构决策。\n\n### 没有 quint-code 时\n- 决策理由散落在会议录音和即时通讯软件中，新入职工程师无法快速理解历史背景与技术约束。\n- 业务量增长后缓存策略失效，团队却找不到当初排除其他方案的完整对比数据，只能盲目试错。\n- 缺乏生命周期管理，过时的技术选型像“僵尸文档”一样继续误导后续开发，导致系统性能瓶颈难以根除。\n- 每次复盘都需人工翻找旧记录，耗时费力且容易遗漏关键约束条件，造成重复劳动。\n\n### 使用 quint-code 后\n- quint-code 将决策固化为包含问题框架与权衡依据的工程合同，新人阅读即可掌握全貌，无需追问前人。\n- 证据衰减机制自动监控决策有效性，当缓存指标异常时通过 `\u002Fq-status` 主动提示重新评估，防止隐患累积。\n- 强制平行对比维度，防止因短期性能优化牺牲长期可维护性，确保架构演进逻辑严密且公平。\n- 决策冲突自动预警，结合会话记忆功能，避免重复讨论已解决的技术难题，提升团队协作效率。\n\nquint-code 让工程决策从“一次性谈话”转变为可追踪、会自省且随时间演进的活文档。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fm0n0x41d_quint-code_6084ea38.png","m0n0x41d","ivan zakutni","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fm0n0x41d_069ec290.jpg","there is no rest for the clicked","OCTA","Armenia","zakutnii.ivan@gmail.com",null,"https:\u002F\u002Fivanzakutnii.com","https:\u002F\u002Fgithub.com\u002Fm0n0x41d",[86,90],{"name":87,"color":88,"percentage":89},"Go","#00ADD8",98.6,{"name":91,"color":92,"percentage":93},"Shell","#89e051",1.4,1251,101,"2026-04-05T22:38:26","NOASSERTION","Linux, macOS","未说明",{"notes":101,"python":99,"dependencies":102},"需安装 Go 1.24+ 以从源码构建；运行时需配合支持 MCP 协议的 AI 工具（如 Claude Code, Cursor, Gemini CLI 等）；使用 Cursor 时需在设置中手动启用 MCP 服务器开关；安装脚本基于 Bash，路径结构表明主要支持类 Unix 系统。",[103],"Go 1.24+",[15,45],[106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125],"ai-agents","ai-coding","anthropic","claude","claude-code","cli","developer-tools","first-principles","prompt-engineering","reasoning","structured-reasoning","systems-thinking","codex-cli","cursor","cursor-ai","gemini-cli","context-memory","ai-skills","air","codex","2026-03-27T02:49:30.150509","2026-04-06T08:17:40.477560",[129,134,139,144,148,153],{"id":130,"question_zh":131,"answer_zh":132,"source_url":133},1050,"在 WSL2 终端中使用 Codex CLI 时，quint_decide 阶段报错 'Transport Closed' 怎么办？","这是 WSL2 环境下的已知问题，已在 v4.1.1 版本修复。请运行以下脚本重新安装 quint-code：`curl -fsSL https:\u002F\u002Fquint.codes\u002Finstall.sh | bash`。如果问题依旧，请检查是否有其他传输相关错误。","https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F21",{"id":135,"question_zh":136,"answer_zh":137,"source_url":138},1051,"为什么 `quint_problem select` 会返回已弃用（deprecated）的问题？","这是一个过滤逻辑的 Bug，旧版本中 `select` 查询未正确过滤 `status` 字段。该问题已在 v5.3.2 版本修复，请更新到该版本或更高版本以确保只返回活跃问题。","https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F38",{"id":140,"question_zh":141,"answer_zh":142,"source_url":143},1052,"为什么模块覆盖率报告显示为 'blind'（无决策），即使存在关联文件的决策？","这是由项目隔离设计导致的。每个 `quint-code init` 创建独立的领域边界和数据库。父项目无法查看子项目的决策。要查看特定模块的决策，请在该 `.quint` 目录所在的工作区直接打开 AI 工具，Quint 会自动连接对应的项目数据库。","https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F40",{"id":145,"question_zh":146,"answer_zh":147,"source_url":143},1053,"重命名包含扫描模块的项目文件夹后，Quint 会出现什么问题？","`codebase_modules.path` 在扫描时设置，不会随目录重命名自动更新。这会导致路径过时，父项目跟踪的子模块路径失效。目前需要手动更新数据库来解决，尚无 CLI 命令支持路径迁移。",{"id":149,"question_zh":150,"answer_zh":151,"source_url":152},1054,"能否让多个 Agent 同时使用 quint-code？如何配置？","可以支持多 Agent 协作。建议配置 MCP 命令调用 `quint-launcher.sh`（内部调用 `quint-launcher.py`），它能自动检测正确的工作空间目录并设置 `.quint` 环境。这样可以避免依赖全局环境变量，实现更稳定的多 Agent 状态管理。","https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F23",{"id":154,"question_zh":155,"answer_zh":156,"source_url":157},1055,"如何将自定义的“反技能退化”（anti-deskilling）技能与 Quint 的 FPF 推理结合使用？","建议采用“人类优先”的工作流：1. 让反技能退化层先触发，暂停并要求用户在关键决策点进行推理；2. 用户使用 `\u002Fq-frame` 记录并验证自己的思考；3. 最后由 Quint 的 FPF 流程（compare → decide）对已确认的决策进行结构化。避免 Quint 直接接管决策空间导致人类思考能力退化。","https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F36",[159,164,169,174,179,184,189,193,198,203,208,213,218,223,228],{"id":160,"version":161,"summary_zh":162,"released_at":163},110374,"v5.3.2","### Added\n- **Status column in Decisions tab** — board now shows Shipped\u002FPending status per decision.