[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-76262":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":12,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":14,"stars7d":15,"stars30d":12,"stars90d":14,"forks30d":14,"starsTrendScore":14,"compositeScore":16,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":17,"fork":17,"defaultBranch":18,"hasWiki":19,"hasPages":17,"topics":20,"createdAt":9,"pushedAt":9,"updatedAt":21,"readmeContent":22,"aiSummary":23,"trendingCount":14,"starSnapshotCount":14,"syncStatus":24,"lastSyncTime":25,"discoverSource":26},76262,"Coding_Corgi_flow","ricoyudog\u002FCoding_Corgi_flow","ricoyudog","OpenSpec GitFlow — structured AI engineering workflows with issue tracking",null,"TypeScript",110,8,1,0,4,42.66,false,"master",true,[],"2026-06-12 04:01:21","**English** | [繁體中文](README.zh-TW.md)\n\n# 🐕 Coding Corgi Flow\n\n> **Your AI pipeline, structured.**  \n> A workflow toolkit that turns any AI coding assistant into a disciplined engineering partner — proposal to archive, tracked and reviewable.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fdeveloper_tools_banner.png\" alt=\"Coding Corgi Flow — Your AI pipeline, structured\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n---\n\n## 🐾 Before & After\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"50%\">\u003Cb>😫 Without Corgi\u003C\u002Fb>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"50%\">\u003Cb>🐕 With Corgi Flow\u003C\u002Fb>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg src=\"docs\u002Farticles\u002Fcorgi_developer_chaos.png\" alt=\"AI coding chaos without workflow management\"\u002F>\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"docs\u002Farticles\u002Fcorgi_developer_confident.png\" alt=\"Structured AI coding with Coding Corgi Flow\"\u002F>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\">No pipeline. No tracking.\u003Cbr\u002F>Code spaghetti. Repeated mistakes.\u003C\u002Ftd>\n    \u003Ctd align=\"center\">Schema-driven planning. Issue tracking.\u003Cbr\u002F>Checkpoint execution. 5-axis review.\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## 🗺️ The Pipeline\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fcorgi_journey_illustration.png\" alt=\"Corgi journey: Propose → Apply → Verify → Review → Archive\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n\u003Cdetails>\n\u003Csummary>Precise diagram (Mermaid)\u003C\u002Fsummary>\n\n```mermaid\nflowchart LR\n    A[\"\u002Fcorgi-propose\"] --> B[\"proposal.md\\nspecs\u002F\\ndesign.md\\ntasks.md\"]\n    B --> C[\"Issues\\n(parent + children)\"]\n    C --> D[\"\u002Fcorgi-apply\"]\n    D --> E{\"Group done?\"}\n    E -->|Yes| F[\"\u002Fcorgi-verify\"]\n    F --> G{\"Pass?\"}\n    G -->|No| D\n    G -->|Yes| H[\"\u002Fcorgi-review\"]\n    H --> I{\"Approved?\"}\n    I -->|Yes, more groups| D\n    I -->|Rejected| J[\"Fix tasks added\"]\n    J --> D\n    I -->|All done| K[\"\u002Fcorgi-archive\"]\n```\n\n\u003C\u002Fdetails>\n\n---\n\n## 🔧 What This Is\n\nCoding Corgi Flow is the **community extension** of [OpenSpec](https:\u002F\u002Fgithub.com\u002FFission-AI\u002FOpenSpec) by [Fission AI](https:\u002F\u002Fgithub.com\u002FFission-AI). We layer custom schemas, AI skills, and CLI tooling on top of OpenSpec's core artifact pipeline to add what real teams need:\n\n| Superpower | Why you need it |\n|---|---|\n| 📌 **Automatic Issue Tracking** | Parent + child issues on GitLab or GitHub, labels synced |\n| 🛑 **Checkpoint-based Apply** | One Task Group at a time — never lose control of your AI |\n| ✅ **Automated Verify Gate** | Lint, build, tests, spec coverage — blocks review on failure |\n| 🔍 **5-Axis Review** | Architecture · Security · Performance · Quality · Completeness |\n| 🧠 **Cross-Session Memory** | 3-layer system — your AI remembers across sessions (≤3000 tokens at startup) |\n| 🌿 **Worktree Isolation** | Parallel changes, each in its own git worktree (opt-in) |\n| 🧩 **Composable Skills** | Atoms → Molecules → Compounds with validated metadata |\n| 📦 **One-command Install** | `npm i -g corgispec` → `corgispec bootstrap` → done |\n\nIt ships as an npm CLI (`corgispec`), a Claude Code \u002F Codex plugin, and a set of slash commands for OpenCode, Claude Code, and Codex.