ClawdBot: Build Your Personal AI Assistant (2026)

ClawdBot is an open-source personal AI assistant framework that lets you build a customizable, always-available AI companion for your daily workflows. Whether you need help managing tasks, summarizing documents, writing code, or automating repetitive work, ClawdBot gives you the building blocks to create an assistant tailored to your needs.

This guide walks you through setting up ClawdBot from scratch, customizing its behavior, and connecting it to your favorite tools and services.

What Is ClawdBot?

ClawdBot is a modular AI assistant framework built on top of large language models. Unlike generic chatbots, ClawdBot is designed to be self-hosted, privacy-respecting, and deeply customizable. You define its personality, capabilities, and integrations.

Key Features

Feature	Description
Multi-model support	Connect to Claude, GPT, Gemini, or local models via Ollama
Plugin system	Extend functionality with first-party and community plugins
Memory management	Persistent conversation memory with configurable retention
Tool calling	Execute shell commands, API calls, and file operations
Multi-channel	Deploy to Discord, Slack, Telegram, or use via CLI
Privacy-first	Self-hosted with full control over your data

Prerequisites

Before you begin, make sure you have:

Node.js 20+ installed
npm or pnpm package manager
An API key from at least one LLM provider (Anthropic, OpenAI, or a local model via Ollama)
A terminal you are comfortable working in

Step 1: Install ClawdBot

Clone the repository and install dependencies:

git clone https://github.com/clawdbot/clawdbot.git
cd clawdbot
pnpm install

Copy the example environment file:

cp .env.example .env

Step 2: Configure Your LLM Provider

Open the .env file and add your API credentials. ClawdBot supports multiple providers simultaneously, so you can configure as many as you need.

# Primary provider
ANTHROPIC_API_KEY=sk-ant-your-key-here
DEFAULT_MODEL=claude-sonnet-4-20250514

# Optional: secondary providers
OPENAI_API_KEY=sk-your-openai-key
OLLAMA_BASE_URL=http://localhost:11434

Provider Comparison

Provider	Best For	Cost	Latency
Anthropic Claude	Complex reasoning, coding	$3-15/M tokens	Low
OpenAI GPT-5	General tasks, vision	$2-30/M tokens	Low
Google Gemini	Long context, multimodal	$1-7/M tokens	Medium
Ollama (local)	Privacy, offline use	Free (hardware cost)	Varies

Step 3: Define Your Assistant's Personality

ClawdBot uses a configuration file to define the assistant's behavior. Create or edit config/persona.yaml:

name: "Atlas"
personality:
  tone: professional
  verbosity: concise
  humor: subtle
instructions: |
  You are Atlas, a personal productivity assistant.
  Always prioritize actionable advice over theory.
  When asked about tasks, check the task list first.
  Format responses in markdown when helpful.
capabilities:
  - task_management
  - code_review
  - document_summary
  - web_search
  - calendar_integration

You can create multiple persona files and switch between them:

clawdbot --persona config/persona-coding.yaml
clawdbot --persona config/persona-writing.yaml

Step 4: Add Plugins

Plugins extend ClawdBot's capabilities. Install them from the plugin registry:

# Task management plugin
clawdbot plugin add @clawdbot/tasks

# Web search plugin
clawdbot plugin add @clawdbot/web-search

# Calendar integration
clawdbot plugin add @clawdbot/google-calendar

# Code execution sandbox
clawdbot plugin add @clawdbot/code-runner

Each plugin is configured in config/plugins.yaml:

plugins:
  tasks:
    storage: sqlite
    db_path: ./data/tasks.db
  web-search:
    provider: tavily
    api_key: ${TAVILY_API_KEY}
  google-calendar:
    credentials_path: ./config/google-credentials.json
  code-runner:
    languages: [python, javascript, bash]
    timeout: 30s
    sandbox: docker

Step 5: Set Up Memory and Context

ClawdBot can remember previous conversations and reference them in future interactions. Configure memory in config/memory.yaml:

memory:
  enabled: true
  backend: sqlite
  db_path: ./data/memory.db
  retention:
    conversations: 90d
    facts: permanent
    preferences: permanent
  context_window:
    max_messages: 50
    summary_threshold: 30

When the conversation exceeds the summary threshold, ClawdBot automatically summarizes older messages to preserve context while staying within token limits.

