dominds 0.1.0

package/LICENSE

@@ -0,0 +1,157 @@
+# GNU LESSER GENERAL PUBLIC LICENSE
+
+Version 3, 29 June 2007
+
+Copyright (C) 2007 Free Software Foundation, Inc.
+<https://fsf.org/>
+
+Everyone is permitted to copy and distribute verbatim copies of this
+license document, but changing it is not allowed.
+
+This version of the GNU Lesser General Public License incorporates the
+terms and conditions of version 3 of the GNU General Public License,
+supplemented by the additional permissions listed below.
+
+## 0. Additional Definitions.
+
+As used herein, "this License" refers to version 3 of the GNU Lesser
+General Public License, and the "GNU GPL" refers to version 3 of the
+GNU General Public License.
+
+"The Library" refers to a covered work governed by this License, other
+than an Application or a Combined Work as defined below.
+
+An "Application" is any work that makes use of an interface provided
+by the Library, but which is not otherwise based on the Library.
+Defining a subclass of a class defined by the Library is deemed a mode
+of using an interface provided by the Library.
+
+A "Combined Work" is a work produced by combining or linking an
+Application with the Library. The particular version of the Library
+with which the Combined Work was made is also called the "Linked
+Version".
+
+The "Minimal Corresponding Source" for a Combined Work means the
+Corresponding Source for the Combined Work, excluding any source code
+for portions of the Combined Work that, considered in isolation, are
+based on the Application, and not on the Linked Version.
+
+The "Corresponding Application Code" for a Combined Work means the
+object code and/or source code for the Application, including any data
+and utility programs needed for reproducing the Combined Work from the
+Application, but excluding the System Libraries of the Combined Work.
+
+## 1. Exception to Section 3 of the GNU GPL.
+
+You may convey a covered work under sections 3 and 4 of this License
+without being bound by section 3 of the GNU GPL.
+
+## 2. Conveying Modified Versions.
+
+If you modify a copy of the Library, and, in your modifications, a
+facility refers to a function or data to be supplied by an Application
+that uses the facility (other than as an argument passed when the
+facility is invoked), then you may convey a copy of the modified
+version:
+
+-   a) under this License, provided that you make a good faith effort
+    to ensure that, in the event an Application does not supply the
+    function or data, the facility still operates, and performs
+    whatever part of its purpose remains meaningful, or
+-   b) under the GNU GPL, with none of the additional permissions of
+    this License applicable to that copy.
+
+## 3. Object Code Incorporating Material from Library Header Files.
+
+The object code form of an Application may incorporate material from a
+header file that is part of the Library. You may convey such object
+code under terms of your choice, provided that, if the incorporated
+material is not limited to numerical parameters, data structure
+layouts and accessors, or small macros, inline functions and templates
+(ten or fewer lines in length), you do both of the following:
+
+-   a) Give prominent notice with each copy of the object code that
+    the Library is used in it and that the Library and its use are
+    covered by this License.
+-   b) Accompany the object code with a copy of the GNU GPL and this
+    license document.
+
+## 4. Combined Works.
+
+You may convey a Combined Work under terms of your choice that, taken
+together, effectively do not restrict modification of the portions of
+the Library contained in the Combined Work and reverse engineering for
+debugging such modifications, if you also do each of the following:
+
+-   a) Give prominent notice with each copy of the Combined Work that
+    the Library is used in it and that the Library and its use are
+    covered by this License.
+-   b) Accompany the Combined Work with a copy of the GNU GPL and this
+    license document.
+-   c) For a Combined Work that displays copyright notices during
+    execution, include the copyright notice for the Library among
+    these notices, as well as a reference directing the user to the
+    copies of the GNU GPL and this license document.
+-   d) Do one of the following:
+    -   0) Convey the Minimal Corresponding Source under the terms of
+        this License, and the Corresponding Application Code in a form
+        suitable for, and under terms that permit, the user to
+        recombine or relink the Application with a modified version of
+        the Linked Version to produce a modified Combined Work, in the
+        manner specified by section 6 of the GNU GPL for conveying
+        Corresponding Source.
+    -   1) Use a suitable shared library mechanism for linking with
+        the Library. A suitable mechanism is one that (a) uses at run
+        time a copy of the Library already present on the user's
+        computer system, and (b) will operate properly with a modified
+        version of the Library that is interface-compatible with the
+        Linked Version.
