myaiforone 1.0.0

package/docs/appcreator_plan.md

@@ -0,0 +1,348 @@
+# App Creator Plan
+
+## Vision
+
+The App Creator is a platform agent that takes a user's natural language description and autonomously produces a deployed, production-quality web application — matching the quality bar of apps the platform owner has personally built. The user touches it twice: the initial prompt and the "deploy" button. Everything between is autonomous.
+
+---
+
+## Stack (Opinionated, Non-Negotiable)
+
+| Layer | Choice | Why |
+|-------|--------|-----|
+| API Server | Express 5 | Proven, flexible, matches existing apps (P&L Analyzer, EscrowService) |
+| Frontend | React 19 + Vite | Fast dev server, clean build, matches existing apps |
+| Language | TypeScript (strict) | Build-time error catching, enables verify loop |
+| ORM | Prisma | Type-safe, schema-as-code, auto-migrations |
+| Database | PostgreSQL | Production-grade, Railway-native |
+| Styling | Tailwind CSS + shadcn/ui | AI-friendly utility classes + polished pre-built components |
+| Runtime | Node.js 20+ | Standard, Railway-supported |
+
+If someone needs Python, mobile, or something outside this stack — the agent says "this isn't what I build" rather than producing mediocre output.
+
+---
+
+## Pipeline: User Prompt → Deployed App
+
+```
+┌─────────────────────────────────────────────────────────┐
+│  USER: "Build me an expense tracker with auth and       │
+│         a dashboard that shows spending by category"    │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 1: SCAFFOLD (deterministic, no AI)               │
+│                                                         │
+│  Creates ~/Desktop/APPs/{app-name}/                     │
+│  ├── backend/                                           │
+│  │   ├── package.json (Express 5, Prisma, cors, etc.)  │
+│  │   ├── tsconfig.json                                  │
+│  │   ├── prisma/schema.prisma (base config)            │
+│  │   └── src/index.ts (Express boilerplate)            │
+│  ├── frontend/                                          │
+│  │   ├── package.json (React 19, Vite, Tailwind)       │
+│  │   ├── tsconfig.json                                  │
+│  │   ├── vite.config.ts                                │
+│  │   ├── tailwind.config.ts                            │
+│  │   └── src/ (base App.tsx, main.tsx, index.css)      │
+│  └── .gitignore                                         │
+│                                                         │
+│  Runs: npm install in both dirs                         │
+│  Exit: all config files exist, npm install succeeded    │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 2: PLAN (AI, quick)                              │
+│                                                         │
+│  Claude analyzes the user's request and produces:       │
+│  - Database schema (Prisma models)                      │
+│  - API routes needed                                    │
+│  - Frontend pages/components                            │
+│  - Auth approach (if needed)                            │
+│  - 10-point architectural plan                          │
+│                                                         │
+│  Reports plan to user in chat (transparency)            │
+│  Exit: plan produced, user can see it                   │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 3: BUILD (AI, iterative)                         │
+│                                                         │
+│  Claude writes all application code:                    │
+│  - Prisma schema + seed data                            │
+│  - Express routes, middleware, controllers              │
+│  - React components, pages, hooks                       │
+│  - shadcn/ui component usage                            │
+│  - Auth (if needed)                                     │
+│                                                         │
+│  Uses Write/Edit/Bash tools                             │
+│  Follows app_patterns skill (YOUR conventions)          │
+│  Applies relevant sub-skills (auth, multitenant, etc.)  │
+│  Exit: all files written                                │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 4: VERIFY (autonomous loop)                      │
+│                                                         │
+│  Loop:                                                  │
+│    1. Run prisma migrate dev                            │
+│    2. Run npm run build (backend)                       │
+│    3. Run npm run build (frontend)                      │
+│    4. If any fails → read error → fix → goto 1         │
+│    5. Max 5 retries per step                            │
+│                                                         │
+│  Exit checklist:                                        │
+│  [ ] prisma migrate succeeds                            │
+│  [ ] backend builds clean (tsc)                         │
+│  [ ] frontend builds clean (vite build)                 │
+│  [ ] no TypeScript errors                               │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 5: PREVIEW (local, for user)                     │
+│                                                         │
+│  Start both servers:                                    │
