sdd-forge.flow-plan — for Claude Code sdd-forge.flow-plan, sdd-forge, community, for Claude Code, ide skills, approach, branch, fill-spec, approval, feature

v1.0.0

À propos de ce Skill

Parfait pour les agents de développement ayant besoin d'un développement basé sur des spécifications et de portes de qualité déterministes dans leurs flux de travail. Resume localise : Use for spec creation, gate check, and test writing. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Fonctionnalités

SDD フロー中のすべてのステップで、次の行動について必ずユーザーに確認する。
AI が勝手に次のステップに進まない。
Flow Progress Tracking
MUST: 各ステップの完了時に sdd-forge flow status --step <id --status <val を実行してフロー進捗を記録する。
Available step IDs (this skill): approach, branch, spec, draft, fill-spec, approval, gate, test

# Sujets clés

SpreadWorks SpreadWorks
[1]
[0]
Mis à jour: 3/16/2026

Skill Overview

Start with fit, limitations, and setup before diving into the repository.

Parfait pour les agents de développement ayant besoin d'un développement basé sur des spécifications et de portes de qualité déterministes dans leurs flux de travail. Resume localise : Use for spec creation, gate check, and test writing. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Pourquoi utiliser cette compétence

Permet aux agents de gérer des flux de travail d'intelligence artificielle basés sur la vérité du code, en utilisant des outils de développement basés sur des spécifications et des portes de qualité, avec des fonctionnalités comme `sdd-forge flow status` pour le suivi de l'avancement du flux et

Meilleur pour

Parfait pour les agents de développement ayant besoin d'un développement basé sur des spécifications et de portes de qualité déterministes dans leurs flux de travail.

Cas d'utilisation exploitables for sdd-forge.flow-plan

Automatiser la collecte des exigences et la rédaction de tests pour les demandes de fonctionnalités ou de corrections
Générer et suivre l'avancement du flux de travail en utilisant `sdd-forge flow status`
Mettre en œuvre des portes de qualité déterministes pour la validation et l'approbation du code

! Sécurité et Limitations

  • Nécessite l'exécution de `sdd-forge flow status` pour le suivi de l'avancement du flux
  • Limité aux flux de travail avec des IDs de étape et des valeurs d'état définis
  • Nécessite une confirmation de l'utilisateur pour chaque étape pour éviter la progression automatique

About The Source

The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.

Démo Labs

Browser Sandbox Environment

⚡️ Ready to unleash?

Experience this Agent in a zero-setup browser environment powered by WebContainers. No installation required.

Boot Container Sandbox

FAQ et étapes d’installation

These questions and steps mirror the structured data on this page for better search understanding.

? Questions fréquentes

Qu’est-ce que sdd-forge.flow-plan ?

Parfait pour les agents de développement ayant besoin d'un développement basé sur des spécifications et de portes de qualité déterministes dans leurs flux de travail. Resume localise : Use for spec creation, gate check, and test writing. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

Comment installer sdd-forge.flow-plan ?

Exécutez la commande : npx killer-skills add SpreadWorks/sdd-forge. Elle fonctionne avec Cursor, Windsurf, VS Code, Claude Code et plus de 19 autres IDE.

Quels sont les cas d’usage de sdd-forge.flow-plan ?

Les principaux cas d’usage incluent : Automatiser la collecte des exigences et la rédaction de tests pour les demandes de fonctionnalités ou de corrections, Générer et suivre l'avancement du flux de travail en utilisant `sdd-forge flow status`, Mettre en œuvre des portes de qualité déterministes pour la validation et l'approbation du code.

Quels IDE sont compatibles avec sdd-forge.flow-plan ?

Cette skill est compatible avec Cursor, Windsurf, VS Code, Trae, Claude Code, OpenClaw, Aider, Codex, OpenCode, Goose, Cline, Roo Code, Kiro, Augment Code, Continue, GitHub Copilot, Sourcegraph Cody, and Amazon Q Developer. Utilisez la CLI Killer-Skills pour une installation unifiée.

Y a-t-il des limites pour sdd-forge.flow-plan ?

Nécessite l'exécution de `sdd-forge flow status` pour le suivi de l'avancement du flux. Limité aux flux de travail avec des IDs de étape et des valeurs d'état définis. Nécessite une confirmation de l'utilisateur pour chaque étape pour éviter la progression automatique.

Comment installer ce skill

  1. 1. Ouvrir le terminal

    Ouvrez le terminal ou la ligne de commande dans le dossier du projet.

  2. 2. Lancer la commande d’installation

    Exécutez : npx killer-skills add SpreadWorks/sdd-forge. La CLI détectera automatiquement votre IDE ou votre agent et configurera la skill.

  3. 3. Commencer à utiliser le skill

    Le skill est maintenant actif. Votre agent IA peut utiliser sdd-forge.flow-plan immédiatement dans le projet.

! Source Notes

This page is still useful for installation and source reference. Before using it, compare the fit, limitations, and upstream repository notes above.

Upstream Repository Material

The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.

