Skip to content

10x-Anit/10x-Accountability-Coach

BranchesTags

Repository files navigation

10X Accountability Coach

Your Personal AI-Powered Accountability System by Team 10X

A dual-system accountability platform: Claude Code CLI as the brain + Next.js UI as the display. Create agents, challenges, and plans from the terminal — interact with them in the browser.

How It Works

┌─────────────────────────────────────────────────────────────┐
│                    10X Accountability Coach                   │
├─────────────────────────────────────────────────────────────┤
│                                                               │
│  ┌──────────────┐         ┌──────────────┐                   │
│  │  CLI Terminal │ writes  │  Local Files  │  reads           │
│  │  (Claude Code)│───────▶│  (./data/)    │◀────────┐        │
│  │  37 Commands  │         │  JSON + MD    │         │        │
│  └──────────────┘         └──────────────┘         │        │
│                                                     │        │
│  ┌──────────────┐         ┌──────────────┐         │        │
│  │  OpenAnalyst  │◀───────│  UI (Next.js) │─────────┘        │
│  │  API          │───────▶│  localhost:3000│                   │
│  │  (Dual AI)    │         │  Auto-refresh │                   │
│  └──────────────┘         └──────────────┘                   │
│                                                               │
└─────────────────────────────────────────────────────────────┘
  • CLI = Create agents, challenges, todos, plans, check-ins via slash commands
  • UI = View, interact, chat, complete tasks, reschedule in the browser
  • OpenAnalyst API = Powers both CLI and UI with context-aware AI
  • Local Files = All data in ./data/ — no cloud required

Features

  • 37 Slash Commands — Full CLI coverage for every feature
  • Dynamic Agent System — Create custom AI agents (fitness coach, nutritionist, etc.) with personality, skills, and references
  • Context-Aware Chat — Conversation history preserved across messages (last 10 messages, ~3000 token budget)
  • Chat Persistence — Messages saved to disk, survive page refresh
  • Challenge Tracking — 30-day challenges with streaks, daily tasks, and insights
  • 20+ AI Skills — Structured coaching operations (motivation, scheduling, check-ins, etc.)
  • Small-Model Optimized — Commands use tagged format (~500 tokens each) for efficient parsing
  • MCP Integration — Model Context Protocol for extensibility
  • Multi-Agent Chat — Select agents in sidebar, chat uses their personality
  • Vision Boards, Plans, Contracts — Full accountability toolkit

Quick Start

Option A: Using Claude Code (Recommended)

  1. Open this project folder in Claude Code
  2. Say: "start my app"
  3. Provide your API key when asked (get one at https://10x.in/dashboard)
  4. Claude Code handles everything — dependencies, config, and startup!

The app MUST run on port 3000. If the port is busy, setup will kill existing processes first.

Option B: Using OpenCode (Full Compatibility)

This project is a fully compatible OpenCode skills plugin:

  1. Clone this repo
  2. Run opencode in the project directory
  3. OpenCode auto-loads: opencode.json, .opencode/agents/, .opencode/commands/, .opencode/skills/
  4. Say: "start my app" — the setup agent handles everything
  5. Use any model (Victoire, Kimi 2.5, Claude, etc.)

Option C: Any AI CLI / Free Models

Commands work with any AI CLI that reads commands/ directory:

  1. Clone this repo
  2. Install: npm install && cd ui && npm install
  3. Create ui/.env.local (see Environment Setup below)
  4. Start: npx kill-port 3000 && cd ui && npm run dev
  5. Open http://localhost:3000

Option D: Manual Setup

1. Clone & Install

git clone https://github.com/Anit-1to10x/10x-Accountabilty-Coach.git
cd 10x-Accountability-Coach
npm install
cd ui && npm install

2. Environment Setup

Create ui/.env.local:

# Required: OpenAnalyst AI API
OPENANALYST_API_URL=https://api.openanalyst.com/api
OPENANALYST_API_KEY=sk-oa-v1-YOUR-KEY-HERE
OPENANALYST_MODEL=openanalyst-beta