\n\n### Fixed\n- **Drift scan requires diff review before action** — `FormatDriftResponse` adds a guard: \"REQUIRED: read `git diff` on modified files before taking action.\" Classification rubric (cosmetic\u002Fmaterial\u002Fincidental) presented to agent.\n- **Drift output uses raw signals instead of interpretive labels** — \"likely implemented\" \u002F \"not yet implemented\" replaced with \"git activity detected \u002F not detected after decision date\". Tool outputs observable facts; agent performs all interpretation.\n- **`quint_problem(action=\"select\")` returns deprecated problems** ([#38](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F38)) — `SelectProblems` and `FindActiveProblem` applied status filter in Go after SQL `LIMIT`, so deprecated rows could push active ones out of the result window. Added `ListActiveByKind` with SQL-level `status='active'` filter.\n\n### Changed\n- **Board no longer reloads data on tab switch** — 3-second periodic refresh provides sufficient data freshness.\n- **CI: remove duplicate push trigger on dev branch** — push trigger now only fires on main (post-merge), preventing double CI runs on PRs.","2026-03-26T08:33:44",{"id":165,"version":166,"summary_zh":167,"released_at":168},110375,"v5.3.1","## Fixed\n\n- **NavStrip no longer triggers agent auto-execution** — \"Next:\" replaced with \"Available:\" + guard line. Slash commands instead of tool call syntax. Mode-aware actions show the right options per depth mode (tactical\u002Fstandard\u002Fdeep).\n- **`quint_solution(action=\"compare\")` rejected valid dimensions** — MCP string serialization caused silent parsing failure. Now uses `parseStringArrayFromArgs`\u002F`parseNestedStringMapFromArgs` helpers consistently.\n\n## Added\n\n- **NavStrip interpretation in q-reason skill** — documents that Available = menu, fewer steps ≠ fewer checkpoints, Path 3 override rules.\n\nFull changelog: https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fblob\u002Fmain\u002FCHANGELOG.md#531--2026-03-25","2026-03-25T13:18:42",{"id":170,"version":171,"summary_zh":172,"released_at":173},110376,"v5.3.0","## Quint Code v5.3.0\n\nFPF-native engineering decision tool. 6 MCP tools + interactive terminal dashboard. Works with Claude Code, Cursor, Gemini CLI, Codex CLI.\n\n### Install\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\nSee [CHANGELOG](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fblob\u002Fmain\u002FCHANGELOG.md) for full details.\n\n### Interactive Terminal Dashboard\n\nThe headline feature: `quint-code board` — a Bubbletea v2 TUI that gives you a real-time view of your project's decision health without starting an agent session.\n\n- **Overview** — health summary, activity timeline, depth distribution, module coverage, context map, evidence freshness\n- **Problems** — backlog with drill-in to full signal\u002Fconstraints\u002Facceptance\n- **Decisions** — list with R_eff scores and drift status, drill-in with glamour markdown rendering\n- **Modules** — coverage tree with decision counts per module\n- **Live refresh** — auto-updates every 3 seconds, no restart needed\n- **`--check` mode** — exit code 1 if critical issues exist, usable as pre-push hook\n\nNavigation: `tab`\u002F`1-4` switch views, `j\u002Fk` navigate, `enter` drill in, `esc` back, `q` quit.\n\n### Other Changes\n\n- **Decision mode computed from artifact chain** — mode now reflects the actual reasoning path taken (problem with characterization + portfolio with comparison = standard minimum), not what the agent declared. Agent can escalate but not downgrade.