@raftlabs/raftstack 1.9.2 → 1.10.0

package/.claude/commands/raftstack/inject.md

@@ -2,6 +2,15 @@
 
 Surface relevant standards and skills for the current work context.
 
+## 🔒 Planning Protocol
+
+This command follows the RaftStack Planning Protocol:
+- **No file modifications** - This command is read-only (context injection for planning)
+- If user wants to implement after injecting context, recommend `/raftstack/shape [task]`
+- That command will handle planning and approval gates
+
+**Reference:** See `_planning-protocol.md` for full protocol details.
+
 ## Arguments
 
 - `$ARGUMENTS` - (Optional) Specific domain or task context (e.g., "React component", "API endpoint", "database migration")
@@ -15,6 +24,15 @@ Surface relevant standards and skills for the current work context.
 
 Never use plain text questions - always use the structured `AskUserQuestion` tool.
 
+## Important Note About Implementation
+
+**This command does NOT implement code.** It only surfaces relevant context for planning.
+
+After injecting context, if the user wants to proceed with implementation:
+1. Recommend running `/raftstack/shape [task description]`
+2. The shape command will plan the work and require approval before implementation
+3. Never proceed directly to implementation from this command
+
 ## Phase 1: Context Detection
 
 If no specific context provided, infer from:
@@ -39,30 +57,30 @@ If context is unclear, use `AskUserQuestion`:
 
 ## Phase 2: Standards Discovery
 
-Scan for standards files in likely locations:
+Scan for standards files:
 
-1. **Check common standard locations:**
+1. **Canonical locations:**
    - `.claude/standards/` and subdirectories
-   - `docs/standards/`
-   - `standards/`
-   - Files matching `*.standard.md`
+   - `.claude/context/constitution.md`
+
+2. **Legacy fallback (suggest migration):**
+   - `docs/standards/`, `standards/`, `*.standard.md`
+   - `CONSTITUTION.md`, `docs/constitution.md`
+
+   If found at legacy locations, suggest migration to `.claude/`.
 
-2. **Match by domain:**
+3. **Match by domain:**
    - React work → look for component, state, hooks standards
    - API work → look for route, validation, error standards
    - Database work → look for schema, query, migration standards
 
-3. **Check for constitution/context:**
-   - `.claude/context/constitution.md`
-   - `CONSTITUTION.md`
-   - `docs/constitution.md`
-
 ## Phase 3: Skills Matching
 
 Based on the detected domain, identify relevant RaftStack skills:
 
 | Domain | Skill | Path |
 |--------|-------|------|
+| Testing/TDD | Test-Driven Development | `.claude/skills/tdd/SKILL.md` |
 | React/Frontend | React Development | `.claude/skills/react/SKILL.md` |
 | API/Backend | Backend Development | `.claude/skills/backend/SKILL.md` |
 | Database | Database Design | `.claude/skills/database/SKILL.md` |
@@ -70,6 +88,28 @@ Based on the detected domain, identify relevant RaftStack skills:
 | General | Code Quality | `.claude/skills/code-quality/SKILL.md` |
 | Asana Integration | Asana Workflow | `.claude/skills/asana/SKILL.md` |
 
+## Phase 3.5: Plugin Recommendations
+
+Based on the detected domain, identify required Claude Code plugins:
+
+### Plugin Enforcement Matrix
+
+| Domain | Plugin | Skill/Tool | When to Trigger |
+|--------|--------|------------|-----------------|
+| React/Frontend/JSX | `frontend-design` | `/frontend-design` skill | Any UI component, JSX, CSS work |
+| React/Frontend/JSX | `figma` | `/implement-design` | Figma files to code |
+| Testing | `playwright` | Browser tools | E2E tests, browser automation |
+| Testing | `superpowers` | `/tdd` skill | Test-driven development |
+| Research/Documentation | `context7` | Documentation lookup | Getting library docs, API references |
+| Code Review | `code-review` | `/code-review` | Before commits, PR creation |
+| Code Review | `pr-review-toolkit` | `/review-pr` | Specialized review agents |
+| Git/Commits | `commit-commands` | `/commit`, `/commit-push-pr` | Any git operations |
+| Git/Commits | `github` | GitHub tools | Issue/PR management |
+| Backend/API | `security-guidance` | Security warnings | API routes, handlers, auth |
+| Deployment | `vercel` | Vercel tools | Deployment-related tasks |
+| Project Management | `linear` | Linear tools | Task tracking with Linear |
+| Project Management | `asana` | Asana tools | Task tracking with Asana |
+
 ## Phase 4: Context-Aware Summary
 
 Present the injected context as a structured summary:
@@ -89,6 +129,13 @@ Present the injected context as a structured summary:
 | [Name] | `[Path]` | [Trigger condition] |
 | [Name] | `[Path]` | [Trigger condition] |
 
+### 🔌 Recommended Plugins
+| Plugin | Purpose | Skill to Use |
+|--------|---------|--------------|
+| [plugin] | [why needed for this domain] | `/[skill-name]` or [tool description] |
+
+**Important:** Always use `context7` when researching libraries or getting documentation.
+
 ### Quick Reference (Most Important Rules)
 1. **[Rule]** - [Brief explanation]
 2. **[Rule]** - [Brief explanation]

package/.claude/commands/raftstack/shape.md

@@ -2,6 +2,15 @@
 
 Plan a feature or task with scale-adaptive depth. Simple tasks get quick flow, complex tasks get full specs.
 