\n\n---\n\n## 🚀 Quick Start\n\n### Prerequisites\n\n- **Node.js 18+**\n- **An LLM Agent** — OpenCode, Claude Code, Cursor, AmpCode, etc.\n- **`gh` CLI** (for GitHub) or **`glab` CLI** (for GitLab)\n\n### Install & Bootstrap\n\nChoose your path:\n\n**A. npm (recommended)**\n\n```bash\nnpm install -g corgispec\ncorgispec bootstrap --path \u002Fpath\u002Fto\u002Fyour-project --schema github-tracked\n```\n\n**B. Claude Code \u002F Codex Plugin**\n\n```text\n# Claude Code\n\u002Fplugin marketplace add ricoyudog\u002FCoding_Corgi_flow\n\u002Fplugin install corgispec@corgispec\n\n# Codex\ncodex plugin install corgispec\n```\n\n**C. Bootstrap via AI Agent**\n\nPaste this into your agent:\n\n```text\nFetch and follow instructions from https:\u002F\u002Fraw.githubusercontent.com\u002Fricoyudog\u002FCoding_Corgi_flow\u002Fmain\u002F.opencode\u002FINSTALL.md\n```\n\n### Initialize Memory (recommended)\n\n```text\n# OpenCode\n\u002Fcorgi-memory-init\n\n# Claude Code\n\u002Fcorgi:memory-init\n```\n\n### Start Building\n\n```text\n# OpenCode\n\u002Fcorgi-propose Add user authentication with JWT and refresh tokens\n\n# Claude Code\n\u002Fcorgi:propose Add user authentication with JWT and refresh tokens\n```\n\nThen: `apply` → `verify` → `review` → `archive`. One Task Group at a time.\n\n---\n\n## 🎮 Commands\n\n| Command | What it does |\n|---|---|\n| `\u002Fcorgi-propose` | Generate planning artifacts (proposal, specs, design, tasks) + create issues |\n| `\u002Fcorgi-apply` | Execute one Task Group, sync closeout, pause for review |\n| `\u002Fcorgi-verify` | Automated quality gate — lint, build, tests, spec coverage |\n| `\u002Fcorgi-review` | 5-axis review with evidence gathering, approve\u002Freject\u002Fdiscuss |\n| `\u002Fcorgi-archive` | Close issues, sync delta specs, extract knowledge, cleanup |\n| `\u002Fcorgi-explore` | Thinking partner — explore ideas, clarify requirements |\n| `\u002Fcorgi-install` | Project-local asset install, update, or verify |\n| `\u002Fcorgi-memory-init` | Initialize 3-layer memory (`memory\u002F` + `wiki\u002F`) |\n| `\u002Fcorgi-migrate` | Import existing knowledge into memory\u002Fwiki |\n| `\u002Fcorgi-lint` | 11-check memory health validation |\n| `\u002Fcorgi-ask` | Answer questions from the vault with budget-aware retrieval |\n\n> Claude Code uses `\u002Fcorgi:\u003Ccommand>` syntax (e.g., `\u002Fcorgi:propose`). Platform auto-detected from `config.yaml`.\n\n---\n\n## ✨ Feature Showcase\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd width=\"50%\">\n      \u003Cb>📋 Checkpoint-based Apply\u003C\u002Fb>\u003Cbr\u002F>\n      One Task Group at a time, pauses for review — never lose control.\n      \u003Cbr\u002F>\u003Cbr\u002F>\n      \u003Cimg src=\"docs\u002Farticles\u002Fimages\u002Fgroup check point.png\" alt=\"Checkpoint-based apply\" width=\"100%\"\u002F>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"50%\">\n      \u003Cb>📌 Automatic Issue Tracking\u003C\u002Fb>\u003Cbr\u002F>\n      Parent + child issues on GitLab or GitHub, labels synced automatically.\n      \u003Cbr\u002F>\u003Cbr\u002F>\n      \u003Cimg src=\"docs\u002Farticles\u002Fimages\u002Fissue_board_example.png\" alt=\"Issue board\" width=\"100%\"\u002F>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd>\n      \u003Cb>✅ Task Management\u003C\u002Fb>\u003Cbr\u002F>\n      Tasks broken into groups with clear checklist tracking.\n      \u003Cbr\u002F>\u003Cbr\u002F>\n      \u003Cimg src=\"docs\u002Farticles\u002Fimages\u002Ftask_list.png\" alt=\"Task list\" width=\"100%\"\u002F>\n    \u003C\u002Ftd>\n    \u003Ctd>\n      \u003Cb>🔍 5-Axis Review\u003C\u002Fb>\u003Cbr\u002F>\n      Architecture · Security · Performance · Quality · Completeness.