\n- **FTS5 search keyword enrichment** — `search_keywords` column on artifacts for agent-generated synonyms at write time. Accepted on `quint_note` and `quint_decision`.\n- **C\u002FC++ header-only module detection** — `-I` include directories registered as modules so dependency edges aren't dropped.\n- **`\u002Fq-refresh scan` rescans modules** — keeps dependency graph fresh alongside drift and stale checks.\n- **Notes excluded from drift detection** — observations don't need baselines.\n- **Module scanner excludes `.claude` and `.context`** — worktrees and reference repos no longer inflate module count.\n- **q-reason context-aware entry** — skill no longer always falls through into full FPF cycle. Three paths based on user intent: think-and-respond, prepare-and-wait (human drives), full autonomous. Default is prepare-and-wait.\n- **C\u002FC++ symlink-safe include resolution** — fixes silent edge loss on macOS symlinked checkouts.\n\n### Acknowledgments\n\nThanks to [@karabelaselias](https:\u002F\u002Fgithub.com\u002Fkarabelaselias) for contributing C\u002FC++ language support, reporting the variant parsing bug (#33), and raising the question about q-reason's interaction with anti-deskilling workflows that led to the context-aware entry fix.\n\n## What's Changed\n* 5.3.0 by @m0n0x41d in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F37","2026-03-25T09:07:30",{"id":175,"version":176,"summary_zh":177,"released_at":178},110377,"v5.2.0","## Quint Code v5.2.0\n\nFPF-native engineering decision tool. 6 MCP tools. Works with Claude Code, Cursor, Gemini CLI, Codex CLI.\n\n### Install\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\nSee [CHANGELOG](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fblob\u002Fmain\u002FCHANGELOG.md) for full details.\n\n### Added\n\n- **C\u002FC++ module detection** — `compile_commands.json` as primary source with directory-based fallback\n- **C\u002FC++ import parsing** — `#include \"...\"` edge extraction with `-I` flag resolution\n- **C\u002FC++ extensions** — `.c`, `.h`, `.cpp`, `.cc`, `.cxx`, `.hpp`, `.hxx` registered in language registry\n\n### Fixed\n\n- **`quint_solution(action=\"explore\")` rejected valid variants** — MCP clients serializing `variants` as JSON strings caused silent parse failure ([#33](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fissues\u002F33))\n- **Status always rescans modules** — fresh scan on every `quint_query(action=\"status\")`\n- **Symlink-safe path resolution** for C\u002FC++ on macOS\n\n## What's Changed\n* feat: add C\u002FC++ module detection and import parsing by @karabelaselias in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F32\n* Release 5.2.0: C\u002FC++ support, explore variant parsing fix by @m0n0x41d in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F35","2026-03-23T06:26:46",{"id":180,"version":181,"summary_zh":182,"released_at":183},110378,"v5.1.0","## Quint Code v5.1.0\n\n**Complete product redesign.** v5 is a new product — not backward-compatible with v4.\n\nNew: FPF E.9 decision records, computed R_eff with evidence decay, problem lifecycle (Backlog → In Progress → Addressed), diversity check, archive recall, parity enforcement, indicator roles, Goldilocks signals, lemniscate feedback loop, universal artifact refresh, and 4243 indexed FPF spec sections.\n\n6 MCP tools. Works with Claude Code, Cursor, Gemini CLI, Codex CLI.\n\n### Install\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\nSee [CHANGELOG](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fblob\u002Fmain\u002Fsrc\u002Fmcp\u002FCHANGELOG.md) for full details.