Step 6: Deploy to a Channel

CLI Mode (Default)

Start ClawdBot in your terminal:

clawdbot start

You can now chat directly in the terminal:

You: Summarize my tasks for today
Atlas: Here are your 3 pending tasks for today:
  1. Review PR #142 - frontend refactor (due 2pm)
  2. Write API documentation for /users endpoint
  3. Prepare slides for Thursday standup

Discord Bot

Add Discord credentials to your .env:

DISCORD_BOT_TOKEN=your-discord-bot-token
DISCORD_CHANNEL_ID=123456789

Start in Discord mode:

clawdbot start --channel discord

Slack Bot

SLACK_BOT_TOKEN=xoxb-your-slack-bot-token
SLACK_SIGNING_SECRET=your-signing-secret

clawdbot start --channel slack

Step 7: Create Custom Tools

You can define custom tools that ClawdBot can invoke during conversations. Create a tool file in tools/:

// tools/check-website-status.ts
import { defineTool } from '@clawdbot/sdk';

export default defineTool({
  name: 'check_website_status',
  description: 'Check if a website is up and return its HTTP status code',
  parameters: {
    url: {
      type: 'string',
      description: 'The URL to check',
      required: true,
    },
  },
  async execute({ url }) {
    const response = await fetch(url, { method: 'HEAD' });
    return {
      url,
      status: response.status,
      ok: response.ok,
      latency: response.headers.get('x-response-time'),
    };
  },
});

tools:
  custom:
    - ./tools/check-website-status.ts
    - ./tools/deploy-preview.ts

Practical Automation Examples

Morning Briefing

Set up a cron job to get a daily briefing:

# crontab entry - runs at 8am every day
0 8 * * * clawdbot run "Give me my morning briefing: today's calendar, pending tasks, and a weather summary" --output ./briefing.md

PR Review Automation

# In your CI pipeline
gh pr diff $PR_NUMBER | clawdbot run "Review this pull request diff. Flag potential bugs, security issues, and style problems. Be concise."

Document Summarization Pipeline

# Summarize all PDFs in a folder
for file in ./documents/*.pdf; do
  clawdbot run "Summarize this document in 3 bullet points" --attach "$file" >> summaries.md
done

Troubleshooting

Issue	Solution
"API key not found"	Verify your `.env` file has the correct key and the variable name matches
Memory database locked	Stop all ClawdBot instances, then restart
Plugin fails to load	Run `clawdbot plugin update` and check version compatibility
Slow responses	Switch to a faster model or use a local Ollama model for simple tasks
Context too long	Reduce `max_messages` in memory config or enable auto-summarization

Tips for Getting the Most Out of ClawdBot

Be specific in your persona instructions. The more context you give about how you want the assistant to behave, the better the results.
Use multiple personas. Create separate configurations for coding, writing, and personal use.
Leverage tool chaining. ClawdBot can call multiple tools in sequence to accomplish complex tasks.
Review memory regularly. Check what facts ClawdBot has stored and correct anything outdated.
Start with CLI mode. Get comfortable with the basics before deploying to Discord or Slack.

Conclusion

ClawdBot gives you the flexibility to build an AI assistant that fits your exact workflow. From simple task management to complex multi-tool automations, the modular architecture means you can start small and expand as your needs grow.

If you are working on AI-powered projects and need reliable infrastructure for image generation, video creation, or other media tasks, Hypereal AI provides production-ready APIs that integrate seamlessly with tools like ClawdBot. Check it out to add powerful media generation capabilities to your personal assistant setup.