CLONEBOT is a chatbot built on Ollama (gpt-oss:20b) that doesn't just run the model — it wraps it in an intelligent system. Advanced context management, reasoning chains, a suite of tools — everything is tuned to squeeze the maximum from an open-source LLM and bring the experience close to frontier models.

No subscriptions. No API keys. No compromises on privacy.

Just your hardware, your data, your control.

Project status

• ✅ Module 1: Message Bus (Pub/Sub)
• ✅ Module 2: Session Store
• ✅ Module 3: Context Assembler
• ✅ Module 4: Model Adapter
• ✅ Module 5: Tool Executor
• ✅ Module 6: Agent Loop
• ✅ Module 7: Block Streamer (Markdown formatting)
• ✅ Module 8: Gateway Server (HTTP/WebSocket)
• ✅ Module 9: Telegram Adapter
• ✅ Module 10: Memory Store (long-term memory)

Features

• Agent loop execution: tool calling, multi‑turn runs, retries, and streaming events.
• Context assembly: session history, skills, tool availability, and optional user profile context.
• Tool execution: validated parameters, permissions, timeouts, result normalization, and audit hooks.
• Memory: session storage plus long‑term memory/user profile integration.
• Interfaces:
  • Web UI (streaming NDJSON)
  • Telegram bot integration
  • CLI runner for tests and checkpoints
• Artifacts: create and view generated documents, SVG, mermaid, web, and React outputs.
• Skills registry: load and execute skill tools with filtering and permissions.
• Gateway server: HTTP/WebSocket API for chat, artifacts, projects, and memory endpoints.

Core functionality

• Message Bus: pub/sub messaging across modules.
• Session Store: persistent chat sessions with artifacts and metadata.
• Model Adapter: provider‑agnostic model interface and streaming deltas.
• Context Assembler: system prompt build, tool list collection, token budgeting.
• Tool Executor: parameter schema validation, permission gating, timeouts, error handling.
• Block Streamer: markdown rendering for streamed responses.

Requirements

• Node.js + npm
• Ollama running on http://localhost:11434 (used by default)

Setup

1. Install dependencies

   npm install

2. Ensure Web UI static assets are present

   cp -r art/web/static/* web/static/

3. Start the Web UI server

   ./start-dev.sh

4. Stop the dev server

   ./stop-dev.sh

Installation

1. Clone the repository

   git clone <repo_url> cd CLONEBOT

2. Install dependencies

   npm install

3. Ensure Web UI static assets are present

   cp -r art/web/static/* web/static/

4. Start Ollama locally

   ollama serve

Run (three entry points)

1. Web UI (full integration)

   ./start-dev.sh

2. Gateway server only

   npm run start:gateway

3. CLI / Telegram

   npm run cli

   or Telegram (requires TELEGRAM_BOT_TOKEN)

   npm run start

Configuration

Environment variables used by the dev script:

• WEB_PORT (default: 3001)
• WEB_HOST (default: 127.0.0.1)
• DATA_DIR (default: ./data)

Optional runtime variables (CLI / gateway):

• TELEGRAM_BOT_TOKEN (Telegram bot token)
• GATEWAY_PORT (Gateway server port)

Scripts

• npm run build — TypeScript build
• npm test / npm run test:watch / npm run test:coverage
• npm run cli — CLI entry
• npm run chat — CLI chat
• npm run start:gateway — Gateway server
• npm run web — Web UI server
• npm run web:dev — Web UI on port 3000
• npm run checkpoint and npm run checkpoint:* — integration checkpoints

CLI

Entry point: cli.ts (commands: start, telegram, gateway, test, build, info).

Project structure (high level)

• src/ — core framework (agent loop, tools, adapters, gateway)
• web/ — Next.js app + static Web UI assets
• art/ — Artifact Studio and deep-thinking reference implementation
• skills/ — skills registry and examples
• data/ — chats, artifacts, projects, memory
• scripts/ — helper scripts

Tests

Jest configuration is in jest.config.js. Tests live under src/**/__tests__.

Contributing

Contributions are welcome. Keep changes focused and documented.

• Open an issue for bugs or feature proposals.
• Follow existing folder structure.
• Add or update tests when changing core behavior.
• For UI changes, include screenshots or short notes.
• Keep PRs small and single‑purpose when possible.