\n\n## What's Changed\n* docs: split install commands to prevent accidental cd your-project by @m0n0x41d in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F29\n* Release 5.1.0 — Codebase Awareness, Cross-Project Recall, Decision Integrity by @m0n0x41d in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F31\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fcompare\u002Fv5.0.0...v5.1.0\n\n","2026-03-20T17:30:43",{"id":185,"version":186,"summary_zh":187,"released_at":188},110379,"v5.0.0","## Quint Code v5.0.0\n\n**Complete product redesign.** v5 is a new product — not backward-compatible with v4.\n\nNew: FPF E.9 decision records, computed R_eff with evidence decay, problem lifecycle (Backlog → In Progress → Addressed), diversity check, archive recall, parity enforcement, indicator roles, Goldilocks signals, lemniscate feedback loop, universal artifact refresh, and 4243 indexed FPF spec sections.\n\n6 MCP tools. Works with Claude Code, Cursor, Gemini CLI, Codex CLI.\n\n### Install\n\n```bash\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fquint-code\u002Fmain\u002Finstall.sh | bash\n```\n\nSee [CHANGELOG](https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fblob\u002Fmain\u002Fsrc\u002Fmcp\u002FCHANGELOG.md) for full details.\n\n## What's Changed\n* v5.0.0 — Complete product redesign by @m0n0x41d in https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fpull\u002F28\n\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fm0n0x41d\u002Fquint-code\u002Fcompare\u002Fv4.1.1...v5.0.0\n\n","2026-03-18T06:42:11",{"id":190,"version":191,"summary_zh":82,"released_at":192},110380,"v4.1.1","2026-01-02T09:00:56",{"id":194,"version":195,"summary_zh":196,"released_at":197},110381,"v4.1.0","### Added\n\n- **DRR Linking (FPF E.9)**: Decision records now create graph relations to hypotheses.\n  - `quint_decide` accepts new `rejected_ids` parameter for rejected alternatives.\n  - Creates `selects` relation: DRR → winner hypothesis.\n  - Creates `rejects` relations: DRR → each rejected alternative.\n  - Enables queries: \"What alternatives were considered for this DRR?\"\n\n- **Context Markdown Formatting**: `quint_record_context` now normalizes input to proper markdown.\n  - Vocabulary entries formatted as definition list (`- **Term**: definition`).\n  - Invariants formatted as numbered list with line breaks.\n\n- **Audit Tree Shows MemberOf Relations**: `quint_audit_tree` now displays `memberOf` groupings.\n  - Decision context alternatives are shown under `[members]` section.\n  - Member R scores displayed for comparison (no WLNK propagation per FPF spec).\n\n- **Improved Tool Descriptions**: `quint_propose` parameters now have actionable descriptions.\n  - `depends_on`: Explains WLNK impact and when to use.\n  - `decision_context`: Explains grouping purpose and when to use.\n  - `q1-hypothesize.md`: Added \"Linking Checklist\" before proposing hypotheses.\n\n- **Merged `\u002Fq-audit` into `\u002Fq-query`**: Simplified command set.\n  - `\u002Fq-query` now shows R_eff and dependency trees for found holons.\n  - Removed standalone `\u002Fq-audit` utility (redundant with enhanced `\u002Fq-query`).\n  - `\u002Fq4-audit` (FPF phase) remains unchanged.\n\n- **sqlc Integration**: Type-safe database queries generated from SQL.\n  - All database operations now use sqlc-generated code with proper type safety.\n  - New `db\u002Fstore.go` wrapper provides clean API while preserving schema bootstrap.\n  - Added comprehensive test suite for database operations (`db\u002Fstore_test.go`).\n\n- **GetHolon Query**: Added query to fetch hypothesis metadata by ID (foundation for future Kind-CAL work).