+│  - Backend: npm run dev (port 3001)                     │
+│  - Frontend: npm run dev (port 5173)                    │
+│                                                         │
+│  Report to user:                                        │
+│  "Preview ready at http://localhost:5173"                │
+│  Canvas shows iframe of the running app                 │
+│                                                         │
+│  STOP and wait for user response                        │
+│  User reviews → requests changes or says "deploy"       │
+│                                                         │
+│  If changes requested → goto PHASE 3 (modify mode)     │
+│  If "deploy" → goto PHASE 6                            │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 6: DEPLOY (autonomous)                           │
+│                                                         │
+│  1. git init + git add + git commit                     │
+│  2. Create GitHub repo (agenticledger org)              │
+│  3. git push                                            │
+│  4. railway init + railway up                           │
+│  5. Set env vars (DATABASE_URL, etc.)                   │
+│  6. Wait for deployment                                 │
+│  7. Check railway logs for errors                       │
+│  8. If errors → read logs → fix code → push → redeploy │
+│  9. Loop until healthy                                  │
+│                                                         │
+│  Exit checklist:                                        │
+│  [ ] GitHub repo created and code pushed                │
+│  [ ] Railway services running (API + frontend)          │
+│  [ ] Database provisioned and migrated                  │
+│  [ ] Health check passes                                │
+│  [ ] Public URL accessible                              │
+└────────────────────────┬────────────────────────────────┘
+                         │
+                         ▼
+┌─────────────────────────────────────────────────────────┐
+│  PHASE 7: REGISTER (platform integration)               │
+│                                                         │
+│  1. Register app via create_app MCP tool                │
+│     - name, URL, GitHub repo, Railway, category         │
+│  2. Create developer agent via create_agent MCP tool    │
+│     - workspace = app directory                         │
+│     - CLAUDE.md = tailored to this app                  │
+│     - heartbeat = "check for issues, fix bugs"          │
+│  3. Report to user:                                     │
+│     "Live at https://app-name.up.railway.app"           │
+│     "Developer agent @app-dev created to maintain it"   │
+│                                                         │
+│  Exit: app in registry, developer agent active          │
+└─────────────────────────────────────────────────────────┘
+```
+
+---
+
+## Skill Architecture
+
+### Core Skills (the vanilla framework)
+
+| Skill | Purpose | Type |
+|-------|---------|------|
+| `app_build_orchestrator` | Master runbook — the agent reads this first, follows phases in order, checks exit criteria at each step | Orchestration |
+| `app_scaffold` | Deterministic project scaffolding — writes all config files, runs npm install | Deterministic |
+| `app_patterns` | YOUR architecture patterns — Express route structure, Prisma conventions, React component patterns, auth approach | Patterns |
+| `app_verify` | Build verification loop — run builds, read errors, fix, retry | Verification |
+| `app_deploy` | GitHub + Railway deployment — git push, railway up, check logs, fix, redeploy | Deployment |
+| `app_register` | Platform registration — create_app MCP + create_agent MCP | Integration |
+
+### Sub-Skills (specialized patterns, added over time)
+
+| Sub-Skill | When Applied | What It Does |
+|-----------|-------------|--------------|
+| `app_pattern_auth` | User mentions login, users, accounts | Adds NextAuth/Passport, user model, protected routes |
+| `app_pattern_multitenant` | User mentions orgs, tenants, teams | Adds org model, tenant isolation, scoped queries |
+| `app_pattern_ai_agent` | User wants AI features, chat | Adds Anthropic SDK, chat endpoint, streaming |
+| `app_pattern_payments` | User mentions billing, subscriptions | Adds Stripe integration, webhook handler |
+| `app_pattern_realtime` | User mentions live updates | Adds WebSocket/SSE, real-time data push |
+| `app_pattern_file_upload` | User mentions documents, images | Adds multer, S3/local storage, file preview |
+| `app_pattern_dashboard` | User wants analytics, charts | Adds Recharts, stat cards, date range filters |
+| `app_pattern_email` | User mentions notifications, email | Adds Resend/Nodemailer, email templates |
+| `app_pattern_api_keys` | User wants external API access | Adds API key management, rate limiting |
+
+The agent reads the user's prompt, determines which sub-skills are relevant, and layers them onto the base scaffold during the BUILD phase.
+
+---
+
+## Orchestrator Design (Master Skill)
+
+The `app_build_orchestrator` skill is the state machine. Each phase has:
+
+```markdown
+## Phase N: {Name}
+
+### Entry Criteria
+- [what must be true before starting this phase]
+
+### Actions
+1. [step one]
+2. [step two]
+...
