midas-mcp 1.0.0

package/docs/SPEC.md

@@ -0,0 +1,330 @@
+# Midas MCP Technical Specification
+
+Technical design for the midas-mcp server.
+
+---
+
+## Architecture
+
+```
+midas-mcp/
+├── src/
+│   ├── index.ts              # MCP server entry point
+│   ├── server.ts             # Server setup and configuration
+│   ├── tools/
+│   │   ├── index.ts          # Tool exports
+│   │   ├── phase.ts          # Phase tracking tools
+│   │   ├── audit.ts          # 12 ingredients audit
+│   │   ├── docs.ts           # Check docs completeness
+│   │   ├── oneshot.ts        # Oneshot paradigm helper
+│   │   ├── tornado.ts        # Tornado cycle trigger
+│   │   └── horizon.ts        # Horizon expansion helper
+│   ├── prompts/
+│   │   ├── index.ts          # Prompt exports
+│   │   ├── session.ts        # Master/execution prompts
+│   │   ├── development.ts    # Feature/bug prompts
+│   │   ├── review.ts         # Security/performance review
+│   │   └── research.ts       # Research prompts
+│   ├── resources/
+│   │   ├── index.ts          # Resource exports
+│   │   ├── methodology.ts    # Core methodology content
+│   │   ├── prompts.ts        # Prompt library content
+│   │   └── ingredients.ts    # 12 ingredients content
+│   └── state/
+│       └── phase.ts          # Phase state machine
+├── docs/
+│   ├── METHODOLOGY.md        # Source content
+│   ├── PROMPTS.md            # Source content
+│   ├── INGREDIENTS.md        # Source content
+│   ├── USER_RULES.md         # Source content
+│   └── SPEC.md               # This file
+├── package.json
+├── tsconfig.json
+└── README.md
+```
+
+---
+
+## Phase State Machine
+
+```typescript
+type EagleSightStep = 
+  | 'IDEA'
+  | 'RESEARCH'
+  | 'BRAINLIFT'
+  | 'PRD'
+  | 'GAMEPLAN';
+
+type BuildStep =
+  | 'RULES_LOADED'
+  | 'CODEBASE_INDEXED'
+  | 'FILES_READ'
+  | 'RESEARCHING'
+  | 'IMPLEMENTING'
+  | 'TESTING'
+  | 'DEBUGGING';
+
+type Phase =
+  | { phase: 'IDLE' }
+  | { phase: 'EAGLE_SIGHT'; step: EagleSightStep }
+  | { phase: 'BUILD'; step: BuildStep }
+  | { phase: 'SHIPPED' };
+
+interface PhaseState {
+  current: Phase;
+  history: Phase[];
+  startedAt: string;
+  docs: {
+    brainlift: boolean;
+    prd: boolean;
+    gameplan: boolean;
+  };
+}
+```
+
+---
+
+## Tools Specification
+
+### midas_start_project
+
+Initializes a new project with Eagle Sight.
+
+**Input:**
+```typescript
+{
+  projectName: string;
+  projectPath?: string; // defaults to cwd
+}
+```
+
+**Behavior:**
+1. Creates `docs/` folder
+2. Creates `docs/brainlift.md` with template
+3. Creates `docs/prd.md` with template
+4. Creates `docs/gameplan.md` with template
+5. Sets phase to `EAGLE_SIGHT.IDEA`
+6. Returns guidance for next step
+
+---
+
+### midas_get_phase
+
+Returns current phase and recommended actions.
+
+**Input:** none
+
+**Output:**
+```typescript
+{
+  current: Phase;
+  nextSteps: string[];
+  prompt?: string; // suggested prompt for this phase
+}
+```
+
+---
+
+### midas_set_phase
+
+Manually set the current phase.
+
+**Input:**
+```typescript
+{
+  phase: 'EAGLE_SIGHT' | 'BUILD' | 'SHIPPED';
+  step?: string;
+}
+```
+
+---
+
+### midas_audit
+
+Audit project against 12 ingredients.
+
+**Input:**
+```typescript
+{