+## 🔒 Planning Protocol
+
+This command follows the RaftStack Planning Protocol:
+- All changes are planned before implementation
+- User approval is required before any file modifications
+- Use `AskUserQuestion` for all approval gates
+
+**Reference:** See `_planning-protocol.md` for full protocol details.
+
 ## Arguments
 
 - `$ARGUMENTS` - Description of the feature or task to plan
@@ -13,6 +22,7 @@ Plan a feature or task with scale-adaptive depth. Simple tasks get quick flow, c
 - Location negotiation
 - Option selection
 - Clarification questions
+- **Implementation approval** (required before any file operations)
 
 Never use plain text questions - always use the structured `AskUserQuestion` tool.
 
@@ -63,12 +73,14 @@ After assessment, use `AskUserQuestion` to confirm the complexity level:
 
 ## Phase 2: Execute Based on Scale
 
+**IMPORTANT:** Each flow level (Quick/Light/Full) MUST include an approval gate before implementation.
+
 ### Quick Flow (Simple Tasks)
 
 1. Clarify scope in 2-3 sentences
 2. Identify the file(s) to change
 3. Describe the change briefly
-4. Proceed directly to implementation
+4. **WAIT FOR APPROVAL** - Do NOT implement yet
 
 Output format:
 ```markdown
@@ -80,13 +92,34 @@ Output format:
 **Scope:** [What needs to change]
 **Files:** [1-2 files]
 **Approach:** [Brief description]
+**Tests First:** [What test to write before implementation]
 
-Ready to implement.
-
-**Your options:** [A] Proceed to implement [B] Add more detail [C] Change approach [D] Cancel
+### 🔌 Use These Plugins
+- [Plugin based on domain detected]
+- `tdd` (mandatory for all implementation work)
 ```
 
-Use `AskUserQuestion` for options.
+#### ⚠️ PLANNING GATE (Quick Flow)
+
+**DO NOT IMPLEMENT WITHOUT USER APPROVAL**
+
+Before implementing:
+
+1. **Present the Plan Above** (already done)
+
+2. **Request Approval** using `AskUserQuestion` with these options:
+   - [A] Proceed to implement (Recommended)
+   - [B] Add more detail
+   - [C] Change approach
+   - [D] Cancel
+
+3. **Implementation Rules:**
+   - ✅ Wait for explicit [A] selection
+   - ✅ If [B] selected, provide more detail and re-present
+   - ✅ If [C] selected, revise approach and re-present
+   - ❌ Never skip approval
+   - ❌ Never implement without [A] selection
+   - ❌ Never create/modify files before approval
 
 ### Light Spec (Medium Tasks)
 
@@ -94,6 +127,7 @@ Use `AskUserQuestion` for options.
 2. Identify related code patterns to follow
 3. List files to create/modify
 4. Note any standards to apply
+5. **WAIT FOR APPROVAL** - Do NOT implement yet
 
 Output format:
 ```markdown
@@ -109,18 +143,46 @@ Output format:
 [Reference similar code in the codebase to follow]
 
 ### Implementation Plan
-1. [Step with file path]
-2. [Step with file path]
-3. [Step with file path]
+1. Write failing test for [core behavior]
+2. Implement [step with file path]
+3. Verify test passes
+4. Add edge case tests
+
+### 🔌 Plugins to Use
+| Plugin | Purpose | When |
+|--------|---------|------|
+| `tdd` | Test-first development | All implementation work (mandatory) |
+| [Plugin] | [Why needed] | [Trigger condition] |
+
+**Important:** Always use `context7` when researching libraries or getting documentation.
 
 ### Standards to Apply
 - [Reference any discovered standards: @.claude/standards/...]
 - [Reference any skills: @.claude/skills/...]
-
-**Your options:** [A] Proceed to implement [B] Save spec to file [C] Modify plan [D] Upgrade to full spec
 ```
 
-Use `AskUserQuestion` for options.
+#### ⚠️ PLANNING GATE (Light Spec)
+
+**DO NOT IMPLEMENT WITHOUT USER APPROVAL**
+
+Before implementing:
+
+1. **Present the Plan Above** (already done)
+
+2. **Request Approval** using `AskUserQuestion` with these options:
+   - [A] Proceed to implement (Recommended)
+   - [B] Save spec to file
+   - [C] Modify plan
+   - [D] Upgrade to full spec
+
+3. **Implementation Rules:**
+   - ✅ Wait for explicit [A] selection
+   - ✅ If [B] selected, save spec and await further instructions
+   - ✅ If [C] selected, revise plan and re-present
+   - ✅ If [D] selected, transition to Full Spec workflow
+   - ❌ Never skip approval
+   - ❌ Never implement without [A] selection
+   - ❌ Never create/modify files before approval
 