+
+### Exit Checklist
+- [ ] [condition 1]
+- [ ] [condition 2]
+
+### Max Retries: N
+
+### On Failure
+- [what to do if exit checklist fails after max retries]
+- Either: fix and retry, or escalate to user
+```
+
+The agent reads the orchestrator, executes phase by phase, and STOPS between Phase 5 (Preview) and Phase 6 (Deploy) to wait for user confirmation. This is the only human-in-the-loop checkpoint.
+
+---
+
+## Scaffold Templates (Deterministic Files)
+
+These files are NEVER written by AI — they're templated with variables:
+
+### Backend
+- `package.json` — Express 5, Prisma, cors, dotenv, TypeScript, ts-node-dev
+- `tsconfig.json` — strict mode, ES2022, node module resolution
+- `prisma/schema.prisma` — PostgreSQL provider, base config
+- `src/index.ts` — Express app boilerplate with CORS, JSON parsing, error handler
+- `.env.example` — DATABASE_URL, PORT templates
+
+### Frontend
+- `package.json` — React 19, Vite, Tailwind CSS, shadcn/ui deps
+- `tsconfig.json` — strict, JSX react-jsx
+- `vite.config.ts` — proxy to backend API, port config
+- `tailwind.config.ts` — content paths, shadcn preset
+- `postcss.config.mjs` — tailwind + autoprefixer
+- `src/main.tsx` — React root mount
+- `src/App.tsx` — Router setup skeleton
+- `src/index.css` — @tailwind directives
+- `components.json` — shadcn/ui config
+
+### Root
+- `.gitignore` — node_modules, dist, .env, .prisma
+
+---
+
+## What Makes This Different from FinneyBuilder
+
+| Aspect | FinneyBuilder | Our App Creator |
+|--------|--------------|-----------------|
+| Sandbox | E2B cloud VM ($) | Local machine (free) |
+| Stack | Next.js (one codebase) | Express + React (matches YOUR apps) |
+| AI | Claude API with custom tool loop | Claude Code agent (native tools) |
+| Database | Supabase tracking layer | No tracking DB needed |
+| Skills | Basic prompt injection | Full skill system with sub-skills |
+| Preview | E2B sandbox URL | localhost iframe in Lab canvas |
+| Deploy | Railway via GraphQL API | Railway CLI via Bash |
+| Quality | AI-generic output | YOUR patterns via curated skills |
+| Multi-tenant | Yes (orgs, BYOI) | No (single user) |
+| Modify/Fix | Separate modes | Same agent, ongoing conversation |
+
+---
+
+## Canvas Integration
+
+The Lab canvas panel serves dual purpose during app creation:
+
+1. **During BUILD phase** — shows code artifacts (file contents) as they're written
+2. **During PREVIEW phase** — switches to iframe showing `http://localhost:5173`
+3. **During DEPLOY phase** — shows deployment logs and status
+
+The canvas auto-opens when the first artifact appears and switches modes as the build progresses.
+
+---
+
+## Developer Agent (Post-Deploy)
+
+After deployment, a developer agent is automatically created:
+
+```json
+{
+  "name": "Expense Tracker Dev",
+  "workspace": "~/Desktop/APPs/expense-tracker",
+  "tools": ["Read", "Edit", "Write", "Glob", "Grep", "Bash"],
+  "mcps": ["github", "myaiforone"],
+  "heartbeatInstructions": "Check the GitHub repo for new issues. Check Railway logs for errors. If anything needs fixing, fix it and push.",
+  "org": [{ "organization": "Apps", "function": "Development", "title": "Developer" }]
+}
+```
+
+This agent maintains the app going forward — the user can chat with it to make changes, and its heartbeat keeps the app healthy.
