By ZTE-AICloud
Reverse-engineer codebases to extract architecture, interfaces, entities, scenarios, requirements, and rules, generating standardized SDD documentation in JSON and Markdown. Supports full-process or element-specific reverse engineering modes.
当您需要识别代码库的整体架构结构,包括分层架构、业务领域划分、关键模块和功能层次时,请使用此代理。此代理应在需要进行架构识别分析时触发。示例:<example> 上下文:用户需要对代码库进行架构识别分析。 用户:"请识别该项目的架构结构" 助手:"我将使用Task工具启动architecture-identifier代理来进行架构识别分析。" <commentary> 由于用户请求进行架构识别分析,请使用architecture-identifier代理来识别代码库的整体架构结构。 </commentary> </example>
当您需要分析代码库中的接口入口和调用链,识别调用链归属的模块层级时,请使用此代理。此代理应在需要进行调用链分析时触发。示例:<example> 上下文:用户需要分析代码库的调用链关系。 用户:"请分析该项目的调用链" 助手:"我将使用Task工具启动call-chain-analyzer代理来进行调用链分析。" <commentary> 由于用户请求进行调用链分析,请使用call-chain-analyzer代理来分析接口入口和调用链关系。 </commentary> </example>
当您需要深度识别代码库的整体架构结构(模块分层、模块边界、模块依赖关系等)并按模板输出深度架构识别报告时,请使用此代理。此代理应在需要进行深度架构识别分析时触发。示例:<example> 上下文:用户需要对代码库进行深度架构识别分析。 用户:"请深度识别该项目的架构结构和模块关系" 助手:"我将使用Task工具启动deep-architecture-identifier代理来进行深度架构识别分析。" <commentary> 由于用户请求进行深度架构识别分析,请使用deep-architecture-identifier代理来识别代码库的深度架构结构、模块边界和模块关系。 </commentary> </example>
当您需要从指定的接口批次文件中分析接口并生成相应的详细文档时,请使用此代理。此代理应在接口批次准备就绪且需要详细分析时触发。示例:<example> 上下文:用户已准备接口批次文件进行详细分析并希望处理它们。 用户:"处理第1批接口的详细信息" 助手:"我将使用Task工具启动interface-analyzer代理来分析第1批中的接口并生成详细文档。" <commentary> 由于用户请求对接口批次进行详细分析,请使用interface-analyzer代理处理接口批次文件并生成详细文档。 </commentary> </example>
当您需要从指定的批处理文件中识别接口并生成相应的JSON输出文件时,请使用此代理。此代理应在批处理准备就绪且需要接口识别时触发。示例:<example> 上下文:用户已准备批处理文件进行接口分析并希望处理它们。 用户:"处理第5批以进行接口识别" 助手:"我将使用Task工具启动interface-recognizer代理来识别第5批中的接口。" <commentary> 由于用户请求对特定批次进行接口识别,请使用interface-recognizer代理处理批处理文件并生成接口列表。 </commentary> </example>
全流程反构编排 Skill。当用户执行 reverse --target all 时,自动编排执行逻辑架构、接口、功能、实体、场景、需求、外部接口、规则共 8 个阶段,生成完整的代码库逆向工程文档。适用于:代码库分析、知识提取、架构文档生成、存量代码逆向等场景。
从代码库自动识别并生成深度逻辑架构文档,提取核心业务逻辑层、数据流模式、模块依赖关系和决策逻辑集中点,输出 logic_architecture.md。适用于代码架构分析、遗留系统改造、技术尽职调查等场景。当 reverse 的 --target 为 deep_logic_architecture 时触发。
实体清单与实体关系反构的编排Skill. 当 reverse 的 --target 为 entities 或 all 的实体阶段时触发.
- **职责**:识别代码库中来自**外部模块**且在本仓库中有调用示例的接口,并生成标准化文档。
功能清单与功能详情文档反构编排Skill。当用户提到"功能反构"、"reverse functions"、或执行 `reverse --target functions` 时自动触发。编排5个阶段:阶段0缓存检查、阶段1项目入口识别、阶段2场景识别、阶段3功能划分、阶段4功能文档生成。支持批量并发处理和缓存复用。
Uses power tools
Uses Bash, Write, or Edit tools
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Build high-quality software faster with Spec-Driven Development.
Co-OmniSpec is an open-source toolkit that turns intent into structured specifications, design artefacts, and dependency-aware tasks, and then drives an AI coding agent through implementation. It is a localized and extended adaptation of GitHub Spec Kit, optimised for Claude Code and other agent environments.
