dflow-sdd-ddd 0.1.1 → 0.3.0

package/docs/using-with-codex.en.md

@@ -0,0 +1,248 @@
+# Using Dflow with Codex CLI
+
+> [繁體中文](using-with-codex.md) | **English**
+
+A walk-through of what Dflow looks like when your AI coding agent is
+[Codex CLI](https://developers.openai.com/codex/cli). About 10 minutes to
+read.
+
+This guide focuses on the Codex CLI experience specifically. For the
+tool-neutral evaluation flow, see
+[`docs/evaluating-dflow.en.md`](evaluating-dflow.en.md). For the full Get Started
+and feature list, see [`README.md`](../README.en.md).
+
+## Who This Guide Is For
+
+You are using or evaluating Dflow with Codex CLI as your AI coding agent.
+This guide covers what Codex sees after `init`, how the `AGENTS.md` shim
+points to the canonical Dflow guide, and the Codex-specific command and
+permission patterns worth knowing.
+
+You do not need to read this before running `init`. It is most useful after
+you have run `init` once and want to understand what Codex CLI is actually
+loading.
+
+## Prerequisites
+
+- Codex CLI installed and authenticated (see
+  [developers.openai.com/codex/cli](https://developers.openai.com/codex/cli)).
+- Node.js / npm available (Dflow ships through npm). Install globally with
+  `npm install -g dflow-sdd-ddd`, or use `npx dflow-sdd-ddd` for the no-install path.
+- A project directory you are comfortable initializing in. A branch or a
+  disposable sample project is recommended for first contact; see the
+  [evaluator guide playbook](evaluating-dflow.en.md#a-30-minute-evaluation-playbook).
+- Codex started from the initialized project root, or with `codex --cd` set
+  to that root, so Codex's `AGENTS.md` discovery includes the Dflow shim.
+
+Running Dflow workflows does not require a separate Dflow service or API key.
+The workflows are Markdown-based instructions and project files.
+
+## What Codex CLI Sees After `init`
+
+Running `dflow init` (or `npx dflow-sdd-ddd init` on the no-install path) and
+selecting `AGENTS.md - Codex / Copilot coding agent` as a target tool creates
+a thin shim at the project root:
+
+```markdown
+# AGENTS.md - Dflow Project Instructions
+
+This project uses Dflow for spec-first AI-assisted development.
+
+Before planning or editing code, read and follow:
+
+- `dflow/specs/shared/AI-AGENT-GUIDE.md`
+
+Keep tool-specific instruction files small. The Dflow guide above is the
+single source of truth for project workflow rules, slash-command behavior,
+spec locations, and SDD/DDD constraints.
+```
+
+Two things matter when Codex starts in this project:
+
+1. Codex CLI reads `AGENTS.md` as project instructions. This is Codex's
+   standard repository-instruction mechanism.
+2. The Dflow shim does not include a Markdown import line. Unlike the
+   Claude Code and Gemini shims, generated `AGENTS.md` does not contain
+   `@dflow/specs/shared/AI-AGENT-GUIDE.md`.
+
+That means Codex sees the pointer immediately, but the canonical Dflow guide
+is not auto-inlined by the shim. Before planning or editing, Codex should
+follow the pointer and read `dflow/specs/shared/AI-AGENT-GUIDE.md`. If Codex
+starts answering a Dflow request without mentioning that file, steer it
+explicitly: "Before continuing, read and follow
+`dflow/specs/shared/AI-AGENT-GUIDE.md`."
+
+The canonical guide is where the real workflow rules live: project context
+(track, tech stack, prose language), the Dflow workflow table,
+source-of-truth file paths, and core SDD/DDD rules. The `AGENTS.md` shim
+stays small so the same canonical guide can serve Codex CLI, Claude Code,
+Gemini CLI, GitHub Copilot, and other tools.
+
+If an `AGENTS.md` already existed in the project, `init` does not overwrite
+it. If the existing file does not already point to
+`dflow/specs/shared/AI-AGENT-GUIDE.md`, `init` writes a merge snippet under
+`dflow/specs/shared/AGENTS-md-snippet.md` that you can merge manually. This
+avoids destroying custom project instructions you already had.