\n\n- **New MCP Tools for Trust Calculus (B.3)**:\n  - `quint_audit_tree`: Visualize assurance tree with R scores, dependencies, and CL penalties.\n  - `quint_calculate_r`: Compute R_eff with detailed breakdown (self score, weakest link, decay penalties).\n  - `quint_check_decay`: Identify holons with expired evidence (epistemic debt detection).\n\n- **Parent ID Chain (FPF Enforcement)**:\n  - Added `parent_id` foreign key to holons table for L0→L1→L2 progression tracking.\n  - New queries: `GetHolonsByParent`, `GetHolonLineage` for traversing hypothesis chains.\n  - `CreateHolon` now accepts parent_id parameter for linking hypothesis progression.\n  - Enables auditable chain from L2 decision back to original L0 hypothesis.\n\n- **Derived Phase (FPF Enforcement)**:\n  - Phase is now computed from holons.layer data instead of stored in state.json.\n  - New `DerivePhase()` method computes phase from database state.\n  - New `GetPhase()` returns derived phase when DB available, falls back to State.Phase.\n  - Prevents AI bypass of FPF phase controls via direct file manipulation.\n\n- **Audit Logging (FPF Enforcement)**:\n  - New `audit_log` table tracks all MCP tool invocations.\n  - Captures: tool name, operation, actor, target ID, input hash, result, and details.\n  - Instrumented tools: `quint_propose`, `quint_verify`, `quint_decide`, `quint_move`.\n  - Enables detection of FPF bypasses through audit trail analysis.\n  - Context-aware logging supports multi-session isolation.\n\n- **Self-Healing Signed Projections (FPF Enforcement)**:\n  - All hypothesis\u002Fevidence\u002FDRR files now include `content_hash` in YAML frontmatter.\n  - New `WriteWithHash()` function adds cryptographic hash (SHA-256 truncated) on write.\n  - New `ValidateFile()` detects tampering by comparing stored vs computed hash.\n  - New `ReadWithValidation()` on Tools automatically detects and logs tampering.\n  - When tampering detected: regenerates file from DB (DB is source of truth).\n  - Tampering events logged to audit_log for violation tracking.\n\n- **DRR Holon Tracking**:\n  - `FinalizeDecision` now creates DRR holon in database (enables derived phase detection).\n  - DRR holons linked to winner hypothesis via parent_id.\n\n- **Tool Preconditions (FPF Enforcement)**:\n  - All MCP tools now validate preconditions before execution.\n  - `quint_propose`: Validates title, content, and kind fields.\n  - `quint_verify`: Confirms hypothesis exists in L0, validates verdict.\n  - `quint_test`: Ensures hypothesis is in L1 (not L0), validates verdict.\n  - `quint_audit`: Confirms hypothesis is in L2 before audit.\n  - `quint_decide`: Requires L2 hypotheses exist, validates winner_id and title.\n  - `quint_calculate_r` \u002F `quint_audit_tree`: Validates holon existence in DB.\n  - Precondition failures logged to audit_log with BLOCKED status.\n  - Each error includes actionable suggestion for the user.\n\n- **Command Contracts (FPF Enforcement)**:\n  - All FPF command prompts (q0-q5) now include formal enforcement contracts.\n  - YAML frontmatter with `pre`, `post`, `invariant`, and `required_tools` fields.\n  - RFC 2119 bindings (MUST, MUST NOT, SHALL) for mandatory behaviors.\n  - \"In","2025-12-21T20:01:29",{"id":199,"version":200,"summary_zh":201,"released_at":202},110382,"v4.0.0","### Added\n\n- **MCP Server Architecture**:\n  - This release introduces the MCP (Model Context Protocol) server as the core of Quint Code, replacing the previous prompt-only approach. The server provides structured tools for AI assistants to interact with the FPF knowledge base.\n  - The MCP server has been restructured into `cmd` and `internal` packages for better organization and maintainability.