\n      \u003Cbr\u002F>\u003Cbr\u002F>\n      \u003Cimg src=\"docs\u002Farticles\u002Fimages\u002Fissue_card_example.png\" alt=\"Review card\" width=\"100%\"\u002F>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n---\n\n## 🧠 Cross-Session Memory\n\nAI sessions are stateless by default. Corgi Flow adds a **3-layer memory system** that persists knowledge across sessions — ≤2900 tokens at startup, self-compacting, Obsidian-compatible.\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fcorgi_knowledge_vault.png\" alt=\"3-layer memory system\" width=\"80%\"\u002F>\n\u003C\u002Fp>\n\n\u003Cdetails>\n\u003Csummary>Precise diagram (Mermaid)\u003C\u002Fsummary>\n\n```mermaid\nflowchart LR\n    subgraph \"Layer 1: memory\u002F (always loaded)\"\n        A[\"MEMORY.md\"] --- B[\"session-bridge.md\"] --- C[\"pitfalls.md\"]\n    end\n    subgraph \"Layer 2: wiki\u002F (on-demand)\"\n        D[\"hot.md\"] --- E[\"index.md\"] --- F[\"patterns\u002F sessions\u002F decisions\u002F ...\"]\n    end\n    subgraph \"Layer 3: docs\u002F (untouched)\"\n        G[\"existing docs\"]\n    end\n\n    B -.->|\"startup read\"| D\n    D -.->|\"navigate\"| E\n    E -.->|\"wikilinks\"| F\n    F -.->|\"references\"| G\n```\n\n\u003C\u002Fdetails>\n\n> 📸 See it in action: ![](docs\u002Farticles\u002Fimages\u002Fobisidian_wiki_example.png)\n\n| Scenario | Command |\n|---|---|\n| New project | Paste Quick Start prompt → `corgispec bootstrap` |\n| Add memory to existing | `\u002Fcorgi-memory-init` |\n| Migrate existing KB | `\u002Fcorgi-migrate` |\n| Health check | `\u002Fcorgi-lint` |\n\n→ **[Full Memory Documentation](docs\u002Fcross-session-memory.md)**\n\n---\n\n## 🧩 Skill Architecture\n\nSkills are organized in a **composable 3-tier hierarchy**:\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fassets\u002Fcoding_corgi_architecture.png\" alt=\"Coding Corgi Flow System Architecture\" width=\"100%\"\u002F>\n\u003C\u002Fp>\n\n| Tier | Role | Dependencies |\n|---|---|---|\n| **Atom** | Single reusable operation (resolve config, parse tasks) | None |\n| **Molecule** | Workflow combining atoms (propose, apply, review) | Atoms only |\n| **Compound** | End-to-end orchestration (the full pipeline) | Molecules only |\n\nEach skill has two files:\n- `SKILL.md` — AI-readable instructions\n- `skill.meta.json` — Machine-readable metadata (tier, deps, platform, version)\n\nValidate and visualize with the `ds-skills` CLI:\n\n```bash\ncd tools\u002Fds-skills && npm install\nnode bin\u002Fds-skills.js validate --path ..\u002F..    # schema + tier + cycle checks\nnode bin\u002Fds-skills.js graph --path ..\u002F..        # dependency graph (Mermaid)\nnode bin\u002Fds-skills.js list --path ..\u002F.. --tier atom --platform github\n```\n\n---\n\n## 📐 Schemas\n\nA schema defines the artifact pipeline. Both bundled schemas (`gitlab-tracked`, `github-tracked`) produce the same 4-artifact pipeline:\n\n| Artifact | File | Purpose |\n|---|---|---|\n| **Proposal** | `proposal.md` | Motivation, scope, capabilities, impact |\n| **Specs** | `specs\u002F\u003Ccapability>\u002Fspec.md` | Formal WHEN\u002FTHEN scenarios (one per capability) |\n| **Design** | `design.md` | Technical decisions, architecture, risks, trade-offs |\n| **Tasks** | `tasks.md` | Numbered Task Groups with checkboxes — each becomes a child issue |\n\nPipeline: `proposal → specs → design → tasks → apply`\n\nKey decisions:\n- **Capability-driven specs** — one spec file per capability, traceable contracts\n- **Delta spec model** — ADDED\u002FMODIFIED\u002FREMOVED\u002FRENAMED operations accumulate into canonical specs\n- **Task Groups as checkpoints** — each `## N. Group` = one child issue, one apply session, one review cycle\n\n\u003Cdetails>\n\u003Csummary>Creating a custom schema\u003C\u002Fsummary>\n\nCreate `openspec\u002Fschemas\u002Fmy-schema\u002F`:\n\n```\nmy-schema\u002F\n├── schema.yaml\n└── templates\u002F\n    ├── proposal.md\n    └── tasks.md\n```\n\n`schema.yaml`:\n\n```yaml\nname: my-schema\nversion: 1\ndescription: Lightweight workflow with proposal and tasks\n\nartifacts:\n  - id: proposal\n    generates: proposal.md\n    description: What and why\n    template: proposal.md\n    instruction: |\n      Write the proposal explaining the change motivation and scope.