/workflow
Complete Research → Plan → Implement workflow in one command. Invokes chief-architect to orchestrate all phases with quality gates and self-correction.
How it fires
How this command gets triggered: by you, by Claude, or both.
- Fires itselfClaude auto-loads it when your prompt matches the work.
- You can call itInvoke it directly when you want it.
- Slash command
/workflow
Context preview
What this command does when you run it.
Complete Research → Plan → Implement workflow in one command. Invokes chief-architect to orchestrate all phases with quality gates and self-correction.
Command definition
workflow.mdname: workflow
description: Complete Research → Plan → Implement workflow in one command. Invokes chief-architect to orchestrate all phases with quality gates and self-correction.
/workflow Command
Execute complete Research → Plan → Implement workflow using chief-architect orchestration.
Usage
/workflow <feature description>
What This Does
**Complete automation of all three phases**:
1. **Research Phase**
- Invokes `@docs-researcher`
- Applies `research-methodology` skill
- Creates ResearchPack
- Validates score ≥ 80
2. **Planning Phase**
- Invokes `@implementation-planner`
- Applies `planning-methodology` skill
- Creates Implementation Plan
- Validates score ≥ 85
- Runs API matcher (no hallucination)
3. **Implementation Phase**
- Checks circuit breaker (must be closed)
- Invokes `@code-implementer`
- Executes plan with self-correction
- Validates tests pass
- Records circuit breaker state
4. **Knowledge Capture**
- Applies `pattern-recognition` skill
- Updates knowledge-core.md
- Records decisions and learnings
Examples
/workflow Add Redis caching to product service with TTL
/workflow Implement JWT authentication middleware
/workflow Create user profile page with avatar upload
/workflow Add Stripe payment integration with webhooks
/workflow Migrate database from SQLite to PostgreSQL
Output
You'll receive a **complete project report**:
📈 Project Summary
- Goal: [Your request]
- Outcome: [What was delivered]
- Duration: [Actual time]
- Agents Used: researcher, planner, implementer
🛠️ Phase Results
**Research**:
- Library/API researched
- Version detected
- APIs documented (count)
- ResearchPack score
**Planning**:
- Files to change (count)
- Implementation steps (count)
- Risks identified (count)
- Plan score
**Implementation**:
- Files created/modified
- Tests passing/failing
- Self-corrections made (if any)
- Circuit breaker state
📚 Artifacts Created
- ResearchPack.md
- ImplementationPlan.md
- Code files (list)
- Test files (list)
🧠 Knowledge Captured
- Patterns identified
- Decisions recorded
- Suggestions for knowledge-core.md
Quality Gates (Automatic)
At each phase transition:
**Research → Planning**:
- ✅ ResearchPack score ≥ 80
- ✅ Library version identified
- ✅ Minimum 3 APIs documented
- ✅ Code examples present
- ⛔ If fail: Blocks planning, requests fixes
**Planning → Implementation**:
- ✅ Plan score ≥ 85
- ✅ APIs match ResearchPack exactly
- ✅ Rollback plan present
- ✅ Risk assessment complete
- ⛔ If fail: Blocks implementation, requests fixes
**Implementation → Completion**:
- ✅ Circuit breaker closed
- ✅ All tests passing
- ✅ Build successful
- ⛔ If fail: Up to 3 self-corrections, then block
Advantages Over Manual Steps
**Manual** (3 commands):
/research [topic]
# wait, review
/plan [feature]
# wait, review
/implement
# wait, review
**Automatic** (1 command):
/workflow [feature]
# orchestrator handles everything
**Benefits**:
- ✅ No waiting between phases
- ✅ Automatic quality gate enforcement
- ✅ Context preserved throughout
- ✅ Single comprehensive report
- ✅ Automatic knowledge capture
When to Use `/workflow` vs Manual Commands
**Use `/workflow` when**:
- ✅ Clear, well-defined feature
- ✅ Trust orchestrator to handle details
- ✅ Want complete automation
- ✅ Standard feature (not experimental)
**Use manual commands when**:
- ⚠️ Need to review research before planning
- ⚠️ Want to modify plan before implementing
- ⚠️ Uncertain about approach
- ⚠️ Learning the system
Time
Typical completion by feature complexity:
| Complexity | Time | Example | |------------|------|---------| | Simple | 10-15 min | API integration | | Medium | 15-25 min | Caching layer, Auth middleware | | Complex | 25-40 min | Database migration, Payment flow | | Very Complex | 40-60 min | Multi-service integration |
Error Handling
**If research fails**:
- Chief-architect reports research issues
- Suggests fixes or retry
- Does not proceed to planning
**If planning fails**:
- Chief-architect reports planning issues
- May suggest research update
- Does not proceed to implementation
**If implementation fails** (after 3 self-corrections):
- Circuit breaker opens
- Chief-architect provides complete failure analysis
- Lists all attempts made
- Recommends manual intervention
- Requires circuit breaker reset before retry
Progress Tracking
Chief-architect reports progress throughout:
🏛️ Starting analysis for [feature]
🗺️ Designing multi-agent execution plan...