 ### Full Spec (Complex Tasks)
 
@@ -142,17 +204,15 @@ Use `AskUserQuestion` for options.
    - Error handling strategy
 
 4. **Break into phases:**
-   - Phase 1: Core functionality
-   - Phase 2: Edge cases/polish
-   - Phase 3: Testing/docs
+   - Phase 1: Core tests (TDD) - [N test files]
+   - Phase 2: Core functionality - [N files]
+   - Phase 3: Edge cases/polish - [N files]
 
-5. **Location negotiation for spec folder:**
+5. **Spec folder location:**
 
-Use `AskUserQuestion` with these options:
-- Option A: `.claude/specs/[feature-slug]/` - Claude-specific planning (Recommended)
-- Option B: `docs/specs/[feature-slug]/` - With documentation
-- Option C: `specs/[feature-slug]/` - At project root
-- Option D: Other (let user specify)
+Specs are always saved at: `.claude/specs/{YYYY-MM-DD-HHMM}-{feature-slug}/`
+
+**Note:** For business requirements (PRD, user stories), use `docs/prd/` instead.
 
 ## Full Spec Folder Structure
 
@@ -204,6 +264,13 @@ After creating the spec folder, present:
 - **references.md:** [N] similar patterns found
 - **architecture.md:** Technical design
 
+### 🔌 Plugins & Skills
+| Plugin | Purpose | When |
+|--------|---------|------|
+| [Plugin] | [Why needed] | [Trigger condition] |
+
+**Important:** Always use `context7` when researching libraries or getting documentation.
+
 ### 🔍 Implementation Phases
 1. **Phase 1:** [Core functionality] - [N files]
 2. **Phase 2:** [Edge cases] - [N files]
@@ -213,22 +280,40 @@ After creating the spec folder, present:
 - [Risk or complexity] - **Mitigation:** [How to handle]
 - [Dependency or blocker] - **Status:** [Current state]
 
-### 🔄 Ready for Implementation
+### 📁 Spec Location
 Spec folder created at: `{path}`
-
-**Your options:** [A] Start implementation [B] Review spec files [C] Modify architecture [D] Add more detail
 ```
 
-Use `AskUserQuestion` for options.
+#### ⚠️ PLANNING GATE (Full Spec)
+
+**DO NOT IMPLEMENT WITHOUT USER APPROVAL**
+
+Before implementing:
+
+1. **Present the Plan Above** (already done)
+
+2. **Request Approval** using `AskUserQuestion` with these options:
+   - [A] Start implementation (Recommended)
+   - [B] Review spec files in detail
+   - [C] Modify architecture
+   - [D] Add more detail
+
+3. **Implementation Rules:**
+   - ✅ Wait for explicit [A] selection
+   - ✅ If [B] selected, review spec files and re-present
+   - ✅ If [C] selected, revise architecture and re-present
+   - ✅ If [D] selected, expand spec and re-present
+   - ❌ Never skip approval
+   - ❌ Never implement without [A] selection
+   - ❌ Never create/modify code files before approval
+   - ✅ Spec folder creation is allowed (it's planning, not implementation)
 
 ## Phase 3: Standards Injection
 
 Before finalizing the plan, check for relevant standards and skills:
 
-1. Scan for standards files in likely locations:
+1. Scan for standards at canonical location:
    - `.claude/standards/`
-   - `docs/standards/`
-   - Project root (`*.standard.md`)
 
 2. Identify relevant skills based on the task domain:
    - React work → `@.claude/skills/react/SKILL.md`

package/.claude/settings.json

@@ -1,3 +1,27 @@
 {
-  "model": "opusplan"
+  "model": "opusplan",
+  "enabledPlugins": {
+    "frontend-design@claude-plugins-official": true,
+    "context7@claude-plugins-official": true,
+    "code-review@claude-plugins-official": true,
+    "github@claude-plugins-official": true,
+    "feature-dev@claude-plugins-official": true,
+    "typescript-lsp@claude-plugins-official": true,
+    "code-simplifier@claude-plugins-official": true,
+    "ralph-loop@claude-plugins-official": true,
+    "commit-commands@claude-plugins-official": true,
+    "playwright@claude-plugins-official": true,
+    "security-guidance@claude-plugins-official": true,
+    "pr-review-toolkit@claude-plugins-official": true,
+    "figma@claude-plugins-official": true,
+    "superpowers@claude-plugins-official": true,
+    "explanatory-output-style@claude-plugins-official": true,
+    "linear@claude-plugins-official": true,
+    "learning-output-style@claude-plugins-official": true,
+    "vercel@claude-plugins-official": true,
+    "asana@claude-plugins-official": true,
+    "claude-md-management@claude-plugins-official": true,
+    "claude-code-setup@claude-plugins-official": true,
+    "document-skills@anthropic-agent-skills": true
+  }
 }