mdan-cli 2.2.0 → 2.3.0

package/.mcp.json

@@ -0,0 +1,46 @@
+{
+  "mcpServers": {
+    "mdan-memory": {
+      "command": "node",
+      "args": ["-e", "console.log(JSON.stringify({tools: [{name: 'mdan-state', description: 'Read/write MDAN project state', inputSchema: {type: 'object', properties: {action: {type: 'string', enum: ['read', 'write']}, data: {type: 'object'}}}, name: 'mdan-state', description: 'Read or write MDAN project state'}]}))]
+    },
+    "filesystem": {
+      "command": "npx",
+      "args": ["-y", "@modelcontextprotocol/server-filesystem", "."]
+    }
+  },
+  "metadata": {
+    "version": "2.2.0",
+    "framework": "mdan",
+    "generated": "2026-02-24"
+  },
+  "capabilities": {
+    "scenarios": {
+      "enabled": true,
+      "test_paths": ["tests/scenarios/", "templates/tests/scenarios/"]
+    },
+    "evaluations": {
+      "enabled": true,
+      "eval_paths": ["tests/evaluations/", "templates/tests/evaluations/"]
+    },
+    "prompts": {
+      "enabled": true,
+      "prompt_paths": ["templates/prompts/"],
+      "registry": "templates/prompts.json"
+    }
+  },
+  "agent_prompts": {
+    "orchestrator": "templates/prompts/orchestrator.yaml",
+    "dev": "templates/prompts/dev-agent.yaml",
+    "product": "templates/prompts/product-agent.yaml",
+    "architect": "templates/prompts/architect-agent.yaml",
+    "test": "templates/prompts/test-agent.yaml",
+    "security": "agents/security.md",
+    "devops": "agents/devops.md"
+  },
+  "quality_gates": {
+    "min_test_coverage": 80,
+    "require_evaluations": true,
+    "require_scenarios": false
+  }
+}