+-   e) Provide Installation Information, but only if you would
+    otherwise be required to provide such information under section 6
+    of the GNU GPL, and only to the extent that such information is
+    necessary to install and execute a modified version of the
+    Combined Work produced by recombining or relinking the Application
+    with a modified version of the Linked Version. (If you use option
+    4d0, the Installation Information must accompany the Minimal
+    Corresponding Source and Corresponding Application Code. If you
+    use option 4d1, you must provide the Installation Information in
+    the manner specified by section 6 of the GNU GPL for conveying
+    Corresponding Source.)
+
+## 5. Combined Libraries.
+
+You may place library facilities that are a work based on the Library
+side by side in a single library together with other library
+facilities that are not Applications and are not covered by this
+License, and convey such a combined library under terms of your
+choice, if you do both of the following:
+
+-   a) Accompany the combined library with a copy of the same work
+    based on the Library, uncombined with any other library
+    facilities, conveyed under the terms of this License.
+-   b) Give prominent notice with the combined library that part of it
+    is a work based on the Library, and explaining where to find the
+    accompanying uncombined form of the same work.
+
+## 6. Revised Versions of the GNU Lesser General Public License.
+
+The Free Software Foundation may publish revised and/or new versions
+of the GNU Lesser General Public License from time to time. Such new
+versions will be similar in spirit to the present version, but may
+differ in detail to address new problems or concerns.
+
+Each version is given a distinguishing version number. If the Library
+as you received it specifies that a certain numbered version of the
+GNU Lesser General Public License "or any later version" applies to
+it, you have the option of following the terms and conditions either
+of that published version or of any later version published by the
+Free Software Foundation. If the Library as you received it does not
+specify a version number of the GNU Lesser General Public License, you
+may choose any version of the GNU Lesser General Public License ever
+published by the Free Software Foundation.
+
+If the Library as you received it specifies that a proxy can decide
+whether future versions of the GNU Lesser General Public License shall
+apply, that proxy's public statement of acceptance of any version is
+permanent authorization for you to choose that version for the
+Library.

package/README.md

@@ -0,0 +1,250 @@
+# Dominds — DevOps Mindsets — Sustainable Agentic Product Lifecycle
+
+> Ship your product with an AI DevOps team that self-improves while it works.
+
+> Agents should be sustaining the continuous development of your products, one-shot product creation is hallucination.
+
+- 中文版：[README.zh.md](./README.zh.md)
+
+## CAVEATS
+
+- **NO WARRANTY / AT YOUR OWN RISK**: Dominds is powerful automation. If you point it at the wrong repo, give it the wrong goal, or trust output blindly, it can waste time or cause damage.
+  - Keep backups and review changes carefully (especially before merging or deploying).
+  - Prefer running on a disposable clone or non-critical branch until you trust your workflow.
+
+- **No Human Permission Mechanism (won’t-have)**: Dominds does not aim to add a per-action “please approve” prompt for tool usage.
+  - The intended safety model is _hard guardrails via workflow/policy_ (planned), not human-in-the-loop popups.
+  - Put high-risk tools behind principled agents so their judgment enforces policies and mitigates risk.
+  - Use least-privilege credentials and avoid production secrets in `.env` when experimenting.
+  - Treat your workspace as sensitive: dialogs, logs, and memory may persist locally on disk.
+
+- **Authorize Your Agents (or don’t use Dominds)**: Dominds is not pair‑programming. Once authorized, agents will act — assume mistakes and review outcomes from a distance.
+  - If you prefer to work closely with your agents, use a more traditional copilot‑style tool.
+  - You’re not a hands‑on driver; you’re fully responsible for the consequences of the team you define and animate — trust them to execute, stay remote, and accept the outcomes, good or bad.
+
+- **Costs / Privacy / Compliance**: using LLM providers can cost money and may send prompts (sometimes including code) to third parties.
+  - Review provider terms, set spending limits, and avoid putting secrets in prompts.
+  - If you need strict privacy/compliance, evaluate self-hosted models and stricter tool policies.
+
+- **Help Yourself (early community)**: Dominds is LGPL, with limited support and a young community.
+  - Expect rough edges, breaking changes, and gaps in docs; issues/PRs are welcome.
