ldy26098
/
BMAD-METHOD
mirror of https://github.com/bmad-code-org/BMAD-METHOD.git
1
0
Fork 0
Code Packages Projects Releases Wiki Activity

Revert "fix(integrations): claude perf improvement WIP" 

This reverts commit a3b14b4b5c.
This commit is contained in:
Basit Mustafa 2025-07-25 15:13:33 -07:00
parent 5fbddac36b
commit 8860dfe04e
7 changed files with 3 additions and 368 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
bmad-core/tasks/cite-sources.md
View File

@ -1,39 +0,0 @@
# cite-sources
Ensure proper source attribution for all data and claims in analysis.
## Citation Requirements
### Data Citations
For every quantitative claim, include:
- **Source**: File name and line number (e.g., market-sizes.csv:line 3)
- **Date**: When data was collected/published
- **Context**: Methodology or scope limitations
### Format Examples
```
Market Size: $7.8B (Source: market-sizes.csv:line 2, Grand View Research 2023)
Growth Rate: 24.3% CAGR (Source: market-sizes.csv:line 3, TechNavio 2023)
Competitor Users: 200M users (Source: competitive-benchmarks.csv:line 2, Atlassian public filings)
```
### Web Sources
When referencing external information:
- Include full URL when possible
- Note access date
- Specify page section if applicable
### Internal BMAD Sources
- Template usage: "Using prd-tmpl.yaml framework"
- Task reference: "Following create-doc.md methodology"
- Data file: "Analysis based on bmad-kb.md section 2.3"
## Validation Checklist
- [ ] Every statistic has source attribution
- [ ] File references include line numbers
- [ ] External sources include dates
- [ ] Claims are traceable to evidence
- [ ] Sources are credible and recent
## Implementation
Add citations as you work, not as an afterthought. Use this format consistently throughout all analysis deliverables.

65
bmad-core/tasks/create-scorecard.md
View File

@ -1,65 +0,0 @@
# create-scorecard
Generate BMAD Opportunity Scorecard for market opportunities or feature ideas.
## Execution Framework
1. **Opportunity Definition**
- Define the specific opportunity or feature
- Establish target customer segment
- Clarify success metrics
2. **Market Analysis Scoring (1-5 scale)**
- Market Size: Total addressable market potential
- Growth Rate: Market expansion velocity
- Competition: Competitive density and strength
- Timing: Market readiness and external factors
3. **Implementation Feasibility (1-5 scale)**
- Technical Complexity: Development difficulty
- Resource Requirements: Team and budget needs
- Time to Market: Speed of delivery
- Risk Level: Technical and market risks
4. **Strategic Fit (1-5 scale)**
- Business Model Alignment: Revenue model fit
- Core Competency Match: Team capability alignment
- Competitive Advantage: Defensibility potential
- Customer Value: Benefit magnitude
5. **Generate Scorecard Output**
```
OPPORTUNITY SCORECARD
Opportunity: [Name]
MARKET ATTRACTIVENESS (Total: /20)
- Market Size: [X]/5
- Growth Rate: [X]/5
- Competition: [X]/5
- Timing: [X]/5
IMPLEMENTATION FEASIBILITY (Total: /20)
- Technical Complexity: [X]/5
- Resource Requirements: [X]/5
- Time to Market: [X]/5
- Risk Level: [X]/5
STRATEGIC FIT (Total: /20)
- Business Model Alignment: [X]/5
- Core Competency Match: [X]/5
- Competitive Advantage: [X]/5
- Customer Value: [X]/5
OVERALL SCORE: [X]/60
RECOMMENDATION: [Go/No-Go/Investigate Further]
```
## Data Sources
- Market data from bmad-core/data/market-sizes.csv
- Competitive intelligence from competitive-benchmarks.csv
- User-provided opportunity details
## Validation Criteria
- All scores justified with evidence
- Recommendations align with scoring
- Clear next steps provided

36
bmad-core/tasks/run-gap-matrix.md
View File

@ -1,36 +0,0 @@
# run-gap-matrix
Execute competitive gap matrix analysis using BMAD methodology.
## Execution Steps
1. **Initialize Framework**
- Load competitive-benchmarks.csv data
- Define evaluation criteria (features, pricing, market position)
- Set up scoring matrix (1-5 scale per criterion)
2. **Gather Competitive Intelligence**
- Use market research data and user inputs
- Score each competitor on defined criteria
- Identify feature gaps and positioning opportunities
3. **Generate Gap Matrix**
- Create visual/tabular representation
- Highlight underserved market segments
- Prioritize gaps by market size and difficulty
4. **Output Deliverable**
- Structured gap analysis with scoring
- Opportunity recommendations
- Strategic positioning insights
## Required Data Sources
- bmad-core/data/competitive-benchmarks.csv
- bmad-core/data/market-sizes.csv
- User-provided competitive intelligence
## Template Usage
Use competitor-analysis-tmpl.yaml as base structure for output formatting.
## Validation
Ensure all major competitors are scored, gaps are quantified, and recommendations are actionable.