\n- **New Commands**:\n  - `\u002Fq-audit`: Visualize assurance tree with R-scores.\n  - `\u002Fq-decay`: Calculate epistemic debt from expired evidence.\n- **Command Updates**:\n  - `\u002Fq-actualize`: Reconcile the knowledge base with recent code changes. This command has been updated for better performance and accuracy.\n- **Renamed Commands**:\n  - `\u002Fq2-check` is now `\u002Fq2-verify`.\n  - `\u002Fq3-research` and `\u002Fq3-test` are consolidated into `\u002Fq3-validate`.\n  - `\u002Fq1-add` has been added for manually adding hypotheses.\n- **SQLite Database (`quint.db`)**:\n  - The project now uses SQLite for deterministic FPF, ensuring consistency and reproducibility.\n  - `holons` table now includes `scope`, `kind`, and `cached_r_score` columns to support advanced FPF features.\n  - `evidence` table now includes a `valid_until` column for evidence decay tracking.\n  - `relations` table now includes a `congruence_level` column for WLNK calculation.\n- **Trust & Assurance Calculator (B.3)**:\n  - Implemented the FPF B.3 standard for calculating trust and assurance.\n  - **Weakest Link (WLNK)**: R-score is now capped by the lowest-scoring dependency in the evidence chain.\n  - **Congruence Penalty**: Low-congruence relations between artifacts now reduce the overall reliability score.\n  - **Evidence Decay**: Expired evidence, as determined by the `valid_until` field, now penalizes the R-score, introducing the concept of \"epistemic debt.\"\n  - **Cycle Detection**: The calculator now detects and flags circular dependencies in the evidence graph.\n- **Typed Reasoning (Kind-CAL)**:\n  - Hypotheses are now classified by `kind` as either `system` (for code and architecture) or `episteme` (for knowledge and theory).\n  - Validation logic now branches based on the hypothesis kind, allowing for more targeted and relevant analysis.\n- **Characteristic Space (C.16)**:\n  - Success metrics are now defined upfront, before testing, as part of the `Characteristic Space`.\n  - These metrics are measured during the induction phase (`\u002Fq3-validate`).\n  - The results are recorded in the Design Rationale Record (DRR) for full traceability.\n- **CI\u002FCD**:\n  - A new GitHub Actions workflow has been added to automate the build and release process.\n- **Testing**:\n  - Added integration tests for the assurance calculator to ensure correctness and stability.\n- **Error Handling**:\n  - Improved error handling to surface previously-silent errors with warnings.\n\n### Changed\n\n- **Project Directory**: Renamed from `.fpf` to `.quint`. Migration: run `\u002Fq-actualize` to migrate, then delete `.fpf`.\n- **Multi-platform Support**:\n  - Installer now supports Claude Code, Cursor, Gemini CLI, and Codex CLI.\n  - Commands are now sourced from `src\u002Fcommands\u002F` and built to platform-specific formats in `dist\u002F`.\n\n### Removed\n\n- **Redundant Agents**: Removed legacy standalone agent files for a cleaner codebase.\n\n---","2025-12-18T19:11:33",{"id":204,"version":205,"summary_zh":206,"released_at":207},110383,"3.2.0","\r\n### Added: Process Hardening & Flexibility\r\n\r\n#### Strict Phase Gating (FPF Integrity)\r\n- **Hard Block in `\u002Fq1-hypothesize`:** Explicitly forbids generating new hypotheses if the cycle has passed Deduction. This prevents the \"Helpfulness Bias\" vulnerability where AI assistants might break process integrity to be \"nice\".\r\n- **Conditional Logic in `\u002Fq2-check`:** The cycle phase now only advances to `DEDUCTION_COMPLETE` when *all* active L0 hypotheses are resolved. If any remain unchecked, the door stays open for extensions.