+
+## Using Dflow Workflow Commands in Codex CLI
+
+Codex CLI has its own built-in slash command layer for controlling the CLI
+session. Commands such as `/permissions`, `/model`, `/status`, `/diff`,
+`/review`, and `/init` are Codex CLI controls, not Dflow workflows.
+
+Dflow's `/dflow:*` entries are workflow names recognized by the AI through
+`AI-AGENT-GUIDE.md`, not registered Codex CLI commands. Raw
+`/dflow:new-feature` passthrough behavior in Codex CLI should be verified
+with the maintainer for the supported Codex version. The reliable form is to
+name the workflow as a plain chat instruction:
+
+```text
+Run the Dflow /dflow:new-feature workflow.
+```
+
+If your Codex CLI version passes unknown slash-prefixed input through to the
+model, this shorter form may also work (verify with maintainer):
+
+```text
+/dflow:new-feature
+```
+
+If Codex reports an unknown slash command, re-send the request in prose:
+
+```text
+Treat /dflow:new-feature as a Dflow workflow name, not as a Codex CLI
+command. Read dflow/specs/shared/AI-AGENT-GUIDE.md and start that workflow.
+```
+
+A typical conversation looks like:
+
+```text
+You: Run the Dflow /dflow:new-feature workflow.
+
+Codex CLI: I'll read dflow/specs/shared/AI-AGENT-GUIDE.md first, then use the
+new-feature workflow. Please describe the user-visible capability or business
+behavior you want to add.
+
+You: Allow expense submitters to attach a receipt image when filing an
+expense.
+
+Codex CLI: I'll start by drafting a feature spec under
+dflow/specs/features/active/. Before I do, I have a few clarifying questions.
+```
+
+The workflow then walks you through spec drafting, behavior examples,
+implementation planning, and finish-feature drift checks. The exact
+sequence depends on which workflow you entered (`/dflow:new-feature`,
+`/dflow:modify-existing`, `/dflow:bug-fix`, etc.).
+
+Available workflow entry points:
+
+| Workflow | Use when |
+|---|---|
+| `/dflow:new-feature` | A new user-visible capability or business behavior is requested. |
+| `/dflow:modify-existing` | Existing behavior needs to change. |
+| `/dflow:bug-fix` | A defect can be described with expected vs actual behavior. |
+| `/dflow:new-phase` | An active feature needs another implementation slice. |
+| `/dflow:finish-feature` | Implementation is complete and needs drift closure. |
+| `/dflow:verify` | Specs, domain docs, implementation, and tests need consistency checks. |
+| `/dflow:pr-review` | A change is ready for SDD/DDD review. |
+| `/dflow:report-dflow-feedback` | You found a Dflow issue or improvement and want a sanitized upstream feedback draft. |
+
+If you forget a workflow name, ask Codex to read
+`dflow/specs/shared/AI-AGENT-GUIDE.md` and list the available Dflow
+workflows.
+
+## Differences vs Other AI Tools
+
+The canonical guide (`dflow/specs/shared/AI-AGENT-GUIDE.md`) is identical
+across tools. Only the root-level shim differs:
+
+| Tool | Generated shim | Loads canonical guide via |
+|---|---|---|
+| Claude Code | `CLAUDE.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| Codex / Copilot coding agent | `AGENTS.md` | Project instructions load the shim; Codex must follow the pointer and read the guide |
+| Gemini CLI | `GEMINI.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| GitHub Copilot | `.github/copilot-instructions.md` | Reads repository instructions directly |
+
+You can run `dflow configure-agents` later to add another tool's shim
+without re-running `init`. Multiple tools can be active in the same project
+and stay synchronized via the canonical guide.
+
+Codex also has its own project-instruction layering. It can read global
+instructions from Codex home and project instructions from `AGENTS.md` files
+between the project root and the current working directory. For Dflow, the
+important practical rule is simple: start Codex at the initialized project
+root, and keep the Dflow pointer in the nearest relevant `AGENTS.md`.
+
+If your team uses both Claude Code and Codex CLI on the same project, no
+extra Dflow coordination is needed. Both tools use the same canonical guide;
+only the shim file and loading mechanism differ.
+
+## Common Patterns and Gotchas
+
+**Keep `AGENTS.md` thin.** If you find yourself adding workflow rules, spec
+locations, or SDD constraints to `AGENTS.md`, those belong in
+`dflow/specs/shared/AI-AGENT-GUIDE.md` instead. The shim stays small so that
+other tools' shims do not drift away from it.