+
+## Table of Contents
+
+- [Dominds — DevOps Mindsets — Sustainable Agentic Product Lifecycle](#dominds--devops-mindsets--sustainable-agentic-product-lifecycle)
+  - [Table of Contents](#table-of-contents)
+  - [What is Dominds?](#what-is-dominds)
+  - [Installation](#installation)
+    - [Prerequisites](#prerequisites)
+    - [Install Dominds](#install-dominds)
+    - [Workspace Setup](#workspace-setup)
+  - [Quick Start](#quick-start)
+  - [Start from scratch](#start-from-scratch)
+  - [Core Philosophy](#core-philosophy)
+    - [1) Clear minds, focused tasks](#1-clear-minds-focused-tasks)
+    - [2) Tools with intent: safe by design](#2-tools-with-intent-safe-by-design)
+    - [3) Diverse, self‑improving team from templates](#3-diverse-selfimproving-team-from-templates)
+  - [Documentation](#documentation)
+    - [Getting Help](#getting-help)
+
+## What is Dominds?
+
+Dominds is an AI-powered DevOps framework that creates autonomous agentic teams with persistent memory and self-improving capabilities. Unlike traditional AI assistants, Dominds agents maintain context across conversations, learn from experience, and collaborate as a cohesive team to handle complex development workflows.
+
+**Key Features:**
+
+- **Local-first Runtime**: The orchestration layer runs on your machine and persists team state locally in your workspace. LLM calls still go to your configured provider (so apply normal cost/privacy discipline).
+
+- **Transparent Workspace Integration**: Agent knowledge, personas, lessons learned, and mindset artifacts are stored locally as part of your workspace. This maximizes transparency to humans and enables version control alongside your product code, creating a complete historical record of both human and AI contributions to your project.
+
+- **Persistent Team Memory**: Agents maintain context across conversations, remember past decisions, and build upon learned patterns. Teams evolve their practices and knowledge over time, with all insights preserved in your workspace.
+
+- **Collaborative Intelligence**: Multiple specialized agents work together on shared tasks, combining their unique capabilities while maintaining clear communication channels and shared understanding through workspace-stored context.
+
+- **Safe Tool Access**: AI-driven guardrails provide secure access to powerful development tools, with all safety policies and access patterns transparently stored and version-controlled within your workspace.
+
+- **Task-Focused Architecture**: Clear, bounded contexts prevent cognitive overload while maintaining rich inter-agent communication and knowledge sharing through the local workspace knowledge base.
+
+## Installation
+
+### Prerequisites
+
+- **Node.js (with npm bundled)**: Version 22.x or later
+- **LLM provider configured for your team**: Dominds ships with a built-in provider catalog [main/llm/defaults.yaml](./main/llm/defaults.yaml) including Codex (ChatGPT) and Anthropic, plus several Anthropic-compatible endpoints (e.g. MiniMax, Z.ai, BigModel). You’ll need valid credentials for at least one provider.
+- **pnpm (optional)**: Recommended only if you’re developing Dominds itself.
+
+### Install Dominds
+
+```bash
+# Global installation (recommended)
+npm install -g dominds
+# (or) pnpm add -g dominds
+
+# Verify installation
+dominds --help
+```
+
+You can also run Dominds with **no global install** (useful for quick trials / CI / keeping your machine clean):
+
+```bash
+# Run the latest release (downloads to npm cache, then executes)
+npx -y dominds@latest --help
+npx -y dominds@latest
+
+# Or pin an exact version (more reproducible)
+# npx -y dominds@1.2.3
+```
+
+Notes:
+
+- `npx dominds` does **not** necessarily mean “always latest”: if your project already has `dominds` installed, or your npm cache already contains an older version, `npx` may reuse that.
+- If you want “resolve whatever `latest` points to”, use `dominds@latest` explicitly. It may still reuse the cached copy of that same version (expected).
+
+For development and any sort of open source contribution, use the in-tree dev wrapper workspace:
+
+https://github.com/longrun-ai/dominds-feat-dev
+
+1. Clone dominds-feat-dev
+2. Clone your dominds fork into dominds-feat-dev/dominds/
+3. Open PRs against [longrun-ai/dominds](https://github.com/longrun-ai/dominds) from that inner repo
+
+### Workspace Setup
+
+There are two common ways to create a workspace:
+
+- **Recommended**: start from a scaffold/template (see [Quick Start](#quick-start)).