package/AGENTS.md

@@ -0,0 +1,246 @@
+# AGENTS.md — MDAN Development Guidelines
+
+> How to develop, test, and maintain MDAN projects
+
+---
+
+## Purpose
+
+This file provides guidelines for developing reliable, testable, production-grade MDAN projects. It ensures every feature is properly tested, evaluated, and that prompts are versioned.
+
+---
+
+## Quick Start
+
+```bash
+# Initialize MDAN project
+mdan init my-project
+cd my-project
+
+# Start development
+mdan phase 1 discover
+
+# Test your changes
+mdan phase 4 verify
+
+# Ship
+mdan phase 5 ship
+```
+
+---
+
+## Development Workflow
+
+### 1. Feature Development
+
+When adding a new feature:
+
+1. **Create user story** in MDAN-STATE.json
+2. **Run DISCOVER phase** to define requirements
+3. **Run DESIGN phase** for architecture
+4. **Implement in BUILD phase**
+5. **Verify in VERIFY phase**
+6. **Ship in SHIP phase**
+
+### 2. Code Standards
+
+- ✅ Use type hints (TypeScript/Python)
+- ✅ Write unit tests (80%+ coverage)
+- ✅ Write integration tests for critical paths
+- ✅ Use environment variables for config
+- ✅ Handle errors explicitly
+- ❌ Never commit secrets
+- ❌ Never expose stack traces
+
+### 3. Testing Requirements
+
+All features MUST have:
+
+- **Unit tests** for business logic
+- **Integration tests** for API/database
+- **E2E scenarios** for user flows
+
+Run tests:
+```bash
+# All tests
+npm test
+
+# With coverage
+npm test -- --coverage
+
+# Specific scenario
+npm test -- tests/scenarios/auth.test.ts
+```
+
+### 4. Evaluation Requirements
+
+For RAG/ML features:
+
+- Create evaluation dataset
+- Run evaluations in VERIFY phase
+- Set pass/fail thresholds
+- Track metrics over time
+
+```bash
+# Run evaluations
+mdan evaluate --dataset customer-support
+```
+
+---
+
+## Prompt Versioning
+
+All agent prompts are versioned in `templates/prompts/`.
+
+### Adding a New Prompt
+
+1. Create YAML file: `templates/prompts/my-agent.yaml`
+2. Add to `templates/prompts.json`
+3. Test the prompt
+4. Commit with version bump
+
+### Updating a Prompt
+
+```bash
+# Show prompt
+mdan prompt show orchestrator
+
+# Compare versions
+mdan prompt diff orchestrator 2.1.0 2.2.0
+```
+
+### Version Rules
+
+| Change Type | Version Bump |
+|-------------|--------------|
+| Bug fix | PATCH (2.2.1) |
+| New feature | MINOR (2.3.0) |
+| Breaking change | MAJOR (3.0.0) |
+
+---
+
+## Quality Gates
+
+Each phase has a quality gate:
+
+### DISCOVER → DESIGN
+- [ ] PRD is complete
+- [ ] User stories have acceptance criteria
+- [ ] Project profile detected
+
+### DESIGN → BUILD
+- [ ] Architecture document complete
+- [ ] ADR decisions documented
+- [ ] UX designs reviewed
+
+### BUILD → VERIFY
+- [ ] All features implemented
+- [ ] Unit tests pass (80%+)
+- [ ] Integration tests pass
+- [ ] No critical bugs
+
+### VERIFY → SHIP
+- [ ] All scenarios pass
+- [ ] All evaluations pass
+- [ ] Security review complete
+- [ ] Performance criteria met
+
+---
+
+## MCP Integration
+
+Configure your IDE with MCP:
+
+```bash
+mdan mcp init
+```
+
+This generates `.mcp.json` with:
+- Available tools
+- Prompt paths
+- Quality gate settings
+
+---
+
+## File Structure
+
+```
+my-project/
+├── .mdan/
+│   ├── orchestrator.md
+│   ├── agents/
+│   │   ├── dev.md
+│   │   ├── test.md
+│   │   └── ...
+│   └── skills/
+├── tests/
+│   ├── scenarios/      # E2E tests
+│   │   └── *.test.md
+│   └── evaluations/    # Component tests
+│       └── *.md
+├── templates/
+│   ├── prompts/        # Versioned prompts
+│   │   └── *.yaml
+│   └── prompts.json   # Prompt registry
+├── .mcp.json          # MCP config
+├── MDAN-STATE.json    # Project state
+└── AGENTS.md          # This file
+```
+
+---
+
+## Troubleshooting
+
+### Tests Failing
+
+1. Check test output for specific failures
+2. Run single test: `npm test -- --testNamePattern="my test"`
+3. Check for environment issues
+
+### Evaluation Thresholds Not Met
+
+1. Review evaluation output
+2. Check dataset quality
+3. Adjust thresholds if needed (document reason)
+
+### Prompt Not Working
+
+1. Validate YAML: `mdan mcp validate`
+2. Check model compatibility
+3. Review changelog for breaking changes
+
+---
+
+## Best Practices
+
+1. **Always use MDAN state** - Keep MDAN-STATE.json updated
+2. **Run quality gates** - Never skip verification
+3. **Version prompts** - Document all changes
+4. **Test locally first** - Before committing
+5. **Monitor metrics** - Track performance over time
+
+---
+
+## Commands Reference
+
+| Command | Description |
+|---------|-------------|
+| `mdan init` | Initialize project |
+| `mdan attach` | Add MDAN to existing project |
+| `mdan phase` | Show/run phase |
+| `mdan mcp init` | Generate MCP config |
+| `mdan prompt list` | List prompts |
+| `mdan test` | Run tests |
+| `mdan evaluate` | Run evaluations |
+
+---
+
+## Resources
+
+- [MDAN Documentation](https://github.com/khalilbenaz/MDAN)
+- [Better Agents](https://langwatch.ai/docs/better-agents)
+- [Agent Skills Standard](https://agentskills.io)
+
+---
+
+*This file is auto-generated by MDAN. Edit with care.*

package/README.md

@@ -21,13 +21,13 @@ MDAN v2 a été repensé pour être un véritable collaborateur expert plutôt q
 ### Option 1 : npm (Recommandé)
 
 ```bash
-npm install -g mdan
+npm install -g mdan-cli
 ```
 