Co-OmniSpec ships as two plugins from the CoMind-plugins marketplace: omni-dsdd for the
Spec-Driven Development pipeline and omni-reverse for the reverse-engineering side. Both
must be installed together.
中文说明 — 请参阅 README-zh-CN.md。
Spec-Driven Development flips the traditional code-first workflow: specifications drive implementation. You write the what and why in natural language; your AI agent translates them into structured specifications, design notes, dependency-aware tasks, and finally code.
Co-OmniSpec provides:
constitution → specify → clarify → design → tasks → implement → archive.Co-OmniSpec is two plugins in one marketplace. They share runtime assets but each focuses on half of the workflow:
| Plugin | Marketplace name | Focus | Plugin-local skills (highlights) | Shared assets |
|---|---|---|---|---|
omni-dsdd | CoMind-plugins | The Spec-Driven Development pipeline: harness, slash commands, shared templates, scripts, workflow YAML. | specify, clarify, design, tasks, implement, archive, analyze, checklist, constitution, routing, workflow-orchestrator, knowledge-retrieval, create-branch, sdd, and the on-demand reverse-on-demand. | Owns .omni-infra/ (config, memory, metamodel, scripts, templates). |
omni-reverse | CoMind-plugins | Reverse engineering of existing code into SDD artefacts. | reverse, reverse-orchestration, reverse-shared, reverse-logic-architecture, reverse-deep-logic-architecture, reverse-interfaces, reverse-external-interfaces, reverse-functions, reverse-entities, reverse-scenarios, reverse-requirements, reverse-rules. | Depends on omni-dsdd; resolves ${DSDD} through scripts/resolve-dsdd-root.{sh, py, ps1}. |
Both plugins must be installed side by side. omni-reverse cannot run without omni-dsdd; the
DSDD pipeline benefits from omni-reverse whenever the project has existing code to document.
| Requirement | Notes |
|---|---|
| Claude Code | code.claude.com with plugin marketplace support. |
| Git | Used for feature branches and the changes/ directory. |
| Bash or PowerShell | Required to execute the shared scripts under .omni-infra/scripts/. |
| Node + pnpm (optional) | Needed only when validating the plugin's own skills locally (pnpm validate inside omni-dsdd/); see CONTRIBUTING.md. |
Add the marketplace and install both plugins:
/plugin marketplace add ZTE-AICloud/Co-OmniSpec
/plugin install omni-dsdd@CoMind-plugins
/plugin install omni-reverse@CoMind-plugins
The first command is also accepted as /market add ZTE-AICloud/Co-OmniSpec.
Equivalent CLI form (for non-interactive shells or CI):
claude plugin marketplace add ZTE-AICloud/Co-OmniSpec
claude plugin install omni-dsdd@CoMind-plugins
claude plugin install omni-reverse@CoMind-plugins
Verify:
/plugin marketplace list (or /market list) — CoMind-plugins is registered./plugin (Discover) — both omni-dsdd and omni-reverse are listed and enabled./constitution, /specify, /reverse,
/routing are visible.npx claudepluginhub zte-aicloud/co-omnispec --plugin omni-reverse面向规范驱动开发(DSDD)的Harness工程解决方案插件包,覆盖按需求反构、规约制定、需求分析、方案设计、任务拆解、实施与归档全流程,通过多种工作流引导AI编程助手高效交付高质量软件。
面向规范驱动开发(DSDD)的Harness工程解决方案插件包,覆盖按需求反构、规约制定、需求分析、方案设计、任务拆解、实施与归档全流程,通过多种工作流引导AI编程助手高效交付高质量软件。
Spec-Driven Development: ADRs, specifications, sprint planning, parallel implementation, code review, and documentation generation.
Reverse engineer clean behavioral specs from any codebase
Spec2Ship: AI-assisted development framework for the full software lifecycle - from specifications to shipping
Three complementary AI agent skills: (1) Spec-Driven Develop — automates full development workflow for large-scale complex tasks: project analysis, task decomposition, project governance, progress tracking, and execution; (2) Deep Discuss — structured deep discussion for problem analysis, brainstorming, and solution design; (3) Review SPD — findings-first code review for uncommitted changes, date-range commits, and branch/PR diffs.
Architecture Deep Research — scan a repo, draft a PRD, run live deep research, get a cited handoff. Use when picking a topology, retrieval architecture, event bus, storage engine, auth pattern, or any system-design decision before code lands.