+
+**Codex does not inline the Dflow guide from `AGENTS.md`.** The generated
+Codex shim has a normal Markdown bullet pointing to the canonical guide, not
+an `@...` import. Ask Codex to read `AI-AGENT-GUIDE.md` if it appears to be
+working from the shim alone.
+
+**`/dflow:*` is not a Codex CLI built-in slash command.** Codex slash
+commands control the Codex session itself. Use Dflow workflow names as plain
+chat instructions when raw slash input is intercepted or rejected. Raw
+`/dflow:*` passthrough behavior should be verified with the maintainer for
+the supported Codex version.
+
+**Do not confuse Codex `/init` with Dflow `init`.** Codex `/init` creates a
+generic `AGENTS.md` scaffold for Codex. Dflow setup is `dflow init` (or
+`npx dflow-sdd-ddd init` on the no-install path), and adding later tool shims
+is `dflow configure-agents`.
+
+**Permission gates and Dflow workflow gates are separate.** Codex may ask
+permission to run a command, edit outside the workspace, or access network
+depending on its sandbox and approval settings. Dflow workflows have their
+own approval gates, such as confirming a spec before implementation. Both
+can appear in the same session; this is expected.
+
+**The common Codex local-work preset is workspace write plus on-request
+approvals.** In current Codex CLI terminology this is
+`--sandbox workspace-write --ask-for-approval on-request`. In that mode,
+Codex can work inside the project and asks before going beyond the sandbox,
+such as writing outside the workspace or accessing network.
+
+**Existing `AGENTS.md` files are preserved.** If Dflow cannot safely write
+the root shim because the file already exists, look under
+`dflow/specs/shared/` for the merge snippet and merge the Dflow pointer into
+your existing project instructions manually.
+
+**Nested `AGENTS.md` files can change what Codex sees.** Codex layers project
+instructions along the path to the current working directory. If a subfolder
+has its own `AGENTS.md` or `AGENTS.override.md`, make sure it does not hide
+or contradict the Dflow pointer you expect Codex to follow.
+
+## Where to Go Next
+
+If you have not run `init` yet:
+
+- Follow the [evaluator guide playbook](evaluating-dflow.en.md#a-30-minute-evaluation-playbook)
+  to try it on a disposable sample project.
+
+If you have run `init` and want to see end-to-end workflow examples:
+
+- Read [`tutorial/01-greenfield/`](../tutorial/01-greenfield/walkthrough-00-setup.md) or
+  [`tutorial/02-brownfield/`](../tutorial/02-brownfield/walkthrough-00-setup.md). The
+  tutorial walk-throughs show conversation flows and the resulting
+  `dflow/specs/` outputs.
+
+If you want to understand the design rationale:
+
+- Read [`docs/why-ddd-for-ai.en.md`](why-ddd-for-ai.en.md).
+
+If something does not work as described:
+
+- File a docs feedback issue (see [`CONTRIBUTING.md`](../CONTRIBUTING.md)).
+  Per-tool documentation is new and feedback specifically about Codex CLI
+  behavior is valuable.