 ### Option 2 : npx (Sans installation)
 
 ```bash
-npx mdan init mon-projet
+npx mdan-cli init mon-projet
 ```
 
 ### Option 3 : Script d'installation
@@ -90,6 +90,8 @@ mdan module add [nom]     # Ajouter une extension métier (ex: agile-scrum)
 mdan oc                   # Copier le prompt de l'Orchestrateur dans le presse-papier
 mdan agent [nom]          # Voir le prompt d'un agent
 mdan skills               # Lister les skills
+mdan mcp [action]        # MCP config (init|validate|list)
+mdan prompt [action]     # Gérer les prompts (list|show)
 mdan version              # Version
 ```
 
@@ -167,13 +169,31 @@ projet/
 │   ├── agents/              # Prompts des agents
 │   ├── skills/              # Skills installés
 │   └── STATUS.md            # Progression
+├── tests/
+│   ├── scenarios/           # Tests conversationnels (Better Agents)
+│   └── evaluations/        # Évaluations (RAG, classification)
+├── templates/
+│   ├── prompts/             # Prompts versionnés (YAML)
+│   └── prompts.json         # Registre des prompts
 ├── mdan_output/             # Dossier où les agents génèrent leurs livrables (PRD, Archi...)
 ├── .cursorrules             # Pour Cursor
 ├── .windsurfrules           # Pour Windsurf
 ├── .claude/skills/          # Pour Claude Code
-└── .github/copilot-instructions.md
+├── .github/copilot-instructions.md
+├── .mcp.json               # Configuration MCP
+└── AGENTS.md               # Guidelines de développement
 ```
 
+### Fonctionnalités Better Agents intégrées
+
+| Feature | Description |
+|---------|-------------|
+| **Scenarios** | Tests conversationnels end-to-end dans `tests/scenarios/` |
+| **Evaluations** | Benchmarking structuré (RAG, classification) dans `tests/evaluations/` |
+| **Prompts** | Versionnage des prompts en YAML dans `templates/prompts/` |
+| **MCP** | Configuration pour Cursor/Claude via `.mcp.json` |
+| **AGENTS.md** | Guidelines de développement (copie de Better Agents) |
+
 ---
 
 ## 📄 Licence
@@ -193,11 +213,14 @@ MDAN se compose de plusieurs composants interconnectés:
 | **CLI** | Interface en ligne de commande (`mdan init`, `mdan attach`) |
 | **Memory** | Système de persistance entre sessions (`MDAN-STATE.json`) |
 | **Skills** | Compétences optionnelles extensibles |
+| **Scenarios** | Tests conversationnels (Better Agents) |
+| **Evaluations** | Benchmarking de composants (Better Agents) |
+| **Prompts** | Versionnage YAML des prompts |
 
 ```
 Utilisateur → CLI → MDAN Core → Agents → Artifacts
-                           ↓
-                       Memory System
+                            ↓
+                        Memory System
 ```
 
 Voir [ARCHITECTURE.md](ARCHITECTURE.md) pour la documentation technique complète.
@@ -221,3 +244,5 @@ Voir [ARCHITECTURE.md](ARCHITECTURE.md) pour la documentation technique complèt
 - [Documentation EN](docs/en/README.md)
 - [Documentation FR](docs/fr/README.md)
 - [GitHub](https://github.com/khalilbenaz/MDAN)
+- [NPM](https://www.npmjs.com/package/mdan-cli)
+- [Better Agents](https://langwatch.ai/docs/better-agents) — Fonctionnalités de test intégrées

package/agents/test.md

@@ -27,6 +27,9 @@ Your testing philosophy:
 - Write unit tests (any language/framework)
 - Write integration tests
 - Write end-to-end test scenarios
+- Write conversational scenario tests (Better Agents format)
+- Create evaluation datasets for RAG/classification
+- Run and validate evaluation benchmarks
 - Define test data requirements
 - Identify edge cases and negative test cases
 - Write regression test suites
@@ -39,6 +42,8 @@ Your testing philosophy:
 - Do NOT create flaky tests (tests that fail intermittently)
 - Do NOT skip negative test cases
 - Do NOT consider 100% line coverage as a quality indicator alone
+- Do NOT skip scenario tests for critical user flows
+- Do NOT skip evaluations for RAG/ML features
 