\n    requires: []\n\n  - id: tasks\n    generates: tasks.md\n    description: Implementation checklist\n    template: tasks.md\n    instruction: |\n      Break implementation into numbered Task Groups with checkboxes.\n    requires:\n      - proposal\n\napply:\n  requires:\n    - tasks\n  tracks: tasks.md\n  instruction: |\n    Execute one Task Group at a time. Mark tasks as [x] when done.\n```\n\nSet `schema: my-schema` in `config.yaml`.\n\n\u003C\u002Fdetails>\n\n---\n\n## ⚖️ Vanilla OpenSpec vs. Corgi Flow\n\n| Capability | Vanilla OpenSpec | Coding Corgi Flow |\n|---|---|---|\n| Issue tracking | None | Parent\u002Fchild issues via `gh` or `glab` |\n| Apply behavior | All tasks at once | Checkpoint-based: one group, pause, review |\n| Progress sync | Local checkboxes only | Rich summaries posted to issues |\n| Workflow labels | None | `backlog → todo → in-progress → review → done` |\n| Review | None | 5-axis automated checks + verify gate + decision loop |\n| Spec format | Generic | Delta ops (ADDED\u002FMODIFIED\u002FREMOVED\u002FRENAMED) |\n| Worktree isolation | None | Opt-in parallel dev via git worktrees |\n| Cross-session memory | None | 3-layer system with self-compaction |\n| Knowledge migration | None | Guided import from docs, archives, vault pages |\n| Memory health | None | 11-check lint (freshness, caps, links, extraction) |\n| Skill architecture | Flat files | Atoms → Molecules → Compounds with schema validation |\n| Plugin marketplace | None | Claude Code `\u002Fplugin install` + Codex marketplace |\n\n---\n\n## ⚙️ Configuration\n\nAll settings live in `openspec\u002Fconfig.yaml`:\n\n```yaml\nschema: github-tracked       # or gitlab-tracked\n\n# Optional: worktree isolation for parallel changes\nisolation:\n  mode: worktree             # worktree | none (default: none)\n  root: .worktrees\n  branch_prefix: feat\u002F\n\n# Optional: project context for AI-generated artifacts\ncontext: |\n  Tech stack: TypeScript, Next.js 14, Prisma, PostgreSQL\n  Domain: e-commerce platform\n\n# Optional: per-artifact rules\nrules:\n  proposal:\n    - Keep proposals under 500 words\n  tasks:\n    - Max 2 hours per task\n```\n\nThe installer manages only the `schema` and `isolation` keys. Add `context` and `rules` yourself.\n\nFor full install\u002Fupdate\u002Fverify reference (fresh install, managed update, local modifications, legacy migration), see [Install \u002F Update \u002F Verify Workflow](#-install--update--verify-reference) below.\n\n---\n\n## 📂 Repository Layout\n\n```\nschemas\u002F\n└── skill-meta.schema.json            # JSON Schema for skill validation\n\npackages\u002Fcorgispec\u002F                   # Unified CLI (npm publishable)\n├── src\u002F                              # TypeScript source\n├── dist\u002F                             # Built output\n└── assets\u002F                           # Bundled assets\n\ntools\u002Fds-skills\u002F                      # Skill CLI (legacy, use corgispec)\n├── bin\u002Fds-skills.js\n├── lib\u002F{loader,validate,list,graph}.js\n└── tests\u002F\n\ndocs\u002F\n├── articles\u002F                         # Comics, screenshots, publish kits\n│   └── images\u002F                       # Feature screenshots\n├── plans\u002F                            # Design & planning documents\n└── specs\u002F                            # Feature design specs\n\nopenspec\u002F\n├── config.yaml\n├── schemas\u002F{gitlab,github}-tracked\u002F  # Schema definitions + templates\n├── specs\u002F                            # Accumulated canonical specs\n└── changes\u002F                          # Active change directories\n\n.opencode\u002F\n├── skills\u002Fcorgispec-*\u002F               # Source of truth: SKILL.md + skill.meta.json\n└── commands\u002Fcorgi-*.md               # Slash command dispatch\n\n.claude\u002F\n├── skills\u002Fcorgispec-*\u002F               # Claude Code skill mirrors\n├── commands\u002Fcorgi\u002F                   # Claude slash command dispatch\n└── settings.json                     # Team auto-install config\n\n.claude-plugin\u002F                       # Claude Code Plugin manifest\n.codex-plugin\u002F                        # Codex Plugin manifest\n.codex\u002Fskills\u002Fcorgispec-*\u002F           # Codex skill symlinks → .claude\u002Fskills\u002F\n```\n\n---\n\n## 📖 Docs\n\n| Article | Lang | Description |\n|---|---|---|\n| [Cross-Session Memory](docs\u002Fcross-session-memory.md) | EN \u002F [中文](docs\u002Fcross-session-memory.zh-TW.md) | Architecture, lifecycle, migration |\n| [OpenSpec 落地 GitHub](docs\u002Fsuperpowers\u002Farticles\u002F2026-04-28-corgispec-github-workflow-zhihu.md) | 中文 | Spec → Issue → Review → Git pipeline integration |\n\n---\n\n## 🤝 Contributing\n\n1. Fork and clone\n2. Create or update a skill under `.opencode\u002Fskills\u002F`\n3. Each skill needs `SKILL.md` (AI instructions) + `skill.meta.json` (metadata)\n4. Validate: `node tools\u002Fds-skills\u002Fbin\u002Fds-skills.js validate --path .`\n5. Test locally, then submit a PR\n6. Sync changes across `.opencode\u002Fskills\u002F`, `.claude\u002Fskills\u002F`, and `.codex\u002Fskills\u002F`\n\n---\n\n## 🔧 Install \u002F Update \u002F Verify Reference\n\nThe installer supports four modes:\n\n### Fresh Install\n\nThe target project has no managed files yet:\n\n```text\n\u002Fcorgi-install --mode fresh --path \u002Fpath\u002Fto\u002Fyour-project\n```\n\nCopies managed files to `.opencode\u002F`, `.claude\u002F`, `openspec\u002Fschemas\u002F`, patches `config.yaml` minimally, writes install manifest and report.\n\n### Managed Update\n\nThe project already has `openspec\u002F.corgi-install.json`:\n\n```text\n\u002Fcorgi-install --mode update --path \u002Fpath\u002Fto\u002Fyour-project\n```\n\nIf local modifications are detected, the installer prints a diff, stops, and asks for manual resolution — it never silently overwrites your changes.\n\n### Verify-Only\n\nHealth check without mutations:\n\n```text\n\u002Fcorgi-install --mode verify --path \u002Fpath\u002Fto\u002Fyour-project\n```\n\n### Legacy Migration\n\nIf managed files exist but no install manifest, the installer classifies it as legacy, creates backups, and asks for confirmation before migrating.\n\n---\n\n## 🙏 Acknowledgments\n\nBuilt on [OpenSpec](https:\u002F\u002Fgithub.com\u002FFission-AI\u002FOpenSpec) by [Fission AI](https:\u002F\u002Fgithub.com\u002FFission-AI). The core CLI, artifact pipeline engine, and change lifecycle are all OpenSpec — we extend it with custom schemas, AI skills, issue tracking, memory, and review automation.\n\nIf you find this useful, please ⭐ [OpenSpec](https:\u002F\u002Fgithub.com\u002FFission-AI\u002FOpenSpec) too.\n\n---\n\n## 📸 Image Credits\n\n- **Hero Banner** & **Pipeline Illustration** & **Architecture Diagram** & **Memory Vault** — AI-generated via the [README visual upgrade plan](wiki\u002Fdecisions\u002Freadme-visual-upgrade.md)\n- **Corgi Comics** (chaos, confident, journey, knowledge) — AI-generated, prompts in [comic workflow guide](docs\u002Farticles\u002Fcorgi-comic-workflow.md)\n- **Feature Screenshots** — from real usage of Coding Corgi Flow on GitHub\u002FGitLab projects\n","Coding Corgi Flow 是一个基于 OpenSpec 的结构化 AI 工程工作流工具包，集成了问题跟踪功能。其核心功能包括自动问题跟踪、基于检查点的应用执行、自动化验证门、五维度代码审查以及跨会话记忆等，旨在将任意 AI 编码助手转变为有条理的工程伙伴。通过使用 TypeScript 开发，该项目提供了从提案到归档的全生命周期管理，确保开发过程中的每个步骤都有迹可循且可被审核。适用于需要增强团队协作效率和代码质量控制的 AI 项目开发场景。",2,"2026-06-11 03:54:50","CREATED_QUERY"]