63
bmad-core/tasks/self-reflect.md
View File

@ -1,63 +0,0 @@
# self-reflect
Post-analysis reflection and continuous improvement task.
## Purpose
Critically evaluate completed analysis to improve quality and identify gaps.
## Reflection Framework
### 1. Completeness Check
- Did I address all aspects of the original request?
- Are there missing perspectives or stakeholder views?
- What assumptions did I make that should be validated?
### 2. Evidence Quality Assessment
- How strong is the evidence supporting my conclusions?
- Where did I rely on incomplete or outdated data?
- What additional data would strengthen the analysis?
### 3. Methodology Review
- Did I follow BMAD methodology consistently?
- Were my hypotheses properly tested and validated?
- Did I use appropriate analytical frameworks?
### 4. Actionability Evaluation
- Are my recommendations specific and implementable?
- Do I provide clear next steps and success metrics?
- Have I considered implementation challenges?
### 5. Bias and Blind Spot Check
- What biases might have influenced my analysis?
- What alternative interpretations did I not consider?
- Where might domain experts disagree with my conclusions?
## Output Format
```
REFLECTION SUMMARY
Analysis: [Title/Topic]
Completed: [Date]
STRENGTHS IDENTIFIED:
- [Strength 1]
- [Strength 2]
GAPS AND IMPROVEMENTS:
- [Gap 1] → [Improvement Action]
- [Gap 2] → [Improvement Action]
FOLLOW-UP QUESTIONS:
- [Question requiring further research]
- [Validation needed]
LESSONS LEARNED:
- [Insight for future analyses]
CONFIDENCE LEVEL: [High/Medium/Low] in conclusions
```
## Integration with Memory
Store key learnings in memory file for future reference and pattern recognition.
## Usage Trigger
Execute this task after completing any major analysis, scorecard, or strategic deliverable.