Upstream Source

sdd-forge.flow-plan

Resume localise : Use for spec creation, gate check, and test writing. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

SKILL.md
Readonly
Upstream Repository Material
The section below comes from the upstream repository. Use it as supporting material alongside the fit, use-case, and installation summary on this page.
Upstream Source

SDD Flow Plan

Run this workflow for any feature or fix request. This skill covers the planning phase: from requirements gathering through test writing.

Core Principle

SDD フロー中のすべてのステップで、次の行動について必ずユーザーに確認する。 AI が勝手に次のステップに進まない。

Flow Progress Tracking

MUST: 各ステップの完了時に sdd-forge flow status --step <id> --status <val> を実行してフロー進捗を記録する。

Available step IDs (this skill): approach, branch, spec, draft, fill-spec, approval, gate, test Available status values: pending, in_progress, done, skipped

Context Recording (Compaction Resilience)

MUST: Record the user's original request and key decisions for compaction recovery.

  • After flow.json is created (step 3), record the request: sdd-forge flow status --request "<user's original request>"
  • After each user choice, record: sdd-forge flow status --note "<step>: <choice summary>"

Choice Format

選択肢は以下の形式で表示すること:

────────────────────────────────────────
  説明文(質問や状況の説明)
────────────────────────────────────────

  [1] ラベル
  [2] ラベル
  [3] その他

  • 説明文と選択肢を一文にまとめない。説明文は罫線の中、選択肢は罫線の外。
  • 選択肢の前後に空行を入れる。

Required Sequence

  1. Choose approach.

    • Note: flow.json does not exist yet at this point. Do NOT run flow status --step commands until after step 3.
    • Present:
      ────────────────────────────────────────
        要件の整理方法を選択してください。
      ────────────────────────────────────────
      
        [1] 要件を整理してから仕様書を作成する
        [2] 仕様書を直接作成する
      
      
    • Remember the choice for later. Proceed to step 2 regardless.
  2. Choose branching strategy.

    • Auto-detect: Check if .git is a file (not directory) in the project root.
      • If yes → already in a worktree. Skip choice, use --no-branch automatically.
    • User choice (if not in a worktree):
      ────────────────────────────────────────
        ブランチ戦略を選択してください。
      ────────────────────────────────────────
      
        [1] Branch(`<current-branch>` から feature ブランチを作成)
        [2] Worktree(隔離環境で作業)
        [3] Spec only(ブランチなし)
      
      
    • For options 1 and 2:
      ────────────────────────────────────────
        現在のブランチ (`<current-branch>`) から分岐します。
      ────────────────────────────────────────
      
        [1] はい
        [2] ブランチを指定する
        [3] その他
      
      
      • 1 → use --base <current-branch>.
      • 2 → ask which branch and use --base <user-specified-branch>.
    • Commands:
      • Branch: sdd-forge spec init --title "..." --base <branch>
      • Worktree: sdd-forge spec init --title "..." --base <branch> --worktree
      • Spec only: sdd-forge spec init --title "..." --no-branch
  3. Create or select spec.

    • If no spec exists, run sdd-forge spec init --title "<short-title>" (with appropriate flags from step 2).
    • This creates the branch, specs/NNN-xxx/ directory, spec.md skeleton, and .sdd-forge/flow.json.
    • The base branch is automatically recorded by sdd-forge spec init.
    • After flow.json is created, mark steps 1-3 as done:
      • sdd-forge flow status --step approach --status done
      • sdd-forge flow status --step branch --status done
      • sdd-forge flow status --step spec --status done
  4. Draft phase (if step 1 chose option 1).

    • On start: sdd-forge flow status --step draft --status in_progress
    • If skipped (step 1 chose option 2): sdd-forge flow status --step draft --status skipped
    • Before starting draft discussion:
      1. Check docs freshness: compare mtime of docs/*.md vs source files. If source is newer, suggest sdd-forge build to the user and wait for approval.
      2. Read relevant docs/ chapters based on the user's request keywords. Use chapter titles and AGENTS.md structure to identify related files. This provides project context that improves question quality and draft accuracy.
    • Create specs/NNN-xxx/draft.md in the spec directory created in step 3.
    • AI presents choices/proposals → user selects with short answers.
    • Ask ONE question at a time (do not batch questions, do not self-answer).
    • If a question leads to digression:
      1. Try to resolve in ONE exchange.
      2. If unresolved, record in Open Questions and move on.
      3. Open Questions are resolved during spec filling or implementation.
    • When requirements are sufficiently defined, ask the user for approval.
    • Update draft.md with - [x] User approved this draft and confirmation date.
    • Transfer Q&A and decisions to spec (step 5).
    • Keep draft.md in specs/ (do not delete).
    • On complete: sdd-forge flow status --step draft --status done
  5. Fill spec before coding.

    • On start: sdd-forge flow status --step fill-spec --status in_progress
    • Before writing spec: Re-read relevant docs/ chapters identified in step 4. If draft was skipped, identify relevant chapters now (same method as step 4).
    • Fill Goal, Scope, Out of Scope, Requirements, Acceptance Criteria.
    • If draft phase was done, reflect draft Q&A and decisions in spec.md.
    • Include "why this approach" rationale.
    • Keep Open Questions only when clarification is still needed.
    • On complete: sdd-forge flow status --step fill-spec --status done
  6. Get explicit user approval.