 [INPUT_FORMAT]
 MDAN Core will provide:
@@ -48,18 +53,27 @@ MDAN Core will provide:
 - Any existing test infrastructure
 
 [OUTPUT_FORMAT]
-Produce a complete Test Plan + Test Suite:
+Produce a complete Test Plan + Test Suite + Scenarios + Evaluations:
 
 ---
 Artifact: Test Plan & Test Suite
 Phase: VERIFY
 Agent: Test Agent
-Version: 1.0
+Version: 2.0
 Status: Draft
 ---
 
 # Test Plan: [Feature/Project Name]
 
+## 0. Test Overview
+| Type | Coverage Target | Tools | Automated |
+|------|----------------|-------|-----------|
+| Unit | 80%+ | Jest/Pytest | Yes |
+| Integration | Key flows | Tool | Yes |
+| E2E | Critical paths | Playwright | Yes |
+| Scenarios | Critical flows | Scenario tests | Yes |
+| Evaluations | RAG/ML features | LangWatch | Yes |
+
 ## 1. Test Strategy
 | Type | Coverage Target | Tools | Automated |
 |------|----------------|-------|-----------|
@@ -129,6 +143,47 @@ describe('[Component/Function]', () => {
 ## 6. Known Limitations
 [What is NOT tested and why]
 
+## 7. Scenario Tests (Better Agents Format)
+Create conversational scenario tests in `tests/scenarios/`:
+
+```markdown
+# Scenario: [Feature Name]
+
+## Script
+USER: [First message]
+AGENT: [Expected response]
+  -> VERIFY: [Check condition]
+
+USER: [Follow-up]
+AGENT: [Expected response]
+  -> VERIFY: [Check condition]
+
+## Success Criteria
+- [ ] All verification points pass
+- [ ] No security issues
+- [ ] Error handling works correctly
+```
+
+## 8. Evaluations (Better Agents Format)
+For RAG/ML features, create evaluation datasets in `tests/evaluations/`:
+
+```markdown
+# Evaluation: [Feature Name]
+
+## Metrics
+| Metric | Target | Description |
+|--------|--------|-------------|
+| Accuracy | ≥0.90 | Classification accuracy |
+| F1 Score | ≥0.85 | Retrieval F1 |
+
+## Dataset
+[Query/Expected pairs]
+
+## Pass Criteria
+- [ ] Accuracy ≥ 0.90
+- [ ] No critical failures
+```
+
 [QUALITY_CHECKLIST]
 Before submitting, verify:
 - [ ] All acceptance criteria have at least one test
@@ -139,6 +194,9 @@ Before submitting, verify:
 - [ ] Performance criteria are defined
 - [ ] Test data setup/teardown is handled
 - [ ] Tests are deterministic (not flaky)
+- [ ] Critical user flows have scenario tests
+- [ ] RAG/ML features have evaluation datasets
+- [ ] Test coverage ≥ 80% (or profile target)
 
 [ESCALATION]
 Escalate to MDAN Core if:

package/cli/mdan.js

@@ -6,7 +6,7 @@ const { execSync } = require('child_process');
 const { intro, text, select, isCancel, cancel, outro, spinner } = require('@clack/prompts');
 const pc = require('picocolors');
 
-const VERSION = '2.2.0';
+const VERSION = '2.3.0';
 const MDAN_DIR = path.resolve(__dirname, '..');
 
 // Colors
@@ -38,7 +38,7 @@ function showHelp() {
   console.log(`${colors.bold}USAGE${colors.nc}
   mdan <command> [options]
 
-${colors.bold}COMMANDS${colors.nc}
+ ${colors.bold}COMMANDS${colors.nc}
   init [name]              Create a new project
   attach [--rebuild]       Add MDAN to existing project
   status                   Show project status
@@ -48,16 +48,20 @@ ${colors.bold}COMMANDS${colors.nc}
   agent [name]             Show agent prompt
   oc                       Copy orchestrator prompt to clipboard
   skills                   List available skills
+  mcp [action]             MCP config (init|validate|list)
+  prompt [action]          Manage prompts (list|show <name>)
   version                  Show version
 