+
+---
+
+## What to Build (Implementation Order)
+
+### Phase 1: Core Skills
+1. `app_build_orchestrator` — master runbook
+2. `app_scaffold` — deterministic scaffolding templates
+3. `app_patterns` — extracted from existing apps (P&L Analyzer, EscrowService)
+4. `app_verify` — build + fix loop
+5. `app_deploy` — GitHub + Railway pipeline
+6. `app_register` — platform integration
+
+### Phase 2: App Creator Agent Update
+1. Update `appcreator` CLAUDE.md — reference the orchestrator skill
+2. Add all skills to the appcreator agent config
+3. Add Railway CLI access (Bash tool already available)
+4. Add GitHub MCP to appcreator
+
+### Phase 3: Lab Canvas Enhancement
+1. Add iframe rendering capability to canvas
+2. Auto-switch canvas to iframe when dev server starts
+3. Add deployment status view in canvas
+
+### Phase 4: Sub-Skills (iterative, over time)
+1. `app_pattern_auth` — first sub-skill (most common need)
+2. `app_pattern_dashboard` — second (visual impact)
+3. Others as needed based on usage
+
+### Phase 5: Test & Refine
+1. Build a test app end-to-end using the system
+2. Identify where the agent gets stuck
+3. Refine skills based on real failures
+4. Repeat
+
+---
+
+## Future: Autonomy Integration
+
+This pipeline is designed to work with the future autonomy system:
+
+1. **User creates a goal**: "Build me 3 internal tools this week"
+2. **Agent proposes tasks**: "Task 1: Expense Tracker, Task 2: Inventory Manager, Task 3: Time Logger"
+3. **User approves via inbox**: approves all 3
+4. **Heartbeat triggers**: agent picks up Task 1, runs the full build pipeline autonomously
+5. **Reports back**: "Expense Tracker deployed at https://..."
+6. **Next heartbeat**: picks up Task 2, builds it
+7. **User reviews**: checks deployed apps, requests modifications
+
+The `source` field on tasks tracks that these were agent-proposed. The heartbeat system provides the execution loop. The orchestrator skill provides the build discipline.

package/docs/platform-mcp-audit.md

@@ -0,0 +1,320 @@
+# Platform MCP Audit — Hub Agent Readiness
+
+> Every action the hub agent needs to perform must be available as an MCP tool.
+> This audit maps every platform API endpoint to its MCP tool coverage.
+> **Status**: ✅ = covered | ❌ = missing | ⚠️ = partial
+
+---
+
+## Summary
+
+| Category | API Endpoints | MCP Tools | Gaps |
+|----------|:---:|:---:|:---:|
+| Agents | 8 | 8 | 0 |
+| Chat & Streaming | 6 | 5 | 1 |
+| Sessions | 3 | 3 | 0 |
+| Tasks & Projects | 7 | 7 | 0 |
+| Goals | 5 | 5 | 0 |
+| Cron Jobs | 5 | 5 | 0 |
+| Automations (list) | 1 | 1 | 0 |
+| Skills | 3 | 3 | 0 |
+| MCPs | 2 | 2 | 0 |
+| MCP Keys & Connections | 5 | 5 | 0 |
+| Marketplace / Registry | 9 | 9 | 0 |
+| Channels | 7 | 7 | 0 |
+| Memory | 3 | 3 | 0 |
+| Activity & Logs | 3 | 3 | 0 |
+| Model Overrides | 3 | 3 | 0 |
+| Cost | 2 | 2 | 0 |
+| Files | 3 | 3 | 0 |
+| Apps | 5 | 5 | 0 |
+| Accounts / Auth | 6 | 6 | 0 |
+| Service Config | 2 | 2 | 0 |
+| Pairing | 3 | 3 | 0 |
+| Heartbeat | 2 | 2 | 0 |
+| Delegation | 1 | 1 | 0 |
+| Webhooks | 1 | 1 | 0 |
+| SaaS Integration | 4 | 4 | 0 |
+| Dashboard | 2 | 2 | 0 |
+| Platform Agents (Lab) | 1 | 1 | 0 |
+| Utilities | 3 | 3 | 0 |
+| **TOTAL** | **105** | **104** | **1** |
+
+> Note: ~21 HTML page routes excluded (not API actions). Only API endpoints counted.
+
+---
+
+## Detailed Mapping
+
+### Agents
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 1 | List all agents | `GET /api/agents` | `list_agents` | ✅ |
+| 2 | Get agent details | `GET /api/agents/:id` | `get_agent` | ✅ |
+| 3 | Get agent instructions | `GET /api/agents/:id/instructions` | `get_agent_instructions` | ✅ |
+| 4 | Create agent | `POST /api/agents` | `create_agent` | ✅ |
+| 5 | Update agent | `PUT /api/agents/:id` | `update_agent` | ✅ |
+| 6 | Delete agent | `DELETE /api/agents/:id` | `delete_agent` | ✅ |
+| 7 | Recover agent | `POST /api/agents/:agentId/recover` | `recover_agent` | ✅ |
+| 8 | Get agent registry | `GET /api/agent-registry` | `get_agent_registry` | ✅ |
+
+### Chat & Streaming
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 9 | Send message (sync) | `POST /api/chat/:agentId` | `send_message` | ✅ |
+| 10 | Start streaming chat | `POST /api/chat/:agentId/stream` | `start_stream` | ✅ |
+| 11 | Get stream output (SSE) | `GET /api/chat/jobs/:jobId/stream` | — | ❌ |
+| 12 | Get raw job output | `GET /api/chat/jobs/:jobId/raw` | `get_chat_job_raw` | ✅ |
+| 13 | Stop chat job | `POST /api/chat/jobs/:jobId/stop` | `stop_chat_job` | ✅ |
+| 14 | Delegate to agent | `POST /api/delegate` | `delegate_message` | ✅ |
+
+> **Gap note:** SSE stream endpoint is browser-only; MCP clients use `get_chat_job_raw` polling instead. **Not a real gap** — can reclassify as N/A.
+
+### Sessions
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 15 | List sessions | `GET /api/agents/:agentId/sessions` | `list_sessions` | ✅ |
+| 16 | Reset session | `POST /api/agents/:agentId/sessions/reset` | `reset_session` | ✅ |
+| 17 | Delete session | `DELETE /api/agents/:agentId/sessions/:senderId` | `delete_session` | ✅ |
+
+### Tasks & Projects
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 18 | List agent tasks | `GET /api/agents/:id/tasks` | `list_tasks` | ✅ |
+| 19 | Create task | `POST /api/agents/:id/tasks` | `create_task` | ✅ |
+| 20 | Update task | `PUT /api/agents/:id/tasks/:taskId` | `update_task` | ✅ |
+| 21 | Delete task | `DELETE /api/agents/:id/tasks/:taskId` | `delete_task` | ✅ |
+| 22 | Get all tasks (cross-agent) | `GET /api/tasks/all` | `get_all_tasks` | ✅ |
+| 23 | Get task stats | `GET /api/agents/:id/tasks/stats` | `get_task_stats` | ✅ |
+| 24 | Create project | `POST /api/agents/:id/projects` | `create_project` | ✅ |
+
+### Goals
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 25 | Create goal | `POST /api/agents/:id/goals` | `create_goal` | ✅ |
+| 26 | Toggle goal | `POST /api/agents/:id/goals/:goalId/toggle` | `toggle_goal` | ✅ |
+| 27 | Trigger goal | `POST /api/agents/:id/goals/:goalId/trigger` | `trigger_goal` | ✅ |
+| 28 | Delete goal | `DELETE /api/agents/:id/goals/:goalId` | `delete_goal` | ✅ |
+| 29 | Get goal history | `GET /api/agents/:id/goals/:goalId/history` | `get_goal_history` | ✅ |
+
+### Cron Jobs
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 30 | Create cron | `POST /api/agents/:id/cron` | `create_cron` | ✅ |
+| 31 | Toggle cron | `POST /api/agents/:id/cron/:index/toggle` | `toggle_cron` | ✅ |
+| 32 | Trigger cron | `POST /api/agents/:id/cron/:index/trigger` | `trigger_cron` | ✅ |
+| 33 | Delete cron | `DELETE /api/agents/:id/cron/:index` | `delete_cron` | ✅ |
+| 34 | Get cron history | `GET /api/agents/:id/cron/:index/history` | `get_cron_history` | ✅ |
+
+### Automations
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 35 | List all automations | `GET /api/automations` | `list_automations` | ✅ |
+
+### Skills
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 36 | Create skill | `POST /api/skills/create` | `create_skill` | ✅ |
+| 37 | Get agent skills | `GET /api/agents/:agentId/skills` | `get_agent_skills` | ✅ |
+| 38 | Get org skills | `GET /api/skills/org/:orgName` | `get_org_skills` | ✅ |
+
+### MCPs
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 39 | List MCPs | `GET /api/mcps` | `list_mcps` | ✅ |
+| 40 | Get MCP catalog | `GET /api/mcp-catalog` | `get_mcp_catalog` | ✅ |
+
+### MCP Keys & Connections