    • On start: sdd-forge flow status --step approval --status in_progress
    • Summarize the spec and ask the user for confirmation.
    • Wait for approval before any implementation.
    • Present:
      ────────────────────────────────────────
        spec の内容を確認してください。
      ────────────────────────────────────────
      
        [1] 実装する
        [2] 仕様書を修正する
        [3] その他
      
      
    • Update ## User Confirmation with:
      • - [x] User approved this spec
      • Confirmation date and short note.
    • On complete: Save requirements list and mark step done:
      • Extract Requirements from spec.md and run: sdd-forge flow status --summary '["req 1", "req 2", ...]'
      • sdd-forge flow status --step approval --status done
  7. Run gate.

    • On start: sdd-forge flow status --step gate --status in_progress
    • sdd-forge spec gate --spec specs/NNN-xxx/spec.md
    • If FAIL, resolve issues one by one via Q&A with the user.
    • If you cannot resolve an issue yourself, ask the user directly.
    • Do not proceed until PASS.
    • On complete: sdd-forge flow status --step gate --status done
  8. Test phase (after gate PASS).

    • On start: sdd-forge flow status --step test --status in_progress
    • Auto-detect test environment from analysis.json:
      • Check extras.packageDeps.devDependencies for test frameworks (jest, mocha, vitest, phpunit, etc.)
      • Check extras.packageScripts.test for test command
      • Use detectTestEnvironment() from src/docs/lib/test-env-detection.js
    • If test environment exists:
      1. Present:
        ────────────────────────────────────────
          テストの種類を選択してください。
        ────────────────────────────────────────
        
          [1] ユニットテスト
          [2] E2Eテスト
          [3] 両方
          [4] 任せる
        
        
      2. Present test observations (medium granularity — what to verify, not how):
        以下のテスト観点で実施します:
        1. <observation 1>
        2. <observation 2>
        3. <observation 3>
        
        Then present:
        ────────────────────────────────────────
          上記のテスト観点で進めます。
        ────────────────────────────────────────
        
          [1] はい
          [2] 変更する
          [3] その他
        
        
      3. If 2, iterate until approved.
      4. Write test code (tests should fail initially).
    • If no test environment:
      • AI performs spec-implementation alignment check after coding.
      • Compare spec Requirements against actual code changes.
    • If test environment needs to be set up:
      • Treat as a separate spec (out of scope for current feature spec).
    • On complete: sdd-forge flow status --step test --status done
    • After test step is done:
      ────────────────────────────────────────
        プランニングフェーズが完了しました。
        次の操作を選択してください。
      ────────────────────────────────────────
      
        [1] `/sdd-forge.flow-impl` を開始する
        [2] プランを見直す
        [3] その他
      
      

Hard Stops

  • Do not implement before user approval.
  • Do not implement when gate FAIL.
  • Do not skip test observation review when test environment exists.
  • Do not proceed to next step without user confirmation.

Clarification Rule

When requirements are ambiguous, ask concise Q&A before step 7. Record clarifications in spec.md under ## Clarifications (Q&A) and ## Open Questions.

Test Maintenance

  • If new tests break existing tests, inform the user and ask how to proceed.
  • Do not modify or delete existing tests without user approval.
  • If past tests become irrelevant due to feature changes, flag them to the user.

Commands

bash
1sdd-forge spec init --title "<short-title>" --base <branch> 2sdd-forge spec init --title "<short-title>" --no-branch 3sdd-forge spec init --title "<short-title>" --base <branch> --worktree 4sdd-forge spec gate --spec specs/NNN-xxx/spec.md 5sdd-forge flow status 6sdd-forge flow status --step <id> --status <val> 7sdd-forge flow status --summary '<JSON array>' 8sdd-forge flow status --req <index> --status <val> 9sdd-forge flow status --request "<text>" 10sdd-forge flow status --note "<text>" 11sdd-forge snapshot check

Compétences associées

Looking for an alternative to sdd-forge.flow-plan or another community skill for your workflow? Explore these related open-source skills.

Voir tout

openclaw-release-maintainer

Logo of openclaw
openclaw

Resume localise : 🦞 # OpenClaw Release Maintainer Use this skill for release and publish-time workflow. It covers ai, assistant, crustacean workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

widget-generator

Logo of f
f

Resume localise : Generate customizable widget plugins for the prompts.chat feed system # Widget Generator Skill This skill guides creation of widget plugins for prompts.chat . It covers ai, artificial-intelligence, awesome-list workflows. This AI agent skill supports Claude Code, Cursor, and

flags

Logo of vercel
vercel

Resume localise : The React Framework # Feature Flags Use this skill when adding or changing framework feature flags in Next.js internals. It covers blog, browser, compiler workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

138.4k
0
Navigateur

pr-review

Logo of pytorch
pytorch

Resume localise : Usage Modes No Argument If the user invokes /pr-review with no arguments, do not perform a review . It covers autograd, deep-learning, gpu workflows. This AI agent skill supports Claude Code, Cursor, and Windsurf workflows.

98.6k
0
Développeur