+  projectPath?: string;
+}
+```
+
+**Output:**
+```typescript
+{
+  scores: {
+    [ingredient: string]: {
+      exists: boolean;
+      score: number; // 0-100
+      issues: string[];
+      suggestions: string[];
+    }
+  };
+  overall: number;
+  level: 'functional' | 'integrated' | 'protected' | 'production';
+}
+```
+
+---
+
+### midas_check_docs
+
+Verify Eagle Sight docs exist and are complete.
+
+**Input:**
+```typescript
+{
+  projectPath?: string;
+}
+```
+
+**Output:**
+```typescript
+{
+  brainlift: { exists: boolean; complete: boolean; issues: string[] };
+  prd: { exists: boolean; complete: boolean; issues: string[] };
+  gameplan: { exists: boolean; complete: boolean; issues: string[] };
+  ready: boolean;
+}
+```
+
+---
+
+### midas_oneshot
+
+Construct a Oneshot retry prompt.
+
+**Input:**
+```typescript
+{
+  originalPrompt: string;
+  error: string;
+  learnings?: string[];
+}
+```
+
+**Output:**
+```typescript
+{
+  prompt: string; // constructed oneshot prompt
+}
+```
+
+---
+
+### midas_tornado
+
+Guide through tornado cycle.
+
+**Input:**
+```typescript
+{
+  problem: string;
+  currentStep?: 'research' | 'logs' | 'tests';
+}
+```
+
+**Output:**
+```typescript
+{
+  nextStep: 'research' | 'logs' | 'tests';
+  guidance: string;
+  prompt: string;
+}
+```
+
+---
+
+### midas_horizon
+
+Expand horizontal context.
+
+**Input:**
+```typescript
+{
+  currentOutput: string;
+  expectedOutput: string;
+}
+```
+
+**Output:**
+```typescript
+{
+  missingContext: string[];
+  expandedPrompt: string;
+  checklist: string[];
+}
+```
+
+---
+
+## Prompts Specification
+
+Each prompt is exposed via MCP prompt capability.
+
+### Session Prompts
+- `master_prompt` - Initialize session
+- `execution_prompt` - Start building
+
+### Development Prompts
+- `feature_planning` - Plan before implementing
+- `feature_implementation` - Implement with TDD
+- `bug_investigation` - Diagnose before fixing
+- `bug_fix` - Fix with test-first
+
+### Review Prompts
+- `security_review` - Check for vulnerabilities
+- `performance_review` - Check for bottlenecks
+- `safe_refactor` - Move code safely
+
+### Problem-Solving Prompts
+- `oneshot_retry` - Retry with full context
+- `tornado_trigger` - Research + Logs + Tests
+- `horizon_expansion` - Add missing context
+
+---
+
+## Resources Specification
+
+Resources are served via MCP resource capability.
+
+| URI | Content |
+|-----|---------|
+| `midas://methodology` | METHODOLOGY.md |
+| `midas://prompts` | PROMPTS.md |
+| `midas://ingredients` | INGREDIENTS.md |
+| `midas://user-rules` | USER_RULES.md |
+| `midas://cheatsheet` | Quick reference |
+
+---
+
+## Implementation Notes
+
+### Dependencies
+- `@modelcontextprotocol/sdk` - MCP SDK
+- `zod` - Input validation
+
+### State Persistence
+Phase state stored in `.midas/state.json` in project root.
+
+### Error Handling
+All tools return structured errors:
+```typescript
+{
+  error: string;
+  code: 'NOT_FOUND' | 'INVALID_INPUT' | 'PHASE_ERROR';
+  suggestion?: string;
+}
+```
+
+### Logging
+Use structured logging for debugging:
+```typescript
+console.error(JSON.stringify({
+  level: 'error',
+  tool: 'midas_audit',
+  message: 'Failed to scan project',
+  error: err.message
+}));
+```