🤝 Delegating task 'research' to @docs-researcher
📊 Research phase progress: 50%...
✅ ResearchPack complete (score: 85/100)
🤝 Delegating task 'planning' to @implementation-planner
📊 Planning phase progress: 60%...
✅ Implementation Plan complete (score: 88/100)
✅ API matching passed
🤝 Delegating task 'implementation' to @code-implementer
📊 Implementation phase: 3/5 files complete...
🧪 Running tests...
✅ All tests passing
🔄 Synthesizing results...
✅ Project complete: [summary]
Rollback
If you need to undo the entire workflow:
# Use rollback procedure from ImplementationPlan.md
git reset --hard HEAD~[N commits]
# Or if not yet committed:
git checkout -- [files modified]
# Or reset completely:
git clean -fd
Each Implementation Plan includes specific rollback instructions.
Tips for Best Results
**Be specific in request**:
- ❌ "Add caching" (too vague)
- ✅ "Add Redis caching to ProductService with 5-minute TTL"
**Include constraints**:
- "Add auth using JWT tokens, not sessions"
- "Use PostgreSQL, not MongoDB"
- "Must work with existing API structure"
**Specify non-functional requirements**:
- "Must handle 1000 req/sec"
- "Must roll back on error"
- "Must log all operations"
Next Steps
After `/workflow` completes:
1. **Review all artifacts**
- ResearchPack.md
- ImplementationPlan.md
- Code changes
2. *
Read more
name: workflow description: Complete Research → Plan → Implement workflow in one command. Invokes chief-architect to orchestrate all phases with quality gates and self-correction.
/workflow Command
Execute complete Research → Plan → Implement workflow using chief-architect orchestration.
Usage
/workflow <feature description>
What This Does
**Complete automation of all three phases**:
1. **Research Phase**
- Invokes `@docs-researcher`
- Applies `research-methodology` skill
- Creates ResearchPack
- Validates score ≥ 80
2. **Planning Phase**
- Invokes `@implementation-planner`
- Applies `planning-methodology` skill
- Creates Implementation Plan
- Validates score ≥ 85
- Runs API matcher (no hallucination)
3. **Implementation Phase**
- Checks circuit breaker (must be closed)
- Invokes `@code-implementer`
- Executes plan with self-correction
- Validates tests pass
- Records circuit breaker state
4. **Knowledge Capture**
- Applies `pattern-recognition` skill
- Updates knowledge-core.md
- Records decisions and learnings
Examples
/workflow Add Redis caching to product service with TTL /workflow Implement JWT authentication middleware /workflow Create user profile page with avatar upload /workflow Add Stripe payment integration with webhooks /workflow Migrate database from SQLite to PostgreSQL
Output
You'll receive a **complete project report**:
📈 Project Summary
- Goal: [Your request]
- Outcome: [What was delivered]
- Duration: [Actual time]
- Agents Used: researcher, planner, implementer
🛠️ Phase Results
**Research**:
- Library/API researched
- Version detected
- APIs documented (count)
- ResearchPack score
**Planning**:
- Files to change (count)
- Implementation steps (count)
- Risks identified (count)
- Plan score
**Implementation**:
- Files created/modified
- Tests passing/failing
- Self-corrections made (if any)
- Circuit breaker state
📚 Artifacts Created
- ResearchPack.md
- ImplementationPlan.md
- Code files (list)
- Test files (list)
🧠 Knowledge Captured
- Patterns identified
- Decisions recorded
- Suggestions for knowledge-core.md
Quality Gates (Automatic)
At each phase transition:
**Research → Planning**:
- ✅ ResearchPack score ≥ 80
- ✅ Library version identified
- ✅ Minimum 3 APIs documented
- ✅ Code examples present
- ⛔ If fail: Blocks planning, requests fixes
**Planning → Implementation**:
- ✅ Plan score ≥ 85
- ✅ APIs match ResearchPack exactly
- ✅ Rollback plan present
- ✅ Risk assessment complete
- ⛔ If fail: Blocks implementation, requests fixes
**Implementation → Completion**:
- ✅ Circuit breaker closed
- ✅ All tests passing
- ✅ Build successful
- ⛔ If fail: Up to 3 self-corrections, then block