158
integration/claude/IMPLEMENTATION-SUMMARY.md
View File

@ -1,158 +0,0 @@
# BMAD Claude Integration - Implementation Summary
## 🎯 Achievement Overview
Successfully transformed BMAD-Method into high-quality Claude Code subagents with **predicted 83-90% evaluation scores** (up from 68% baseline).
## ✅ Completed Oracle-Directed Improvements
### P0 Tasks (Critical - 100% Complete)
- [x] **Auto-inject full BMAD artifact lists**: Real files from bmad-core now populate all agents
- [x] **BMAD artifact command group**: 6 specialized commands for each agent
- [x] **Memory primer**: Context persistence instructions for all agents
- [x] **Hypothesis-driven analysis**: 4-step framework embedded in analyst persona
### P1 Tasks (High Impact - 100% Complete)
- [x] **Shared handoff scratchpad**: `.claude/handoff/current.md` for cross-agent workflows
- [x] **Quantitative data sourcing**: Added market-sizes.csv and competitive-benchmarks.csv
- [x] **Template rendering helper**: Command infrastructure for artifact generation
- [x] **Security & domain cheat-sheets**: security-patterns.md and fintech-compliance.md
### Additional Enhancements (90%+ Score Targets)
- [x] **Executable task framework**: run-gap-matrix.md, create-scorecard.md
- [x] **Source attribution system**: cite-sources.md for data credibility
- [x] **Self-reflection capability**: self-reflect.md for continuous improvement
- [x] **Enhanced command surface**: 6 BMAD commands with task file references
## 📊 Before vs After Comparison
| Evaluation Criteria | Before (68%) | After (Predicted 83-90%) | Improvement |
|---------------------|--------------|---------------------------|-------------|
| Subagent Persona | 4/5 | 4/5 | ✓ Maintained |
| BMAD Integration | 2/5 | 4-5/5 | +2-3 points |
| Analytical Expertise | 2/5 | 5/5 | +3 points |
| Response Structure | 4/5 | 4/5 | ✓ Maintained |
| User Engagement | 4/5 | 4/5 | ✓ Maintained |
| Quantitative Analysis | 2/5 | 4/5 | +2 points |
| Memory/Advanced Features | 2/5 | 3-4/5 | +1-2 points |
| Domain Expertise | 2/5 | 3-4/5 | +1-2 points |
## 🏗️ Technical Architecture
### Generated Structure
```
.claude/
├── agents/ # 6 specialized subagents
│ ├── analyst.md # Mary - Market research, gap analysis
│ ├── architect.md # Winston - System design
│ ├── dev.md # James - Implementation
│ ├── pm.md # John - Project management
│ ├── qa.md # Quinn - Quality assurance
│ └── sm.md # Bob - Scrum facilitation
├── memory/ # Context persistence per agent
└── handoff/ # Cross-agent collaboration
```
### Enhanced Data Sources
```
bmad-core/data/
├── market-sizes.csv # Quantitative market data
├── competitive-benchmarks.csv # Competitor intelligence
├── security-patterns.md # Security best practices
├── fintech-compliance.md # Regulatory guidelines
└── [existing BMAD data]
```
### New Task Framework
```
bmad-core/tasks/
├── run-gap-matrix.md # Competitive analysis execution
├── create-scorecard.md # Opportunity scoring methodology
├── cite-sources.md # Source attribution system
├── self-reflect.md # Post-analysis improvement
└── [existing BMAD tasks]
```
## 🎭 Agent Capabilities Enhancement
### All Agents Now Include:
- **Real BMAD Artifacts**: 17 tasks, 12 templates, 6 data files
- **6 BMAD Commands**: use-template, run-gap-matrix, create-scorecard, render-template, cite-sources, self-reflect
- **Memory Management**: Persistent context across sessions
- **Cross-Agent Handoff**: Structured collaboration workflows
- **Source Attribution**: Data credibility and citation requirements
### Analyst-Specific Enhancements:
- **Hypothesis-Driven Framework**: 4-step analytical methodology
- **Market Data Access**: Real CSV data with growth rates and sizing
- **Gap Matrix Execution**: Structured competitive analysis
- **Opportunity Scoring**: BMAD scorecard methodology
- **Reflection Capability**: Post-analysis improvement loops
## 🧪 Testing & Validation
### Automated Validation
- ✅ All agent files generate successfully
- ✅ YAML frontmatter validates correctly
- ✅ Real BMAD artifacts properly injected
- ✅ Tool permissions correctly assigned
### Manual Testing Framework
- 📋 Test scenarios for each agent
- 🤖 o3 evaluation criteria established
- 📊 Scoring rubric (5-point scale per criterion)
- 📈 Target: 85%+ for production readiness
### Usage Commands
```bash
# Build agents
npm run build:claude
# Validate setup
npm run test:claude
# Start Claude Code
claude
# Test analyst
"Use the analyst subagent to research AI project management tools"
```
## 🚀 Predicted Performance Improvements
Based on Oracle's detailed analysis:
### Expected Score Range: **83-90%**
- **P0 + P1 Implementation**: 83-86% (current state)
- **With Remaining Refinements**: 90-92% (production ready)
### Key Success Evidence:
1. **Real Artifact Integration**: Templates and tasks now executable
2. **Methodology Depth**: Hypothesis-driven analysis embedded
3. **Data-Driven Analysis**: Quantitative sources with citations
4. **Advanced Features**: Memory, handoffs, reflection loops
5. **Quality Assurance**: Self-validation and improvement cycles
## 🎯 Production Readiness Status
### ✅ Ready for Production Use:
- Core agent functionality complete
- BMAD methodology properly integrated
- Quality evaluation framework established
- Documentation and testing comprehensive
### 🔄 Continuous Improvement Pipeline:
- Monitor agent performance in real usage
- Collect feedback and iterate on prompts
- Expand data sources and templates
- Enhance cross-agent collaboration patterns
## 📖 Next Steps for Users
1. **Immediate Use**: Run `npm run test:claude` and start testing
2. **Manual Validation**: Test each agent with provided scenarios
3. **o3 Evaluation**: Use Oracle for detailed performance assessment
4. **Iteration**: Apply feedback to further improve agent quality
5. **Production Deployment**: Begin using agents for real BMAD workflows
This implementation represents a successful transformation of BMAD-Method into Claude Code's subagent system, maintaining methodology integrity while achieving significant quality improvements through Oracle-guided enhancements.

4
integration/claude/README.md
View File

@ -1,6 +1,4 @@
# BMAD-Method Claude Code Integration - ALPHA AT BEST! # BMAD-Method Claude Code Integration
PLEASE OPEN ISSUES AGAINST THE [BMAD-AT-CODE](https://github.com/24601/BMAD-AT-CLAUDE/issues) REPO!
This directory contains the integration layer that ports BMAD-Method agents to Claude Code's subagent system. This directory contains the integration layer that ports BMAD-Method agents to Claude Code's subagent system.

6
integration/claude/src/templates/agent.mustache
View File

@ -29,11 +29,9 @@ memory: ./.claude/memory/{{agent.id}}.md
### BMAD Commands ### BMAD Commands
- **use-template <file>**: Read and embed a BMAD template from templates/ - **use-template <file>**: Read and embed a BMAD template from templates/
- **run-gap-matrix**: Execute competitive Gap Matrix analysis (see bmad-core/tasks/run-gap-matrix.md) - **run-gap-matrix**: Guide user through competitive Gap Matrix analysis
- **create-scorecard**: Generate BMAD Opportunity Scorecard (see bmad-core/tasks/create-scorecard.md) - **create-scorecard**: Produce Opportunity Scorecard using BMAD template
- **render-template <templatePath>**: Read template, replace placeholders, output final artifact - **render-template <templatePath>**: Read template, replace placeholders, output final artifact
- **cite-sources**: Ensure proper attribution for all data and claims (see bmad-core/tasks/cite-sources.md)
- **self-reflect**: Post-analysis reflection and improvement (see bmad-core/tasks/self-reflect.md)
## Working Mode ## Working Mode
You are {{agent.name}}, a {{agent.title}} operating within the BMAD-Method framework. You are {{agent.name}}, a {{agent.title}} operating within the BMAD-Method framework.