\r\n\r\n#### New Command: `\u002Fq1-extend`\r\n- **Legitimate Extension Path:** A dedicated command to add a missed hypothesis during the `ABDUCTION_COMPLETE` phase.\r\n- **Safety Rails:** Strictly blocked once `DEDUCTION_COMPLETE` is reached, ensuring evidence integrity (WLNK validity) during testing.\r\n\r\n### Changed\r\n- **Updated `\u002Fq-status`:** State machine visualization now includes the `(q1-extend)` loop.\r\n- **Refined `\u002Fq3-test` & `\u002Fq3-research`:** Reinforced checks to ensure testing only happens after deduction is fully complete.","2025-12-15T09:52:44",{"id":209,"version":210,"summary_zh":211,"released_at":212},110384,"3.1.0","Accumulated release.\r\n\r\n1. Major up to 3.0.0 due to rebranding of tool and commands (read the changelog for details)\r\n\r\n2. Release 3.1.0\r\n\r\n### Added: Deep Reasoning Capabilities\r\n\r\n1. Context Slicing (A.2.6)\r\n\r\nStructured Context: .fpf\u002Fcontext.md is now structured into explicit slices:\r\n\r\n- Slice: Grounding (Infrastructure, Region)\r\n- Slice: Tech Stack (Language, Frameworks)\r\n- Slice: Constraints (Compliance, Budget, Team)\r\n\r\nContext-Aware Init: \u002Fq0-init now scans package.json, Dockerfile, etc., to auto-populate slices.\r\n\r\n2. Explicit Role Injection (A.2)\r\n\r\nRole-Swapping Prompts: Commands now enforce specific FPF roles to prevent \"AI drift\":\r\n\r\n- \u002Fq1-hypothesize: ExplorerRole (Creative, Abductive)\r\n- \u002Fq2-check: LogicianRole (Strict, Deductive)\r\n- \u002Fq4-audit: AuditorRole (Adversarial, Normative)\r\n\r\nContext Drift Analysis\r\n\r\n- New Audit Step: \u002Fq4-audit now includes a mandatory Context Drift Check.\r\n- Validation: Verifies that hypotheses generated in step 1 still match the constraints in step 4 (preventing \"works on my machine\" architecture).\r\n\r\n### Changed\r\nCommand Prompts: Updated q0, q1, q2, q4 to enforce the new reasoning standards.","2025-12-14T19:41:46",{"id":214,"version":215,"summary_zh":216,"released_at":217},110385,"2.2.0","Multi-Platform Support 🔥\r\n\r\n  Crucible Code now works with four AI coding tools:\r\n  - Claude Code\r\n  - Cursor\r\n  - Gemini CLI\r\n  - Codex CLI\r\n\r\n  One-Liner Install\r\n\r\n  # Global (all projects)\r\n  curl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fcrucible-code\u002Fmain\u002Finstall.sh | bash -s -- -g\r\n\r\n  # Per-project (commit to repo)\r\n  curl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fm0n0x41d\u002Fcrucible-code\u002Fmain\u002Finstall.sh | bash\r\n\r\n  Interactive TUI lets you pick which platforms to install.\r\n\r\n  What's New\r\n\r\n  Multi-Platform Architecture\r\n  - Adapter-based build: src\u002Fcommands\u002F → dist\u002F{platform}\u002F\r\n  - Platform-specific formats (TOML for Gemini, Markdown for others)\r\n  - CI\u002FCD ensures dist\u002F stays in sync\r\n\r\n  TUI Installer\r\n  - Interactive platform selection with arrow keys or vim motions (j\u002Fk)\r\n  - Global (-g) or per-project install\r\n  - Bash 3.x compatible (works on macOS out of the box)\r\n  - Uninstall support (--uninstall)\r\n\r\n  Visual Refresh\r\n  - Melted steel gradient (red → white) for ASCII banner\r\n  - New SVG banner for GitHub README\r\n\r\n  Upgrade\r\n\r\n  Just re-run the installer to get the latest commands. No breaking changes to .fpf\u002F structure.\r\n","2025-12-14T10:52:08",{"id":219,"version":220,"summary_zh":221,"released_at":222},110386,"2.1.0","Release v2.1.0 Summary\r\n\r\n  Here is what has been upgraded in your codebase:\r\n\r\n   1. Agentic Initialization (`\u002Ffpf-0-init`)\r\n       * Now scans the repo (package.json, Dockerfile, etc.) to infer the Tech Stack.