+- **Minimal**: start from an empty folder (see [Start from scratch](#start-from-scratch)).
+
+Dominds uses your current working directory as the runtime workspace (rtws). When you start `dominds`, the WebUI will automatically redirect you to `http://localhost:5666/setup` if the workspace is missing required configuration (for example `.minds/team.yaml` or provider env vars).
+
+Note: In production mode, Dominds enables a local shared-secret auth key by default, so the browser may open with a URL containing `?auth=...`. Treat that token as sensitive.
+
+**Template creation (recommended)**:
+
+```bash
+# Use official scaffold templates (default base = https://github.com/longrun-ai)
+dominds create web-scaffold my-project
+
+# Use custom organization templates
+export DOMINDS_TEMPLATE_BASE="https://github.com/myorg"
+dominds create web-scaffold my-project
+
+# Or use full URLs for any template
+dominds create https://github.com/myorg/custom-template.git my-project
+```
+
+**Template Resolution:**
+
+- Short names like `web-scaffold` resolve to `${DOMINDS_TEMPLATE_BASE}/web-scaffold.git`
+- Default `DOMINDS_TEMPLATE_BASE` is `https://github.com/longrun-ai`
+- Set your own `DOMINDS_TEMPLATE_BASE` for organization-specific templates
+
+For more template options, see the [CLI Usage Guide](docs/cli-usage.md#workspace-creation).
+
+## Quick Start
+
+```bash
+# 1) Create a workspace from a scaffold template
+dominds create web-scaffold my-project
+cd my-project
+
+# 2) Start the WebUI (opens a browser by default)
+dominds
+```
+
+Then:
+
+1. Your browser should land on `http://localhost:5666/setup` (either directly, or via an automatic redirect).
+2. In **Setup**, pick a provider + model. If your template didn’t create `.minds/team.yaml`, use Setup to create/overwrite it (this writes a minimal `member_defaults` config).
+3. Still in **Setup**, set the required provider env var (the name comes from the provider catalog, e.g. `OPENAI_API_KEY`, `ANTHROPIC_API_KEY`, `CODEX_HOME`, etc.).
+   - The setup UI can write it into `~/.zshrc` / `~/.bashrc` in a managed block, and also applies it to the running server process immediately.
+4. Click **Go to App**, create a dialog, and start working.
+
+## Start from scratch
+
+Starting from scratch means: create an empty folder, run `dominds`, let Setup generate the minimal `.minds/team.yaml`, then use the shadow team manager (`@fuxi`) to design a real team for your product.
+
+```bash
+mkdir my-workspace
+cd my-workspace
+
+# Starts the WebUI server and opens a browser (default port: 5666)
+dominds
+```
+
+1. The WebUI should redirect you to `http://localhost:5666/setup` because the workspace has no `.minds/team.yaml` yet.
+2. In **Setup**:
+   - Select a provider + model and click **Create `.minds/team.yaml`**.
+   - Provide the required provider env var and write it to your shell rc (or set it manually). Setup applies it immediately to the running server.
+   - Click **Go to App**.
+3. In the app, create a new dialog with **shadow member** `@fuxi` (Fuxi is hidden by default; use the “Shadow members” picker in the dialog creation modal). Until you add visible members, you’ll mostly work through shadow members.
+4. Tell `@fuxi` your product idea and ask it to propose and apply a suitable agentic team configuration by updating `.minds/team.yaml` (Fuxi has the `team-mgmt` toolset scoped to `.minds/**`).
+
+## Core Philosophy
+
+Dominds is designed for long-running product development and operations, with a “division of labor” constitution:
+
+- Reduce **agent mental overhead** by dividing and distributing work across a team of specialized agents, and by using more effective **dialog round control** (instead of brittle context condensation) to shed conversational/tool-call noise when a dialog becomes cluttered.
+- Prevent **tool misuse** by putting side-effectful tools only in the hands of well-prompted specialist agents, via least-privilege toolsets/policies.
+- Prevent **staleness** (agents acting on outdated assumptions) by encoding decisions, conventions, configuration, and key repo facts into version-tracked workspace artifacts intended to be loaded into agent context (team-shared and agent-individual memory, task docs, and dialog-local reminders), rather than leaving them as “undiscovered” knowledge scattered across the repo.