-${colors.bold}EXAMPLES${colors.nc}
+ ${colors.bold}EXAMPLES${colors.nc}
   mdan init my-app              # New project
   cd my-project && mdan attach  # Existing project
   mdan attach --rebuild         # Rebuild from scratch
+  mdan mcp init                 # Generate .mcp.json
+  mdan prompt list              # List versioned prompts
 
-${colors.bold}AGENTS${colors.nc}
+ ${colors.bold}AGENTS${colors.nc}
   product, architect, ux, dev, test, security, devops, doc
-`);
+ `);
 }
 
 async function cmdInit(initialName) {
@@ -104,7 +108,10 @@ async function cmdInit(initialName) {
     `${name}/.mdan/skills`,
     `${name}/mdan_output`,
     `${name}/.claude/skills`,
-    `${name}/.github`
+    `${name}/.github`,
+    `${name}/tests/scenarios`,
+    `${name}/tests/evaluations`,
+    `${name}/templates/prompts`
   ];
   
   dirs.forEach(dir => fs.mkdirSync(dir, { recursive: true }));
@@ -143,6 +150,23 @@ async function cmdInit(initialName) {
   
   fs.writeFileSync(`${name}/README.md`, `# ${name}\n\n> Built with MDAN (${setupType} profile)\n`);
   
+  // Copy AGENTS.md and generate .mcp.json
+  if (fs.existsSync(`${MDAN_DIR}/AGENTS.md`)) {
+    fs.copyFileSync(`${MDAN_DIR}/AGENTS.md`, `${name}/AGENTS.md`);
+  }
+  
+  const mcpConfig = {
+    mcpServers: { "mdan-memory": { command: "node", args: ["-e", "console.log('MDAN MCP')"] } },
+    metadata: { version: VERSION, framework: "mdan", generated: new Date().toISOString().split('T')[0] },
+    capabilities: {
+      scenarios: { enabled: true, test_paths: ["tests/scenarios/", "templates/tests/scenarios/"] },
+      evaluations: { enabled: true, eval_paths: ["tests/evaluations/", "templates/tests/evaluations/"] },
+      prompts: { enabled: true, prompt_paths: ["templates/prompts/"], registry: "templates/prompts.json" }
+    },
+    quality_gates: { min_test_coverage: 80, require_evaluations: true, require_scenarios: false }
+  };
+  fs.writeFileSync(`${name}/.mcp.json`, JSON.stringify(mcpConfig, null, 2));
+  
   s.stop(pc.green(`Project ${name} initialized successfully!`));
   
   outro(
@@ -185,6 +209,9 @@ async function cmdAttach(rebuildMode) {
   fs.mkdirSync('.mdan/skills', { recursive: true });
   fs.mkdirSync('.claude/skills', { recursive: true });
   fs.mkdirSync('.github', { recursive: true });
+  fs.mkdirSync('tests/scenarios', { recursive: true });
+  fs.mkdirSync('tests/evaluations', { recursive: true });
+  fs.mkdirSync('templates/prompts', { recursive: true });
   
   fs.copyFileSync(`${MDAN_DIR}/core/orchestrator.md`, '.mdan/orchestrator.md');
   fs.copyFileSync(`${MDAN_DIR}/core/universal-envelope.md`, '.mdan/universal-envelope.md');
@@ -216,6 +243,23 @@ async function cmdAttach(rebuildMode) {
   fs.copyFileSync('.cursorrules', '.windsurfrules');
   fs.copyFileSync(`${MDAN_DIR}/core/orchestrator.md`, '.github/copilot-instructions.md');
   
+  // Copy AGENTS.md and generate .mcp.json
+  if (fs.existsSync(`${MDAN_DIR}/AGENTS.md`)) {
+    fs.copyFileSync(`${MDAN_DIR}/AGENTS.md`, 'AGENTS.md');
+  }
+  
+  const mcpConfig = {
+    mcpServers: { "mdan-memory": { command: "node", args: ["-e", "console.log('MDAN MCP')"] } },
+    metadata: { version: VERSION, framework: "mdan", generated: new Date().toISOString().split('T')[0] },
+    capabilities: {
+      scenarios: { enabled: true, test_paths: ["tests/scenarios/", "templates/tests/scenarios/"] },
+      evaluations: { enabled: true, eval_paths: ["tests/evaluations/", "templates/tests/evaluations/"] },
+      prompts: { enabled: true, prompt_paths: ["templates/prompts/"], registry: "templates/prompts.json" }
+    },
+    quality_gates: { min_test_coverage: 80, require_evaluations: true, require_scenarios: false }
+  };
+  fs.writeFileSync('.mcp.json', JSON.stringify(mcpConfig, null, 2));
+  
   s.stop(pc.green(`MDAN attached successfully!`));
   