# Optional: Data source (local or supabase)
DATA_SOURCE=local

# Optional: Gemini for image generation
# GEMINI_API_KEY=your-gemini-key

# Optional: Supabase for cloud sync
# SUPABASE_URL=your-url
# SUPABASE_ANON_KEY=your-key

Get your API key at: https://10x.in/dashboard (format: sk-oa-v1-xxx)

3. Start

npx kill-port 3000
cd ui && npm run dev

Open http://localhost:3000

All 37 Slash Commands

Streaks & Challenges

Command Description
/streak Check in to active challenge
/streak-new Create a new challenge
/streak-list List all challenges
/streak-stats View challenge statistics
/streak-switch Switch active challenge
/streak-insights Cross-challenge insights
/challenge Browse challenge templates

Agents

Command Description
/agent View all agents
/agent-new Create a custom AI agent
/agent-edit Edit agent personality, skills, prompts
/agent-skills Manage agent's assigned skills
/agent-delete Delete an agent
/agent-chat Chat with a specific agent
/agent-ref Add reference files to agent

Todos & Schedule

Command Description
/todo View all todos
/todo-add Add a new todo
/todo-list List todos with filters
/todo-complete Mark todo as done
/schedule View your schedule
/reschedule Reschedule a task

Check-ins & Motivation

Command Description
/checkin Daily accountability check-in
/checkin-history View check-in history
/motivate Get a motivation boost

Planning & Vision

Command Description
/plan View active plan
/plan-new Create a new plan
/visionboard View vision board
/visionboard-new Create a vision board

Accountability

Command Description
/contract Create/view accountability contract
/punishment Manage consequences

Skills & Tools

Command Description
/skill-list List all 20+ AI skills
/skill-new Create a custom skill
/prompt Manage prompt templates

Workspace & System

Command Description
/workspace View workspace overview
/asset Manage assets
/history View activity history
/onboard Run onboarding setup
/help Show all commands

Agent System

Create custom AI agents from the terminal or the UI:

# From Claude Code CLI
/agent-new "Fitness Coach"
# → Creates data/agents/fitness-coach/ with full workspace
# → Auto-assigns fitness skills (workout, nutrition)
# → Sets personality: energetic, motivating
# → Shows in UI sidebar instantly

/agent-ref fitness-coach --add
# → Add workout plans, meal templates as reference files
# → Agent uses these for context-aware responses

Agent personality presets: Drill Sergeant, Best Friend, Mentor, Hype Man, Professor, Custom

Agent types auto-detected: fitness, finance, learning, career, creative, general

How Agents Work

  1. Create via /agent-new (CLI) or "+" button (UI)
  2. Data stored in data/agents.json + data/agent-capabilities.json
  3. Workspace at data/agents/{id}/ (files, notes, chats, generated content)
  4. Select in UI sidebar → chat uses agent's personality + skills
  5. Reference files provide context-aware responses

Project Structure

10X-Accountability-Coach/
├── .claude/                     # Claude Code configuration
│   ├── commands/                # Slash commands (for Claude Code CLI)
│   └── instructions.md          # Claude Code startup behavior
│
├── .opencode/                   # OpenCode configuration
│   ├── agents/                  # Agent definitions (coach.md, setup.md)
│   ├── commands/                # Slash commands (for OpenCode CLI)
│   └── skills/                  # Skills with OpenCode frontmatter
│
├── opencode.json                # OpenCode config (loads CLAUDE.md)
│
├── commands/                    # Slash commands (for any CLI / free models)
│   ├── _format.md              # Tag format reference
│   ├── streak.md ... (6)       # Streak commands
│   ├── agent.md ... (7)        # Agent commands
│   ├── todo.md ... (4)         # Todo commands
│   ├── checkin.md ... (2)      # Check-in commands
│   ├── plan.md ... (2)         # Plan commands
│   ├── visionboard.md ... (2)  # Vision commands
│   └── help.md, etc.           # System commands
│
├── skills/                      # 20+ AI skill definitions
│   ├── streak/                  # Challenge tracking
│   ├── daily-checkin/           # Check-in flow
│   ├── motivation/              # Motivation engine
│   ├── workout-program-designer/# Fitness programs
│   ├── nutritional-specialist/  # Nutrition guidance
│   └── ...                      # 15+ more skills
│
├── data/                        # All user data (local storage)
│   ├── agents.json              # Agent registry
│   ├── agent-capabilities.json  # Agent skills/personality
│   ├── agents/                  # Agent workspaces
│   ├── config.json              # App configuration
│   ├── challenges/              # Challenge templates (30-day content)
│   ├── profiles/                # User profiles (runtime)
│   ├── todos/                   # User todos (runtime)
│   ├── checkins/                # Check-in records (runtime)
│   ├── chats/                   # Chat history (runtime)
│   └── schemas/                 # Supabase SQL schemas
│
├── ui/                          # Next.js 14 frontend
│   ├── .env.local              # API keys (git-ignored)
│   ├── app/                    # Pages & API routes
│   │   ├── api/chat/stream/    # SSE streaming with conversation history
│   │   ├── api/agents/         # Agent CRUD
│   │   ├── (shell)/            # 17+ app pages
│   │   └── ...
│   ├── components/             # React components
│   ├── lib/                    # Stores, utilities, data adapters
│   └── package.json
│
├── CLAUDE.md                    # Setup instructions (for AI assistants)
├── README.md                    # This file
└── package.json                 # Root dependencies

