| title | Get Started with Bolt in Under a Minute |
|---|---|
| sidebarTitle | Quickstart |
| description | Install Bolt, open your project, connect a provider, and run your first AI prompt — all in under 60 seconds with this step-by-step guide. |
This guide walks you through installing Bolt, launching it in a project, connecting an AI provider, and running your first prompt. The whole flow takes under a minute.
The quickest path is the official install script, which works on macOS and Linux:```bash
curl -fsSL https://raw.githubusercontent.com/bolt-builder/bolt-cli/dev/install | bash
```
Prefer a package manager? Install globally with npm, bun, pnpm, or yarn:
```bash
npm i -g @bolt-builder/bolt-cli
```
For the desktop app on macOS (no Gatekeeper detour):
```bash
curl -fsSL https://raw.githubusercontent.com/bolt-builder/bolt-cli/dev/install-desktop | bash
```
See the [Installation guide](/installation) for all methods, including Homebrew, Chocolatey, Scoop, and Windows-specific notes.
```bash
cd my-project
bolt
```
Bolt picks up your project context automatically — reading your file tree, understanding the stack, and making it available to every agent from the moment you open it.
```bash
bolt providers
```
This opens the provider management flow where you can add API keys for OpenAI, Anthropic, and other supported LLM providers. Configuration is saved to `.bolt/bolt.jsonc` in your project root.
```jsonc
{
"$schema": "https://opencode.ai/config.json",
"provider": {
// Provider config goes here
}
}
```
You can also run prompts directly from the command line without opening the TUI:
```bash
# Non-interactive prompt using the default agent
bolt run "explain this codebase"
# Use a specific agent
bolt run --agent ask "what does this project do?"
# Attach the TUI to a running server (local or remote)
bolt attach <url>
```
Switch agents with Tab in the TUI, or pass --agent <name> to bolt run. Each agent has a defined access level to match its purpose.
| Agent | Access | Description |
|---|---|---|
code |
Full | Default: reads, writes, and runs code |
plan |
Read | Explores the codebase and produces a plan without changing files |
ask |
Read | Questions and research — no file edits |
code-review |
Read | Reviews changes for correctness, style, and security |
debug |
Full | Debugs failing tests, crashes, and logic errors |
refactor |
Full | Safe refactoring with test verification at each step |
docs |
Edit | Writes and updates documentation and comments |
security |
Read | Security audit: vulnerabilities, secrets, patterns |
migrate |
Full | Framework upgrades and dependency migrations |
perf |
Full | Performance analysis and optimization |
There are also two built-in subagents for delegation: @general for complex multi-step tasks and @explore for fast read-only codebase research. You can define your own agents as Markdown files or generate one with bolt agent create.
# Review staged changes with a pass/fail exit code
bolt review --staged
# Generate a commit message for staged changes
bolt commit
# List and manage sessions
bolt session list
# View token usage and cost statistics
bolt stats
# Query the local session database
bolt dbbolt --mini
bolt attach --mini <url>bolt run --auto "refactor this module"
bolt run --yolo "refactor this module"
bolt run --dangerously-skip-permissions "refactor this module"bolt completionFollow the printed instructions to source the completion script in your shell config.