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 41 | List MCP keys | `GET /api/agents/:id/mcp-keys` | `list_mcp_keys` | ✅ |
+| 42 | Save MCP key | `POST /api/agents/:id/mcp-keys` | `save_mcp_key` | ✅ |
+| 43 | Delete MCP key | `DELETE /api/agents/:id/mcp-keys/:mcpName` | `delete_mcp_key` | ✅ |
+| 44 | List MCP connections | `GET /api/agents/:id/mcp-connections` | `list_mcp_connections` | ✅ |
+| 45 | Create MCP connection | `POST /api/agents/:id/mcp-connections` | `create_mcp_connection` | ✅ |
+| 46 | Delete MCP connection | `DELETE /api/agents/:id/mcp-connections/:instanceName` | `delete_mcp_connection` | ✅ |
+
+### Marketplace / Registry
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 47 | Browse by type | `GET /api/marketplace/:type` | `browse_registry` | ✅ |
+| 48 | Install item | `POST /api/marketplace/install` | `install_registry_item` | ✅ |
+| 49 | Assign to agent | `POST /api/marketplace/assign` | `assign_to_agent` | ✅ |
+| 50 | Set platform default | `POST /api/marketplace/platform-default` | `set_platform_default` | ✅ |
+| 51 | Scan skills | `GET /api/marketplace/scan-skills` | `scan_skills` | ✅ |
+| 52 | Import skills | `POST /api/marketplace/import-skills` | `import_skills` | ✅ |
+| 53 | Create prompt | `POST /api/marketplace/create-prompt` | `create_prompt` | ✅ |
+| 54 | Get prompt trigger | `GET /api/marketplace/prompt-trigger` | `get_prompt_trigger` | ✅ |
+| 55 | Set prompt trigger | `POST /api/marketplace/prompt-trigger` | `set_prompt_trigger` | ✅ |
+| 56 | Add MCP to registry | `POST /api/marketplace/add-mcp` | `add_mcp_to_registry` | ✅ |
+
+### Channels
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 57 | List channels | `GET /api/channels` | `list_channels` | ✅ |
+| 58 | Update channel | `PUT /api/channels/:channelName` | `update_channel` | ✅ |
+| 59 | Add agent route | `POST /api/channels/:channelName/agents` | `add_agent_route` | ✅ |
+| 60 | Remove agent route | `DELETE /api/channels/:channelName/agents/:agentId` | `remove_agent_route` | ✅ |
+| 61 | Add monitored chat | `POST /api/channels/:channelName/monitored` | `add_monitored_chat` | ✅ |
+| 62 | Remove monitored chat | `DELETE /api/channels/:channelName/monitored` | `remove_monitored_chat` | ✅ |
+| 63 | Get sticky routing | `GET /api/sticky-routing` | `get_sticky_routing` | ✅ |
+
+### Memory
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 64 | Get agent memory | `GET /api/agents/:agentId/memory` | `get_agent_memory` | ✅ |
+| 65 | Search memory | `POST /api/agents/:agentId/memory/search` | `search_memory` | ✅ |
+| 66 | Clear memory context | `DELETE /api/agents/:agentId/memory/context` | `clear_memory_context` | ✅ |
+
+### Activity & Logs
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 67 | Get activity feed | `GET /api/activity` | `get_activity` | ✅ |
+| 68 | Get agent logs | `GET /api/agents/:agentId/logs` | `get_agent_logs` | ✅ |
+| 69 | Get changelog | `GET /api/changelog` | `get_changelog` | ✅ |
+
+### Model Overrides
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 70 | Get model | `GET /api/agents/:agentId/model` | `get_model` | ✅ |
+| 71 | Set model | `PUT /api/agents/:agentId/model` | `set_model` | ✅ |
+| 72 | Clear model | `DELETE /api/agents/:agentId/model` | `clear_model` | ✅ |
+
+### Cost
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 73 | Get agent cost | `GET /api/agents/:agentId/cost` | `get_agent_cost` | ✅ |
+| 74 | Get all costs | `GET /api/cost/all` | `get_all_costs` | ✅ |
+
+### Files
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 75 | Upload file (JSON/base64) | `POST /api/upload/:agentId/json` | `upload_file` | ✅ |
+| 76 | List agent files | `GET /api/agents/:agentId/files` | `list_agent_files` | ✅ |
+| 77 | Download file | `GET /api/agents/:agentId/download` | `download_agent_file` | ✅ |
+
+### Apps
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 78 | List apps | `GET /api/apps` | `list_apps` | ✅ |
+| 79 | Create app | `POST /api/apps` | `create_app` | ✅ |
+| 80 | Update app | `PUT /api/apps/:id` | `update_app` | ✅ |