package/docs/using-with-codex.md

@@ -0,0 +1,224 @@
+# 在 Codex CLI 中使用 Dflow
+
+> **繁體中文** | [English](using-with-codex.en.md)
+
+當你的 AI 程式設計助理是 [Codex CLI](https://developers.openai.com/codex/cli) 時，Dflow 的使用體驗 walk-through。閱讀約需 10 分鐘。
+
+本指南專注於 Codex CLI 的具體使用體驗。工具中立的評估流程請見
+[`docs/evaluating-dflow.md`](evaluating-dflow.md)。完整的 Get Started
+與功能列表請見 [`README.md`](../README.md)。
+
+## 本指南的適用對象
+
+你正在使用或評估以 Codex CLI 作為 AI 程式設計助理的 Dflow。
+本指南說明 `init` 之後 Codex 看到了什麼、`AGENTS.md` 薄 shim 是如何指向
+canonical Dflow 指南的，以及幾個值得了解的 Codex 專屬指令與權限模式。
+
+你不需要在執行 `init` 之前先讀本指南。它最適合在你執行過一次 `init` 之後、
+想了解 Codex CLI 實際載入什麼內容時閱讀。
+
+## 前置條件
+
+- 已安裝並完成認證的 Codex CLI（見
+  [developers.openai.com/codex/cli](https://developers.openai.com/codex/cli)）。
+- 已具備 Node.js / npm 環境（Dflow 透過 npm 發佈）。以
+  `npm install -g dflow-sdd-ddd` 全域安裝，或用 `npx dflow-sdd-ddd` 走免安裝路徑。
+- 有一個你願意在其中執行 init 的專案目錄。首次嘗試建議先用 branch 或
+  可拋棄的範例專案；見
+  [評估者指南 Playbook](evaluating-dflow.md#30-分鐘評估-playbook)。
+- Codex 從已初始化的專案根目錄啟動，或以 `codex --cd` 指向該根目錄，
+  確保 Codex 的 `AGENTS.md` 探索範圍包含 Dflow shim。
+
+執行 Dflow workflow 不需要獨立的 Dflow 服務或 API 金鑰。
+這些 workflow 是 Markdown-based 的指示與專案檔案。
+
+## `init` 之後 Codex CLI 看到了什麼
+
+執行 `dflow init`（或免安裝路徑的 `npx dflow-sdd-ddd init`）並選擇
+`AGENTS.md - Codex / Copilot coding agent` 作為目標工具後，會在專案根目錄
+建立一個薄 shim：
+
+```markdown
+# AGENTS.md - Dflow Project Instructions
+
+This project uses Dflow for spec-first AI-assisted development.
+
+Before planning or editing code, read and follow:
+
+- `dflow/specs/shared/AI-AGENT-GUIDE.md`
+
+Keep tool-specific instruction files small. The Dflow guide above is the
+single source of truth for project workflow rules, slash-command behavior,
+spec locations, and SDD/DDD constraints.
+```
+
+Codex 在這個專案中啟動時，有兩件事值得注意：
+
+1. Codex CLI 將 `AGENTS.md` 作為專案指示讀取。這是 Codex 的
+   標準 repository 指示機制。
+2. Dflow shim 不含 Markdown import 那一行。與 Claude Code 和 Gemini 的
+   shim 不同，產生的 `AGENTS.md` 不含 `@dflow/specs/shared/AI-AGENT-GUIDE.md`。
+
+這意味著 Codex 能立即看到指標，但 canonical Dflow 指南不會由 shim 自動 inline 嵌入。
+在規劃或編輯之前，Codex 應跟著指標讀取 `dflow/specs/shared/AI-AGENT-GUIDE.md`。
+若 Codex 在回應 Dflow 請求時沒有提到該檔案，請明確引導它：「Before continuing,
+read and follow `dflow/specs/shared/AI-AGENT-GUIDE.md`.」
+
+canonical 指南是實際 workflow 規則的所在：專案上下文（track、技術棧、
+文章語言）、Dflow workflow 表、source-of-truth 檔案路徑，以及核心 SDD/DDD 規則。
+`AGENTS.md` shim 刻意保持精簡，這樣 canonical 指南就能同時服務 Codex CLI、
+Claude Code、Gemini CLI、GitHub Copilot 與其他工具。
+
+如果專案中已有 `AGENTS.md`，`init` 不會覆蓋它。若既有檔案尚未指向
+`dflow/specs/shared/AI-AGENT-GUIDE.md`，`init` 會在
+`dflow/specs/shared/AGENTS-md-snippet.md` 下寫入 merge snippet，
+讓你手動合併。這樣可以避免破壞你已有的自訂專案指示。
+
+## 在 Codex CLI 中使用 Dflow Workflow 指令
+
+Codex CLI 有自己的內建 slash command 層，用來控制 CLI session。
+`/permissions`、`/model`、`/status`、`/diff`、`/review`、`/init` 等指令
+都是 Codex CLI 控制項，不是 Dflow workflow。
+
+Dflow 的 `/dflow:*` 項目是 AI 透過 `AI-AGENT-GUIDE.md` 識別的 workflow
+名稱，不是已註冊的 Codex CLI 指令。`/dflow:new-feature` 在 Codex CLI 中
+能否直接傳遞到模型，需依所用的 Codex 版本向 maintainer 確認。
+最可靠的方式是以普通對話指示輸入 workflow 名稱：
+
+```text
+Run the Dflow /dflow:new-feature workflow.