Advantages Over Manual Steps
**Manual** (3 commands):
/research [topic] # wait, review /plan [feature] # wait, review /implement # wait, review
**Automatic** (1 command):
/workflow [feature] # orchestrator handles everything
**Benefits**:
- ✅ No waiting between phases
- ✅ Automatic quality gate enforcement
- ✅ Context preserved throughout
- ✅ Single comprehensive report
- ✅ Automatic knowledge capture
When to Use `/workflow` vs Manual Commands
**Use `/workflow` when**:
- ✅ Clear, well-defined feature
- ✅ Trust orchestrator to handle details
- ✅ Want complete automation
- ✅ Standard feature (not experimental)
**Use manual commands when**:
- ⚠️ Need to review research before planning
- ⚠️ Want to modify plan before implementing
- ⚠️ Uncertain about approach
- ⚠️ Learning the system
Time
Typical completion by feature complexity:
| Complexity | Time | Example | |------------|------|---------| | Simple | 10-15 min | API integration | | Medium | 15-25 min | Caching layer, Auth middleware | | Complex | 25-40 min | Database migration, Payment flow | | Very Complex | 40-60 min | Multi-service integration |
Error Handling
**If research fails**:
- Chief-architect reports research issues
- Suggests fixes or retry
- Does not proceed to planning
**If planning fails**:
- Chief-architect reports planning issues
- May suggest research update
- Does not proceed to implementation
**If implementation fails** (after 3 self-corrections):
- Circuit breaker opens
- Chief-architect provides complete failure analysis
- Lists all attempts made
- Recommends manual intervention
- Requires circuit breaker reset before retry
Progress Tracking
Chief-architect reports progress throughout:
🏛️ Starting analysis for [feature] 🗺️ Designing multi-agent execution plan... 🤝 Delegating task 'research' to @docs-researcher 📊 Research phase progress: 50%... ✅ ResearchPack complete (score: 85/100) 🤝 Delegating task 'planning' to @implementation-planner 📊 Planning phase progress: 60%... ✅ Implementation Plan complete (score: 88/100) ✅ API matching passed 🤝 Delegating task 'implementation' to @code-implementer 📊 Implementation phase: 3/5 files complete... 🧪 Running tests... ✅ All tests passing 🔄 Synthesizing results... ✅ Project complete: [summary]
Rollback
If you need to undo the entire workflow:
# Use rollback procedure from ImplementationPlan.md git reset --hard HEAD~[N commits] # Or if not yet committed: git checkout -- [files modified] # Or reset completely: git clean -fd
Each Implementation Plan includes specific rollback instructions.
Tips for Best Results
**Be specific in request**:
- ❌ "Add caching" (too vague)
- ✅ "Add Redis caching to ProductService with 5-minute TTL"
**Include constraints**:
- "Add auth using JWT tokens, not sessions"
- "Use PostgreSQL, not MongoDB"
- "Must work with existing API structure"
**Specify non-functional requirements**:
- "Must handle 1000 req/sec"
- "Must roll back on error"
- "Must log all operations"
Next Steps
After `/workflow` completes:
1. **Review all artifacts**
- ResearchPack.md
- ImplementationPlan.md
- Code changes
2. *
Research-first development system for Claude Code CLI No API hallucinations. No coding from stale training data. Research → Plan → Implement.
Repo: VAMFI/claude-user-memory
Other commands on claude-user-memory.
- /context
Analyze and optimize context configuration. Reviews CLAUDE.md, knowledge-core.md, and active context for optimization opportunities. Helps prevent context rot (39% improvement, 84% token reduction).
Open command - /implement
Quick command to invoke code-implementer to execute implementation with self-correction (up to 3 intelligent retries).
Open command - /plan
Quick command to invoke implementation-planner to create surgical, reversible implementation plans with rollback procedures.
Open command - /research
Quick command to invoke docs-researcher for rapid documentation research. Fetches version-accurate, authoritative sources and creates ResearchPack.
Open command