+
+### 1) Clear minds, focused tasks
+
+> Split big initiatives into bounded tasks. Each agent should operate with a clear, minimal context; everything else is noise.
+
+- Smaller context reduces hallucination risk and error rates.
+- Bounded scopes make evaluation and iteration faster.
+- Clear task contracts enable parallelism and reuse.
+
+### 2) Tools with intent: safe by design
+
+> Side-effectful tools are powerful, yet dangerous. Safety comes from intent-aware guardrails; vague “be careful” prompts don’t scale, and neither does a human doing constant micro-audits.
+
+Principles
+
+- Least privilege by default; expand access only when justified.
+- For dangerous tools, require per-call intent justification:
+  - Will it cause unexpected side effects?
+  - Could it be misused (even “nerfed”)?
+  - Does it create an “evil leverage” path?
+  - Is there a safer/minimal alternative?
+  - Pattern-based command matching is not enough; safety comes from disciplined specialist agents enforcing policy with judgment.
+- Keep tool grants explicit and reviewable (toolsets, per-agent tool lists, and workspace access scopes).
+
+Result: fewer bad side effects, higher plan fidelity, and more first‑try successes.
+
+### 3) Diverse, self‑improving team from templates
+
+> Start from a template team, then curate and evolve both individual and shared memory as versioned artifacts.
+
+- Bootstrap from a domain-specific scaffold (team + workflows).
+- Keep individual and shared memory in `.minds/**` (personas, lessons, playbooks).
+- Evolve mindsets and governance as the product evolves.
+- Keep it transparent to humans: plain Markdown, version-tracked.
+
+## Documentation
+
+- **[Terminology](docs/dominds-terminology.md)** — Canonical vocabulary across prompts and implementation docs (Taskdoc, Teammate Tellask, etc.)
+- **[CLI Usage Guide](docs/cli-usage.md)** — Commands, options, and usage patterns
+- **[MCP Support](docs/mcp-support.md)** — MCP tool integration
+- **[Encapsulated Taskdocs](docs/encapsulated-taskdoc.md)** — `*.tsk/` Taskdoc packages and parsing
+- **[Context Health](docs/context-health.md)** — Measuring/maintaining context quality
+- **[Keep-going](docs/keep-going.md)** — Auto-continue (diligence) mechanism
+- **[Design](docs/design.md)** — Architecture and key abstractions
+- **[Mottos](docs/mottos.md)** — Short guiding statements
+- **[OEC Philosophy](docs/OEC-philosophy.md)** — Philosophy and safety stance
+
+- **[Team Mgmt Toolset](docs/team-mgmt-toolset.md)** — Managing team members via tools