+```
+
+若你的 Codex CLI 版本會將未知的 slash 前綴輸入傳遞到模型，也可嘗試以下
+較短形式（請先向 maintainer 確認）：
+
+```text
+/dflow:new-feature
+```
+
+若 Codex 回報未知 slash command，改以純文字重新送出：
+
+```text
+Treat /dflow:new-feature as a Dflow workflow name, not as a Codex CLI
+command. Read dflow/specs/shared/AI-AGENT-GUIDE.md and start that workflow.
+```
+
+典型的對話如下：
+
+```text
+You: Run the Dflow /dflow:new-feature workflow.
+
+Codex CLI: I'll read dflow/specs/shared/AI-AGENT-GUIDE.md first, then use the
+new-feature workflow. Please describe the user-visible capability or business
+behavior you want to add.
+
+You: Allow expense submitters to attach a receipt image when filing an
+expense.
+
+Codex CLI: I'll start by drafting a feature spec under
+dflow/specs/features/active/. Before I do, I have a few clarifying questions.
+```
+
+接著這個 workflow 會引導你完成 spec 起草、行為範例、實作計畫，以及
+finish-feature 漂移（drift）檢查。確切的流程取決於你進入的是哪個 workflow
+（`/dflow:new-feature`、`/dflow:modify-existing`、`/dflow:bug-fix` 等）。
+
+可用的 workflow 入口：
+
+| 指令 | 適用情境 |
+|---|---|
+| `/dflow:new-feature` | 需要新增一個使用者可見的功能或業務行為。 |
+| `/dflow:modify-existing` | 需要修改現有行為。 |
+| `/dflow:bug-fix` | 可以用預期行為 vs 實際行為描述的缺陷。 |
+| `/dflow:new-phase` | 進行中的 feature 需要另一個實作 slice。 |
+| `/dflow:finish-feature` | 實作完成後需要進行漂移（drift）收尾。 |
+| `/dflow:verify` | 需要對 spec、領域文件、實作與測試進行一致性檢查。 |
+| `/dflow:pr-review` | 變更已準備好進行 SDD/DDD review。 |
+| `/dflow:report-dflow-feedback` | 你發現了 Dflow 的問題或改進點，想要一份清理過的上游回饋草稿。 |
+
+如果你忘了 workflow 名稱，請 Codex 讀取 `dflow/specs/shared/AI-AGENT-GUIDE.md`
+並列出可用的 Dflow workflow 即可。
+
+## 與其他 AI 工具的差異
+
+canonical 指南（`dflow/specs/shared/AI-AGENT-GUIDE.md`）在各工具之間是相同的。
+只有根目錄層的 shim 有所不同：
+
+| 工具 | 產生的 shim | 載入 canonical 指南的方式 |
+|---|---|---|
+| Claude Code | `CLAUDE.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| Codex / Copilot coding agent | `AGENTS.md` | 專案指示載入 shim；Codex 須跟著指標讀取指南 |
+| Gemini CLI | `GEMINI.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| GitHub Copilot | `.github/copilot-instructions.md` | 直接讀取 repository 指示 |
+
+你可以之後執行 `dflow configure-agents` 來新增另一個工具的 shim，而不需要重跑
+`init`。同一個專案可以同時啟用多個工具，並透過 canonical 指南保持同步。
+
+Codex 也有自己的專案指示分層機制。它可以從 Codex home 讀取全域指示、
+從專案根目錄到當前工作目錄之間的 `AGENTS.md` 檔案讀取專案指示。
+對 Dflow 而言，關鍵的實務規則很簡單：從已初始化的專案根目錄啟動 Codex，
+並將 Dflow 指標保留在最近一層相關的 `AGENTS.md` 中。
+
+如果你的團隊在同一個專案中同時使用 Claude Code 和 Codex CLI，
+不需要額外的 Dflow 協調。兩個工具都讀取相同的 canonical 指南；
+只有 shim 檔案與載入機制不同。
+
+## 常見模式與注意事項
+
+**保持 `AGENTS.md` 精簡。** 如果你發現自己在把 workflow 規則、spec 路徑或
+SDD 約束加入 `AGENTS.md`，這些內容應該放到
