Skip to main content

​
Quick start

This guide will help you get Docbot up and running quickly. You’ll set up Qdrant, initialize your project, index your docs, and run your first documentation task.

​
Prerequisites

Before starting, ensure you have:
  • Bun installed (version 1.3.0 or higher)
  • Docker installed (for local Qdrant)
  • AI_GATEWAY_API_KEY environment variable set
You can get an AI Gateway API key from Vercel AI Gateway. This is required for Docbot to function.

​
Step 1: Start Qdrant

Docbot requires Qdrant for vector storage. Start a local Qdrant instance using Docker: 
docker run --rm -p 6333:6333 -v "$(pwd)/qdrant_storage:/qdrant/storage" qdrant/qdrant
Keep this terminal open while using Docbot, or use the -d flag to run in detached mode.

​
Step 2: Initialize Docbot

Navigate to your project directory and run the init command: 
bunx @helmlabs/docbot init
This command will:
  • Create a .docbot/ directory for local state
  • Generate a docbot.config.jsonc configuration file
  • Set up Qdrant collections based on your project name
  • Add .docbot/ to your .gitignore
  • Optionally start a Qdrant Docker container for you
If you already have a docbot.config.jsonc file, use the --force flag to overwrite it:
bunx @helmlabs/docbot init --force

​
Configuration example

After initialization, your docbot.config.jsonc will look like this: 
{
  // project identifier used for qdrant collection naming
  "projectSlug": "my-project",

  // qdrant vector database settings
  "qdrant": {
    "url": "http://127.0.0.1:6333",
    "collections": {
      "docs": "docbot_my-project_docs",
      "code": "docbot_my-project_code"
    }
  }

  // uncomment to customize models:
  // "models": {
  //   "planning": "openai/gpt-5.2",
  //   "prose": "anthropic/claude-sonnet-4.5",
  //   "fast": "anthropic/claude-haiku-4.5"
  // }

  // uncomment to customize server:
  // "server": {
  //   "port": 3070
  // }
}

​
Step 3: Index your documentation

Index your docs and codebase to enable semantic search: 
bunx @helmlabs/docbot index --docs ./docs --codebase ./src
If you’ve configured paths.docs and paths.codebase in your config file, you can simply run:
bunx @helmlabs/docbot index
The indexing process will:
  1. Scan your documentation files
  2. Scan your codebase
  3. Create embeddings for semantic search
  4. Store the index in Qdrant
  5. Save a manifest to .docbot/manifest.json for incremental updates

​
Step 4: Run your first task

Now you’re ready to run Docbot with a documentation task: 
bunx @helmlabs/docbot run "document the settings page"
Docbot will:
1

Analyze

Read your docs and codebase to understand the current state
2

Plan

Generate a concrete plan with specific file operations
3

Review

Present the plan in the TUI for your approval
4

Execute

Apply the approved changes to your documentation files

​
Interactive mode

By default, Docbot runs in interactive mode with a terminal UI (TUI) where you can:
  • Review the proposed plan
  • Approve or reject changes
  • Monitor progress in real-time
  • Toggle verbose logging with Ctrl+L
To run without the interactive approval step, use --no-interactive:
bunx @helmlabs/docbot run "your task" --no-interactive

​
Common workflows

​
Update documentation for new features

bunx @helmlabs/docbot run "document the new API endpoints in src/api/"

​
Fix stale documentation

bunx @helmlabs/docbot run "update authentication docs to match current implementation"

​
Fill documentation gaps

bunx @helmlabs/docbot run "find and document undocumented configuration options"

​
Next steps

Installation

Learn about global installation and advanced setup options

Commands

Explore all available commands and their options

Configuration

Customize models, paths, and server settings

Search

Use Docbot’s semantic search capabilities

Build docs developers (and LLMs) love

Get started for freeTalk to us