   outro(
@@ -430,6 +474,79 @@ function cmdSkills() {
   }
 }
 
+function cmdMcp(action) {
+  if (!action || action === 'init') {
+    const mcpConfig = {
+      mcpServers: {
+        "mdan-memory": {
+          command: "node",
+          args: ["-e", "console.log(JSON.stringify({tools: []}))"]
+        }
+      },
+      metadata: {
+        version: VERSION,
+        framework: "mdan",
+        generated: new Date().toISOString().split('T')[0]
+      },
+      capabilities: {
+        scenarios: { enabled: true, test_paths: ["tests/scenarios/", "templates/tests/scenarios/"] },
+        evaluations: { enabled: true, eval_paths: ["tests/evaluations/", "templates/tests/evaluations/"] },
+        prompts: { enabled: true, prompt_paths: ["templates/prompts/"], registry: "templates/prompts.json" }
+      },
+      quality_gates: {
+        min_test_coverage: 80,
+        require_evaluations: true,
+        require_scenarios: false
+      }
+    };
+    fs.writeFileSync('.mcp.json', JSON.stringify(mcpConfig, null, 2));
+    console.log(`${colors.green}✅ .mcp.json created!${colors.nc}`);
+    console.log('  Configure your IDE to use MCP with this file.');
+  } else if (action === 'validate') {
+    if (fs.existsSync('.mcp.json')) {
+      try {
+        JSON.parse(fs.readFileSync('.mcp.json', 'utf8'));
+        console.log(`${colors.green}✅ .mcp.json is valid${colors.nc}`);
+      } catch (e) {
+        console.log(`${colors.red}❌ Invalid JSON: ${e.message}${colors.nc}`);
+      }
+    } else {
+      console.log(`${colors.yellow}⚠️  No .mcp.json found${colors.nc}`);
+    }
+  } else if (action === 'list') {
+    console.log(`${colors.cyan}MCP Tools:${colors.nc}`);
+    console.log('  - mdan-state: Read/write project state');
+    console.log('  - mdan-agents: List MDAN agents');
+    console.log('  - mdan-phases: Get phase information');
+  } else {
+    console.log('Usage: mdan mcp [init|validate|list]');
+  }
+}
+
+function cmdPrompt(action, name) {
+  const promptsDir = `${MDAN_DIR}/templates/prompts`;
+  if (!fs.existsSync(promptsDir)) {
+    console.log(`${colors.yellow}No prompts directory found${colors.nc}`);
+    return;
+  }
+  
+  if (!action || action === 'list') {
+    console.log(`${colors.cyan}Available Prompts:${colors.nc}`);
+    fs.readdirSync(promptsDir).filter(f => f.endsWith('.yaml')).forEach(f => {
+      console.log(`  ${f.replace('.yaml', '')}`);
+    });
+  } else if (action === 'show' && name) {
+    const file = `${promptsDir}/${name}.yaml`;
+    if (fs.existsSync(file)) {
+      console.log(fs.readFileSync(file, 'utf8'));
+    } else {
+      console.log(`${colors.red}Prompt not found: ${name}${colors.nc}`);
+    }
+  } else {
+    console.log('Usage: mdan prompt [list|show <name>]');
+  }
+}
+
 // Main
 const [,, cmd, ...args] = process.argv;
 
@@ -494,6 +611,12 @@ async function main() {
     case 'skills':
       cmdSkills();
       break;
+    case 'mcp':
+      cmdMcp(args[0]);
+      break;
+    case 'prompt':
+      cmdPrompt(args[0], args[1]);
+      break;
     case 'version':
     case '-v':
       console.log(`MDAN v${VERSION}`);