+`dflow/specs/shared/AI-AGENT-GUIDE.md`。shim 保持精簡，其他工具的 shim 才不會
+與它產生漂移（drift）。
+
+**Codex 不會從 `AGENTS.md` inline 嵌入 Dflow 指南。** 產生的 Codex shim
+是以普通的 Markdown bullet 指向 canonical 指南，而非 `@...` import。
+若 Codex 看起來只從 shim 工作，請要求它讀取 `AI-AGENT-GUIDE.md`。
+
+**`/dflow:*` 不是 Codex CLI 的內建 slash command。** Codex slash command 控制
+的是 Codex session 本身。當 slash 輸入被攔截或拒絕時，改用普通對話指示
+輸入 Dflow workflow 名稱。`/dflow:*` 的直通行為需依所用的 Codex 版本向
+maintainer 確認。
+
+**不要混淆 Codex `/init` 與 Dflow `init`。** Codex `/init` 為 Codex 建立
+通用的 `AGENTS.md` scaffold。Dflow 的設定是 `dflow init`（或免安裝路徑的
+`npx dflow-sdd-ddd init`），之後新增工具 shim 則是 `dflow configure-agents`。
+
+**Permission gates 與 Dflow workflow gates 是分開的。** Codex 可能會依其
+sandbox 與 approvals 設定，詢問執行指令、workspace 外編輯或存取網路的權限。
+Dflow workflow 有自己的審核關卡，例如在實作前確認 spec。兩者可能在同一個
+session 中同時出現；這是預期行為。
+
+**Codex 本機工作的常見預設是 workspace write 加上 on-request approvals。**
+以目前 Codex CLI 的術語為 `--sandbox workspace-write --ask-for-approval on-request`。
+在這個模式下，Codex 可在專案內工作，並在超出 sandbox 範圍（例如 workspace
+外寫入或存取網路）前先詢問。
+
+**既有的 `AGENTS.md` 會被保留。** 若 Dflow 因檔案已存在而無法安全寫入
+根目錄 shim，請到 `dflow/specs/shared/` 下找 merge snippet，
+手動將 Dflow 指標合併進你現有的專案指示。
+
+**巢狀 `AGENTS.md` 可能改變 Codex 看到的內容。** Codex 沿著到當前工作目錄
+的路徑分層讀取專案指示。若某個子目錄有自己的 `AGENTS.md` 或
+`AGENTS.override.md`，請確認它不會遮蓋或矛盾你預期 Codex 遵循的 Dflow 指標。
+
+## 下一步
+
+如果你還沒有執行 `init`：
+
+- 按照[評估者指南 Playbook](evaluating-dflow.md#30-分鐘評估-playbook)
+  在可拋棄的範例專案中試用。
+
+如果你已執行 `init` 且想查看端到端的 workflow 範例：
+
+- 閱讀 [`tutorial/01-greenfield/`](../tutorial/01-greenfield/walkthrough-00-setup.md) 或
+  [`tutorial/02-brownfield/`](../tutorial/02-brownfield/walkthrough-00-setup.md)。
+  tutorial walk-through 展示了對話流程與產生的 `dflow/specs/` 輸出。
+
+如果你想了解設計理念：
+
+- 閱讀[為什麼 AI 時代 DDD 更重要](why-ddd-for-ai.md)。
+
+如果有任何行為與描述不符：
+
+- 開一個 docs feedback issue（見 [`CONTRIBUTING.md`](../CONTRIBUTING.md)）。
+  Per-tool 文件是新內容，有關 Codex CLI 行為的具體回饋非常有價值。

package/docs/using-with-gemini-cli.en.md

@@ -0,0 +1,200 @@
+# Using Dflow with Gemini CLI
+
+> [繁體中文](using-with-gemini-cli.md) | **English**
+
+A walk-through of what Dflow looks like when your AI coding agent is
+[Gemini CLI](https://github.com/google/gemini-cli). About 10 minutes to read.
+
+This guide focuses on the Gemini CLI experience specifically. For the
+tool-neutral evaluation flow, see
+[`docs/evaluating-dflow.en.md`](evaluating-dflow.en.md). For the full Get Started
+and feature list, see [`README.md`](../README.en.md).
+
+## Who This Guide Is For
+
+You are using or evaluating Dflow with Gemini CLI as your AI coding agent.