+- **[i18n](docs/i18n.md)** — Language and localization rules
+- **[Dialog System](docs/dialog-system.md)** — Dialog runtime model and streaming
+- **[Dialog Persistence](docs/dialog-persistence.md)** — On-disk layout and lifecycle
+- **[Interruption & Resumption](docs/interruption-resumption.md)** — Stop/resume semantics
+- **[Auth](docs/auth.md)** — Authentication and access model
+
+### Getting Help
+
+- Open an issue on [GitHub](https://github.com/longrun-ai/dominds/issues) for bugs or feature requests
+
+---
+
+**License:** [LGPL](./LICENSE) | **Repository:** [github.com/longrun-ai/dominds](https://github.com/longrun-ai/dominds)

package/README.zh.md

@@ -0,0 +1,161 @@
+# Dominds（斟发作们）—— 斟酌开发运作的智能体们
+
+> 以一支具备自我迭代能力的 AI DevOps 团队，实现产品的持续交付。
+
+- English: [README.md](./README.md)
+
+## 重要提示（CAVEATS）
+
+- **无担保 / 风险自担**：Dominds 是一套高性能自动化框架。若配置不当、目标模糊或盲目采信输出结果，可能导致时间损耗或代码损坏。
+  - 合并代码或发布版本前，务必核查所有修改；建议先在测试分支或克隆仓库中试运行验证。
+- **成本 / 隐私 / 合规性**：调用大语言模型（LLM）供应商服务会产生费用，且提示词（部分场景含代码片段）可能传输至第三方。
+  - 请仔细阅读服务条款，设置费用额度上限，切勿将密钥写入提示词或日志文件。
+- **授权即执行**：Dominds 并非“辅助工具”，一旦获得授权，智能体将直接执行操作，你需对最终结果承担全部责任。
+- **项目处于早期阶段**：目前存在功能边界不完善、破坏性更新及文档缺失等问题，欢迎提交 issue 或 PR 参与优化。
+
+## 关于 Dominds
+
+Dominds 是一款面向开发运作（DevOps）场景的智能体框架，它将“团队配置、记忆存储、流程规范、工具权限”整合至统一工作区，使多智能体团队能够长期稳定运行、持续学习，并在明确边界内高效执行任务。
+
+其核心定位并非一次性完成产品开发，而是聚焦“持续交付”的全周期流程，涵盖任务拆分、职责分工、决策可追溯、工具使用管控及上下文健康度维护等关键环节。
+
+## 安装指南
+
+### 环境要求
+
+- **Node.js（含 npm）**：版本 22.x 及以上
+- **至少一个可用的 LLM 服务提供商**：Dominds 内置提供商目录（路径：[main/llm/defaults.yaml](./main/llm/defaults.yaml)），需为其中至少一个提供商配置有效凭证（通过环境变量设置）。
+- **pnpm（可选）**：仅在开发 Dominds 本体时推荐安装。
+
+### 安装 Dominds
+
+```bash
+# 推荐全局安装
+npm install -g dominds
+# 或使用 pnpm 安装
+# pnpm add -g dominds
+
+# 查看帮助文档
+dominds --help
+```
+
+也可以选择 **无需全局安装** 的方式运行（适合临时试用 / CI / 不想污染环境）：
+
+```bash
+# 运行最新版（会临时下载到 npm cache 并执行）
+npx -y dominds@latest --help
+npx -y dominds@latest
+
+# 或固定到某个版本（更利于复现）
+# npx -y dominds@1.2.3
+```
+
+说明：
+
+- 仅写 `npx dominds` **不一定**会自动跑“最新版”：如果你本地项目已安装了 `dominds`，或 npm cache 已存在旧版本，`npx` 可能会直接复用已有版本。
+- 想要“每次都按 latest 标签解析版本”，请显式写 `dominds@latest`；它仍可能复用 cache 中已下载的同版本（属于正常行为）。
+
+### 开发与贡献（树内工作流 / in-tree 工作流）
+
+若你计划为 Dominds 开源项目贡献代码，建议使用树内包装器（in-tree wrapper）工作区：
+
+https://github.com/longrun-ai/dominds-feat-dev
+
+1. 克隆 dominds-feat-dev 仓库
+2. 在 dominds-feat-dev/dominds/ 目录下，克隆你自己的 Dominds fork 仓库
+3. 在内层 dominds/ 目录中，向 longrun-ai/dominds 提交 PR
+
+## 快速上手（推荐：通过模板创建工作区）
+
+```bash
+# 1) 基于模板创建工作区
+dominds create web-scaffold my-project
+cd my-project
+
+# 2) 启动 WebUI（默认自动打开浏览器，默认端口 5666）
+dominds
+```
+
+首次启动后，通常会自动跳转至 `http://localhost:5666/setup` 配置页面，按以下步骤操作：
