Where to next

You've finished Quickstart and the first-run checklist. Here's how to pick the right rabbit hole based on what you actually want to do.

I just want to use Codebolt day-to-day

Go to Using Codebolt. It's reference documentation for every feature — the chat panel, tabs, checkpoints, context, Ctrl+K, the agent marketplace, MCP server management, the CLI, integrations. Read pages as you need them.

First reads in order of usefulness:

I want to accomplish a specific task

Go to Guides & Tutorials. Every guide has one concrete outcome. Examples:

I want to build on Codebolt

Go to Build on Codebolt. This is the extension surface: custom agents, agent extensions, plugins, custom interfaces, remote providers, orchestration design, server internals, self-hosting.

Suggested reading path by goal:

Goal	Start here
Build a custom agent for my codebase	Custom Agents → Overview → Quickstart
Give agents access to my internal APIs	MCP Tools → Overview
Connect a new LLM provider	Remote Environment Providers → LLM Provider
Extend the application without building a new agent	Plugins → Overview
Design a multi-agent workflow	Multi-Agent Orchestration → Overview
Understand how the server actually works	Internals → Architecture Overview
Self-host for my team	Self-Hosting → Overview

I need the exact signature / schema / flag

Go to Reference. Mostly auto-generated, always complete:

I'm curious how it works under the hood

The Internals section was written for you. Start with:

Architecture Overview — the five planes, the process model, where data lives.
Agent Subsystem — what an agent actually is.
Chat message end-to-end — follow one message all the way through.
Memory — the layer stack and the ingestion pipeline.

From there, pick subsystems as they become relevant.

I want to be opinionated about multi-agent

Start with When to use multi-agent — it will likely talk you out of multi-agent for your first project, which is the right answer most of the time. Then:

Plan-Execute-Review — the one pattern that covers most real cases.
Agent Flows — how to freeze a pattern into a reusable graph.
Review & Merge Design — human-in-the-loop without making humans the bottleneck.

Still not sure?

Default path: Using Codebolt → Overview. Everything else is accessible from there.

I just want to use Codebolt day-to-day​

I want to accomplish a specific task​

I want to build on Codebolt​

I need the exact signature / schema / flag​

I'm curious how it works under the hood​

I want to be opinionated about multi-agent​

Still not sure?​