+This guide covers what Gemini CLI sees after `init`, how Dflow's slash
+commands are recognized, and the Gemini-CLI-specific patterns worth knowing.
+
+You do not need to read this before running `init`. It is most useful after
+you have run `init` once and want to understand what Gemini CLI is actually
+loading.
+
+## Prerequisites
+
+- Gemini CLI installed (see [github.com/google/gemini-cli](https://github.com/google/gemini-cli)).
+- Node.js / npm available (Dflow ships through npm). Install globally with
+  `npm install -g dflow-sdd-ddd`, or use `npx dflow-sdd-ddd` for the no-install path.
+- A project directory you are comfortable initializing in. A branch or a
+  disposable sample project is recommended for first contact; see the
+  [evaluator guide playbook](evaluating-dflow.en.md#a-30-minute-evaluation-playbook).
+
+You do not need a separate Dflow service or API key. The workflows are
+text-based and use Gemini CLI's existing capabilities.
+
+## What Gemini CLI Sees After `init`
+
+Running `dflow init` (or `npx dflow-sdd-ddd init` on the no-install path) and
+selecting Gemini CLI as a target tool creates a thin shim at the project root:
+
+```markdown
+# GEMINI.md - Dflow Project Instructions
+
+This project uses Dflow for spec-first AI-assisted development.
+
+Before planning or editing code, read and follow:
+
+- `dflow/specs/shared/AI-AGENT-GUIDE.md`
+
+Keep tool-specific instruction files small. The Dflow guide above is the
+single source of truth for project workflow rules, slash-command behavior,
+spec locations, and SDD/DDD constraints.
+
+If your tool supports Markdown imports, the canonical guide is imported below:
+
+@dflow/specs/shared/AI-AGENT-GUIDE.md
+```
+
+Two things happen when Gemini CLI starts in this project:
+
+1. Gemini CLI automatically loads `GEMINI.md` from the project root into
+   its context. This is Gemini CLI's standard project instructions
+   mechanism.
+2. The trailing `@dflow/specs/shared/AI-AGENT-GUIDE.md` line uses Gemini
+   CLI's Markdown import syntax to inline the canonical Dflow guide. So
+   Gemini CLI effectively reads both files as one set of instructions.
+
+The canonical guide (`dflow/specs/shared/AI-AGENT-GUIDE.md`) is where the
+real workflow rules live: project context (track, tech stack, prose
+language), the `/dflow:*` workflow table, source-of-truth file paths, and
+core SDD/DDD rules. The `GEMINI.md` shim stays small precisely so the
+canonical guide can evolve without Gemini-CLI-specific edits.
+
+If a `GEMINI.md` already existed in the project, `init` does not overwrite
+it. Instead it writes a merge snippet under `dflow/specs/shared/` that you
+can paste into your existing `GEMINI.md` manually.
+
+## Using Dflow Workflow Commands in Gemini CLI
+
+Dflow's `/dflow:*` slash commands are workflow names recognized by the AI
+through the workflow table in `AI-AGENT-GUIDE.md`, not Gemini CLI's
+built-in tool commands. You type them as plain chat:
+
+```text
+/dflow:new-feature
+```
+
+Gemini CLI treats this as input. Because it has the workflow table loaded
+via `GEMINI.md` import, it recognizes the prefix and enters the matching
+workflow. A typical conversation looks like:
+
+```text
+User: /dflow:new-feature
+
+Gemini CLI: Entering new-feature workflow. Please describe the user-facing
+capability or business behavior you want to add.
+
+User: Allow expense submitters to attach a receipt image when filing an
+expense.
+
+Gemini CLI: I'll start by drafting a feature spec under
+dflow/specs/features/active/. Before I do, I need a short answer on:
+[clarifying questions about scope, owner, priority]
+```
+
+The workflow then walks you through spec drafting, behavior examples,
+implementation planning, and finish-feature drift checks. The exact
+sequence depends on which workflow you entered (`/dflow:new-feature`,
+`/dflow:modify-existing`, `/dflow:bug-fix`, etc.). All workflow definitions
+live under the Dflow skill source; Gemini CLI follows them by reading the
+skill files when needed.