\r\n       * Interviews the user to fill in the gaps (Scale, Budget).\r\n       * Creates a living .fpf\u002Fcontext.md file that acts as the \"Base Slice\" for all future reasoning.\r\n\r\n   2. Strict Method\u002FWork Distinction (`\u002Ffpf-1-hypothesize`)\r\n       * The hypothesis template is now split into \"1. The Method\" (Plan) and \"2. The Validation\" (Execution).\r\n       * This enforces FPF Pattern A.15 without forcing the user to learn the ontology.\r\n\r\n   3. NQD & Formality Tracking\r\n       * Hypotheses now carry formality: [0-9] and novelty\u002Fcomplexity tags in frontmatter.\r\n       * \u002Ffpf-status (via the logic update we discussed earlier) will calculate Min\u002FMax Formality and Trust Index.\r\n\r\n   4. Context Awareness\r\n       * Research and Test commands now automatically read .fpf\u002Fcontext.md to ensure evidence congruence is calculated against the actual project\r\n         reality, not an abstract one.","2025-12-13T13:37:13",{"id":224,"version":225,"summary_zh":226,"released_at":227},110387,"2.0.0","Version 2.0.0 is a major release that transforms FPF from advisory guidance into an enforced reasoning framework. The core change: **phases are now gates, not suggestions.**\r\n\r\n### Headline Changes\r\n\r\n**1. Phase Gate Enforcement**\r\n- Commands now block invalid phase transitions. You can't `\u002Ffpf-2-check` without hypotheses, can't gather evidence without L1 claims.\r\n- The ADI cycle is strictly sequential for phases 1→2→3. Audit (phase 4) can be skipped to go directly to `\u002Ffpf-5-decide`, but this is not recommended.\r\n\r\n**2. Transformer Mandate Enforcement**\r\n- Explicit \"AWAIT HUMAN INPUT\" sections at all decision points\r\n- Claude generates options, presents analysis, then stops and waits\r\n- No more implicit decisions buried in command output\r\n\r\n**3. Quantitative WLNK Analysis**\r\n- `R_eff = R_base - Φ(CL)` formula now calculated in audit\r\n- Evidence chain tables show exactly where your argument is weakest\r\n- Congruence penalties formalized: High=0.00, Medium=0.15, Low=0.35\r\n\r\n**4. Plausibility Filters**\r\n- Hypotheses assessed against four filters: Simplicity, Explanatory Power, Consistency, Falsifiability\r\n- Clear verdicts: PLAUSIBLE \u002F MARGINAL \u002F IMPLAUSIBLE\r\n- Ranking table for comparison\r\n\r\n**5. Better Learning Preservation**\r\n- `\u002Ffpf-discard` now captures key insights before cleanup\r\n- Evidence validity windows mandatory\r\n- Epistemic debt tracking in `\u002Ffpf-decay`\r\n\r\n### Breaking Changes\r\n\r\nNone. Existing `.fpf\u002F` directories work without modification.\r\n\r\n**Recommended migration:**\r\n1. Run `\u002Ffpf-decay` to identify evidence needing validity dates\r\n2. Add congruence assessment to existing external evidence\r\n3. Existing DRRs remain valid\r\n\r\n### What's New in Documentation\r\n\r\n- README updated with phase strictness explanation and visual diagram annotations\r\n- Commands Reference table now shows which phases are required vs optional\r\n- \"Common Mistakes to Avoid\" sections in each command\r\n- Anti-pattern tables with explanations\r\n\r\n### Upgrade Path\r\n\r\n```bash\r\ncd crucible-code\r\ngit pull origin main\r\n.\u002Finstall.sh \u002Fpath\u002Fto\u002Fyour\u002Fproject\r\n```\r\n\r\nExisting FPF sessions continue working. New sessions get enforced gates","2025-12-13T12:45:42",{"id":229,"version":230,"summary_zh":231,"released_at":232},110388,"1.0.0","First, the somehow stable release of crucible-code includes a basic ADI reasoning cycle from FPF, which has proven to be valuable for engineering work.","2025-12-13T08:01:32"]