+| 81 | Delete app | `DELETE /api/apps/:id` | `delete_app` | ✅ |
+| 82 | Check app health | `POST /api/apps/:id/check-health` | `check_app_health` | ✅ |
+
+### Accounts / Auth
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 83 | List accounts | `GET /api/config/accounts` | `list_accounts` | ✅ |
+| 84 | Add account | `POST /api/config/accounts` | `add_account` | ✅ |
+| 85 | Delete account | `DELETE /api/config/accounts/:name` | `delete_account` | ✅ |
+| 86 | Check account status | `GET /api/config/accounts/:name/status` | `check_account_status` | ✅ |
+| 87 | Start login | `POST /api/config/accounts/login` | `start_account_login` | ✅ |
+| 88 | Submit login code | `POST /api/config/accounts/login/code` | `submit_login_code` | ✅ |
+
+### Service Config
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 89 | Get service config | `GET /api/config/service` | `get_service_config` | ✅ |
+| 90 | Update service config | `PUT /api/config/service` | `update_service_config` | ✅ |
+
+### Pairing
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 91 | List paired senders | `GET /api/pairing` | `list_paired_senders` | ✅ |
+| 92 | Pair sender | `POST /api/pairing` | `pair_sender` | ✅ |
+| 93 | Unpair sender | `DELETE /api/pairing/:senderKey` | `unpair_sender` | ✅ |
+
+### Heartbeat
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 94 | Trigger heartbeat | `POST /api/agents/:id/heartbeat` | `trigger_heartbeat` | ✅ |
+| 95 | Get heartbeat history | `GET /api/agents/:id/heartbeat-history` | `get_heartbeat_history` | ✅ |
+
+### Webhooks
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 96 | Send webhook | `POST /webhook/:agentId` | `send_webhook` | ✅ |
+
+### Dashboard
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 97 | Get dashboard | `GET /api/dashboard` | `get_dashboard` | ✅ |
+| 98 | Browse directories | `GET /api/browse-dirs` | `browse_dirs` | ✅ |
+
+### Platform Agents (Lab)
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 99 | Get platform agents | `GET /api/platform-agents` | `get_platform_agents` | ✅ |
+
+### Utilities
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 100 | Health check | `GET /health` | `health_check` | ✅ |
+| 101 | Install xbar | `POST /api/install-xbar` | `install_xbar` | ✅ |
+| 102 | Whoami | `GET /api/whoami/:agentId` | `whoami` | ✅ |
+
+### SaaS Integration
+
+| # | Action | API Endpoint | MCP Tool | Status |
+|---|--------|-------------|----------|:---:|
+| 103 | Get SaaS config | `GET /api/saas/config` | `get_saas_config` | ✅ |
+| 104 | Update SaaS config | `PUT /api/saas/config` | `update_saas_config` | ✅ |
+| 105 | Test SaaS connection | `POST /api/saas/test` | `test_saas_connection` | ✅ |
+| 106 | Publish to SaaS | `POST /api/saas/publish` | `publish_to_saas` | ✅ |
+
+---
+
+## Gap Analysis
+
+### ✅ All Gaps Closed
+
+**Phase 1** (2026-04-02 — API coverage):
+- `get_saas_config` — Get SaaS connection settings
+- `update_saas_config` — Configure SaaS base URL and API key
+- `test_saas_connection` — Test SaaS credentials
+- `publish_to_saas` — Publish skill/prompt/agent/app to SaaS platform
+- `upload_file` — Upload a file to an agent's storage (base64/JSON endpoint)
+
+**Phase 2** (2026-04-02 — Hub agent readiness):
+- `list_capabilities` — Structured discovery of all platform capabilities by category
+- `write_memory` — Write to an agent's context.md or daily journal
+- `get_skill_content` — Read full skill file content (markdown)
+- `update_goal` — Modify an existing goal's config without delete+recreate
+- `update_cron` — Modify an existing cron's schedule/message without delete+recreate
+- `restart_service` — Trigger service restart after config changes
+- `get_user_guide` — Full platform user guide as markdown
+
+### Remaining Non-Gaps (1)
+
+| # | Action | API Endpoint | Status | Notes |
+|---|--------|-------------|:---:|-------|
+| 1 | **SSE stream** | `GET /api/chat/jobs/:jobId/stream` | N/A | Browser-only; MCP uses `get_chat_job_raw` polling instead |
+
+### ✅ Total MCP Tools: **113** | Coverage: **100%** of actionable endpoints + 7 hub-agent-specific tools