+
+Available workflow entry points:
+
+| Command | Use when |
+|---|---|
+| `/dflow:new-feature` | A new user-visible capability or business behavior is requested. |
+| `/dflow:modify-existing` | Existing behavior needs to change. |
+| `/dflow:bug-fix` | A defect can be described with expected vs actual behavior. |
+| `/dflow:new-phase` | An active feature needs another implementation slice. |
+| `/dflow:finish-feature` | Implementation is complete and needs drift closure. |
+| `/dflow:verify` | Specs, domain docs, implementation, and tests need consistency checks. |
+| `/dflow:pr-review` | A change is ready for SDD/DDD review. |
+| `/dflow:report-dflow-feedback` | You found a Dflow issue or improvement and want a sanitized upstream feedback draft. |
+
+If you forget a command name, ask Gemini CLI "what dflow workflows are
+available?" — the answer comes from the workflow table it already has
+loaded.
+
+## Differences vs Other AI Tools
+
+The canonical guide (`dflow/specs/shared/AI-AGENT-GUIDE.md`) is identical
+across tools. Only the root-level shim differs:
+
+| Tool | Generated shim | Loads canonical guide via |
+|---|---|---|
+| Gemini CLI | `GEMINI.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| Claude Code | `CLAUDE.md` | `@dflow/specs/shared/AI-AGENT-GUIDE.md` Markdown import |
+| Codex / Copilot coding agent | `AGENTS.md` | Reads file content directly when starting |
+| GitHub Copilot | `.github/copilot-instructions.md` | Reads file content directly |
+
+You can run `dflow configure-agents` later to add another tool's shim
+without re-running `init`. Multiple tools can be active in the same project
+and stay synchronized via the canonical guide.
+
+## Common Patterns and Gotchas
+
+**Keep `GEMINI.md` thin.** If you find yourself adding workflow rules,
+spec locations, or SDD constraints to `GEMINI.md`, those belong in
+`dflow/specs/shared/AI-AGENT-GUIDE.md` instead. The shim stays small so
+that other tools' shims don't drift away from it.
+
+**`/dflow:*` is not a Gemini CLI tool.** The slash commands are plain text
+patterns the AI recognizes from the workflow table. You can use them
+immediately after `init` without any Gemini CLI configuration.
+
+**Permission gates and Dflow workflow gates are separate.** Gemini CLI may
+ask for confirmation before executing a shell command or modifying a file.
+Dflow's workflows have their own approval gates (e.g., "I drafted the spec —
+do you want me to proceed to implementation?"). Both can fire on the same
+action; this is expected and not a sign of misconfiguration.
+
+**The `@` import is not recursive.** `GEMINI.md` imports
+`AI-AGENT-GUIDE.md`, but if `AI-AGENT-GUIDE.md` references other files
+(e.g., feature specs), those are not auto-loaded — Gemini CLI reads them
+on demand when entering the relevant workflow. This keeps context usage
+proportional to active work.
+
+**A pre-existing `GEMINI.md` is preserved.** `init` will not overwrite your
+existing project instructions. Look under `dflow/specs/shared/` for the
+merge snippet `init` wrote and paste the relevant sections into your
+existing `GEMINI.md` manually.
+
+**Cross-machine projects work.** `dflow/specs/` is plain Markdown checked
+into your repo. Anyone cloning the repo and using Gemini CLI in it will
+see the same Dflow setup automatically through the committed `GEMINI.md`
+shim and the canonical guide.
+
+## Where to Go Next
+
+If you have not run `init` yet:
+
+- Follow the [evaluator guide playbook](evaluating-dflow.en.md#a-30-minute-evaluation-playbook)
+  to try it on a disposable sample project.
+
+If you have run `init` and want to see end-to-end workflow examples:
+
+- Read [`tutorial/01-greenfield/`](../tutorial/01-greenfield/walkthrough-00-setup.md) or
+  [`tutorial/02-brownfield/`](../tutorial/02-brownfield/walkthrough-00-setup.md). The
+  tutorial walk-throughs show conversation flows and the resulting
+  `dflow/specs/` outputs.
+
+If you want to understand the design rationale:
+
+- Read [`docs/why-ddd-for-ai.en.md`](why-ddd-for-ai.en.md).
+
+If something does not work as described:
+
+- File a docs feedback issue (see [`CONTRIBUTING.md`](../CONTRIBUTING.md)).
+  Per-tool documentation is new and feedback specifically about Gemini CLI
+  behavior is valuable.