package/docs/USER_RULES.md

@@ -0,0 +1,90 @@
+# User Rules Reference
+
+Expert user rules for .cursorrules files.
+
+---
+
+## Documentation Style
+
+- No emojis, no ALL CAPS, no markdown tables with dashes/pipes
+- No em dashes, no excessive formatting
+- Bold key words inside sentences for emphasis
+- Write tersely, casually, high IQ
+
+---
+
+## Paid Services (Firebase, APIs, etc.)
+
+- **SHOW THE MATH** - Calculate and comment max monthly cost
+- **LIMIT EVERYTHING** - No operation without explicit maximum
+- **PREVENT LOOPS** - Code must not trigger itself without guards
+
+---
+
+## Security
+
+- Build exhaustively secure, unhackable
+- Comprehensive .gitignore from day 1
+- Production-ready always
+- Never git push an API key
+
+---
+
+## Code Quality
+
+- Single responsibility principle
+- Achieve each task in minimum tokens
+- Testflight-deployable builds
+- Add logs and look up documentation to debug
+
+---
+
+## Workflow
+
+- Don't delete scripts when done, move to scripts folder
+- Try to compile/build before git committing, fix errors as needed
+- Simple short git commit after changes, no emojis or special characters
+- Deploy Firebase rules/functions/indexes when changed
+- Deploy Firebase cloud functions when updated
+
+---
+
+## Testing
+
+- Use iPhone 16 or 16 Pro for Xcode iOS simulator
+
+---
+
+## Full Rules Template
+
+```plaintext
+# DOCUMENTATION
+- No emojis, no ALL CAPS, no tables with dashes/pipes
+- No em dashes, no excessive formatting
+- Bold for emphasis, write tersely, casually, high IQ
+
+# PAID SERVICES
+- SHOW THE MATH - Calculate and comment max monthly cost
+- LIMIT EVERYTHING - No operation without explicit maximum
+- PREVENT LOOPS - Code must not trigger itself without guards
+
+# SECURITY
+- Build exhaustively secure, unhackable
+- Comprehensive .gitignore from day 1
+- Production-ready always
+- Never git push an API key
+
+# CODE QUALITY
+- Single responsibility principle
+- Achieve each task in minimum tokens
+- Add logs and look up documentation to debug
+
+# WORKFLOW
+- Don't delete scripts, move to scripts folder
+- Compile/build before git committing
+- Simple short git commit, no emojis or special characters
+- Deploy Firebase rules/functions/indexes when changed
+
+# TESTING
+- Use iPhone 16 or 16 Pro for Xcode iOS simulator
+```

package/package.json

@@ -0,0 +1,42 @@
+{
+  "name": "midas-mcp",
+  "version": "1.0.0",
+  "description": "MCP server for Elite Vibecoding - everything you vibecode turns to gold",
+  "main": "dist/index.js",
+  "bin": {
+    "midas-mcp": "./dist/index.js"
+  },
+  "type": "module",
+  "scripts": {
+    "build": "tsc && cp -r docs dist/",
+    "dev": "tsc --watch",
+    "start": "node dist/index.js",
+    "prepublishOnly": "npm run build"
+  },
+  "keywords": [
+    "mcp",
+    "model-context-protocol",
+    "vibecoding",
+    "cursor",
+    "ai-assistant",
+    "developer-tools"
+  ],
+  "author": "",
+  "license": "MIT",
+  "dependencies": {
+    "@modelcontextprotocol/sdk": "^1.0.0",
+    "zod": "^3.23.0"
+  },
+  "devDependencies": {
+    "@types/node": "^20.0.0",
+    "typescript": "^5.0.0"
+  },
+  "engines": {
+    "node": ">=18.0.0"
+  },
+  "files": [
+    "dist",
+    "docs",
+    "README.md"
+  ]
+}