+
+1. 选择提供商（provider）及模型（model），创建或覆盖 `.minds/team.yaml`（最小化可运行配置文件）。
+2. 根据提示设置提供商所需的环境变量（配置页可协助将变量写入 `~/.zshrc` 或 `~/.bashrc` 的托管区块）。
+3. 进入主界面，创建对话即可开始工作。
+
+## 从零开始（空文件夹启动）
+
+若暂无模板或团队配置，可直接从空目录启动：
+
+```bash
+mkdir my-workspace
+cd my-workspace
+dominds
+```
+
+完成初始化配置后，建议先与影子团队管理者 “伏羲”(`@fuxi`) 对话，由其根据你的产品需求，搭建适配的团队配置（例如生成完整的 `.minds/team.yaml` 文件，分配成员职责及工具权限）。
+
+## 核心理念（Core Philosophy）
+
+Dominds 面向“长期开发运作（DevOps）”场景设计，基于社会化分工理念，所有功能均围绕以下三大核心目标展开：
+
+- 降低 **智能体心智负担**：将工作分配给不同专长的智能体成员，通过“对话分轮、噪音清理”等机制减少上下文污染（而非依赖传统上比较勉强的“上下文压缩”手段）。
+- 防范 **工具滥用风险**：将软件工具交给专职智能体把关使用，确保牠们不受任务上下文影响，能够专心审核使用合规性，以及负责工具的正确使用。
+- 避免 **信息过时问题**：将关键决策、约定及配置，以可版本追踪的工作区文件形式，体现为智能体上下文的直接内容（团队以及个人记忆，差遣牒，对话附加提醒项），而非散落在仓库中等着智能体自己去找。
+
+### 1) 清心定界：上下文清爽、任务明确
+
+> 将宏大目标拆解为边界清晰的细分任务，每个智能体成员仅获取完成当前任务所需的最小上下文。
+
+- 上下文范围越小，智能体产生幻觉及误操作的概率越低
+- 任务边界越清晰，结果评估与迭代优化的效率越高
+- 任务契约越明确，越容易实现并行处理与流程复用
+
+### 2) 意图导向的工具调用：安全的核心是“先判断再约束”
+
+> 强力赋能的工具一般都伴随使用风险，安全保障必须来自基于意图的判断与约束，经常提醒智能体“要小心谨慎”，或者靠人来把关每个小细节，都是反生产力提升的。
+
+核心原则：
+
+- 默认赋予最小权限，仅在必要时扩展权限范围
+- 针对危险工具，每次使用时由专职智能体进行确认：
+  - 会产生意外副作用吗？
+  - 不小心用错了？(防呆)
+  - 被恶意滥用了？
+  - 是否有更安全/更简单的替代手段？
+  - 常规文本匹配对这些问题是力不从心的；更可靠的是由纪律严明的专职智能体做出判断，确保合规。
+- 工具授权需满足可读取、可审计、可追溯（含工具集、成员工具清单、目录读写范围等）
+
+### 3) 基于领域专有模板：多元团队，自我进化
+
+> 基于模板快速搭建团队，将个体及团队记忆作为版本追踪的数字资产，持续迭代优化。
+
+- 通过领域模板引导团队职责划分与工作流搭建
+- 将角色设定（personas）、经验教训（lessons）、操作手册（playbooks）存入 `.minds/**` 目录，对于人类完全透明，方便阅读审核
+- 伴随产品迭代，同步升级治理模式与团队结构
+
+## 相关文档
+
+- **[Terminology](docs/dominds-terminology.md)** — Dominds 专有术语
+- **[CLI Usage Guide](docs/cli-usage.md)** — 命令行工具及使用方法
+- **[MCP Support](docs/mcp-support.md)** — MCP 工具集成指南
+- **[Encapsulated Taskdocs](docs/encapsulated-taskdoc.md)** — 差遣牒（Taskdoc）封装
+- **[Context Health](docs/context-health.md)** — 上下文健康维护
+- **[Keep-going](docs/keep-going.md)** — 鞭策机制
+- **[Design](docs/design.md)** — Dominds 架构设计与核心抽象
+- **[Mottos](docs/mottos.md)** — 警世名言
+- **[OEC Philosophy](docs/OEC-philosophy.md)** — 全方位优化管理法
+
+- **[Team Mgmt Toolset](docs/team-mgmt-toolset.md)** — 团队管理工具集
+- **[i18n](docs/i18n.md)** — 语言支持与本地化配置
+- **[Dialog System](docs/dialog-system.md)** — 对话系统与流式事件
+- **[Dialog Persistence](docs/dialog-persistence.md)** — 磁盘持久化存储结构
+- **[Interruption & Resumption](docs/interruption-resumption.md)** — 任务中断与恢复语义
+- **[Auth](docs/auth.md)** — 认证机制与访问模型（WebUI + API）
+
+## 获取帮助
+
+如需协助，可在 GitHub 提交 issue：[https://github.com/longrun-ai/dominds/issues](https://github.com/longrun-ai/dominds/issues)
+
+---
+
+**License:**[LGPL](./LICENSE) | **Repository:**[https://github.com/longrun-ai/dominds](https://github.com/longrun-ai/dominds)