Command Format (Small-Model Optimized)

Commands use structured XML-like tags for efficient parsing (~500 tokens each instead of ~2000):

---
description: Short description for command matching
---
<cmd name="command-name">What it does</cmd>
<ref>skills/relevant-skill/SKILL.md</ref>

<steps>
1. Step one
2. Step two
</steps>

<data path="data/path/{id}.json">
{field1, field2, field3}
</data>

<output>What user sees after execution</output>

This format works with Claude Code, OpenCode, and free models (Victoire, Kimi 2.5, etc.).

CLI Compatibility

CLI Config Commands Skills Agents
Claude Code CLAUDE.md .claude/commands/ skills/ via commands
OpenCode opencode.json .opencode/commands/ .opencode/skills/ .opencode/agents/
Any CLI CLAUDE.md commands/ skills/ via commands

All three read the same data (./data/) and use the same UI (localhost:3000).

Context-Aware Chat

The chat system maintains conversation memory:

  • History: Last 10 messages sent to API with each request (~3000 token budget)
  • Persistence: Messages saved to data/chats/{date}/{agentId}.md
  • Profile Context: User profile, challenges, todos, streaks injected into system prompt
  • Agent Personality: Selected agent's tone, style, skills shape every response
  • Skill Matching: Messages auto-matched to relevant skills for structured behavior

API Configuration

Variable Required Description
OPENANALYST_API_URL Yes https://api.openanalyst.com/api
OPENANALYST_API_KEY Yes Your key (starts with sk-oa-v1-)
OPENANALYST_MODEL Yes openanalyst-beta
DATA_SOURCE No local (default) or supabase
GEMINI_API_KEY No For image generation
SUPABASE_URL No For cloud sync
SUPABASE_ANON_KEY No For cloud sync

Deployment

Vercel

  1. Connect GitHub repo to Vercel
  2. Add env vars in Vercel dashboard
  3. Deploy

Local Development

npm install && cd ui && npm install
npx kill-port 3000
cd ui && npm run dev

Troubleshooting

Issue Fix
"API Disconnected" Check ui/.env.local has valid key, restart server
Port 3000 in use npx kill-port 3000 3001 3002 then restart
Dependencies broken npx rimraf node_modules ui/node_modules then reinstall
Build errors cd ui && npx rimraf .next && npm run build
Chat has no memory Ensure history is being sent (check browser network tab)

Technology Stack

  • Frontend: Next.js 14, TypeScript, Tailwind CSS, Framer Motion, Zustand
  • Backend: Node.js, Next.js API Routes, Server-Sent Events
  • AI: OpenAnalyst API (Anthropic Messages API compatible)
  • Storage: File system (Markdown + JSON), optional Supabase
  • CLI: Claude Code, OpenCode (full plugin support), any AI CLI
  • MCP: Model Context Protocol for tool extensibility

License

+10x Team Proprietary - For use with 10x.in subscription.

Developed by Team 10X | Powered by OpenAnalyst

About

No description, website, or topics provided.

Resources

Readme

Security policy

Security policy
Activity

Stars

9 stars

Watchers

2 watching

Forks

3 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages