ToolsXpo Code

Quickstart: your first task with ToolsXpo Code

Install the extension, sign in from your browser, and give the agent its first task — about five minutes end to end.

Agent and scout tiers

Two reserved names — toolsxpo-agent and toolsxpo-scout — that let your workspace decide which model runs, without pinning it into the extension.

Bring your own key

Add your Anthropic, OpenAI or Google key once in the dashboard and every ToolsXpo app can use it. Why keys live there and not in your editor.

Checkpoints: undoing what the agent did

A snapshot before each edit, so any change can be rolled back. What they cost, and when to turn them off.

Choosing a model

What the model picker shows you, why some models are greyed out, and how to search a catalogue of hundreds with filters.

Context and compaction

Every model has a limit on how much it can hold. What happens as a task approaches it, and how to keep long sessions working.

Hooks: running your own scripts around the agent

Shell scripts that fire at points in the agent's lifecycle — formatting after an edit, blocking a dangerous command, logging what happened.

Local MCP servers

Running an MCP server as a local process — configuration, environment, and why a local server sees everything you do.

MCP: connecting the agent to your own services

What Model Context Protocol is, what the built-in ToolsXpo server gives you, and when to add your own.

Plan and Code modes

Plan explores and proposes; Code makes the changes. When to use each, and how to run a cheap model for planning and a strong one for the work.

Prompt caching

Providers cache the repeated part of your prompt and charge less for it. What ToolsXpo does automatically, and how to read the saving.

Remote MCP servers

Connecting to a hosted MCP server over HTTP, how authentication works, and what to check when one will not connect.

Rules: teaching the agent your conventions

Persistent instructions the agent follows in every conversation — project conventions, things never to do, how your team writes code.

Settings reference

Every setting in ToolsXpo Code, what it does, what it costs, and which ones sync between machines.

Signing in, and what your editor is allowed to do

How browser sign-in works, which workspace gets billed, what the editor can and cannot do with your account, and how to revoke it.

Skills: procedures the agent loads on demand

Reusable instruction sets the agent activates when a task matches, so a long procedure does not cost context on every request.

Slash commands

Built-in commands for reviewing, fixing, testing, explaining and committing — plus how to add your own.

Subagents and the scout tier

The agent delegates exploration to cheaper models that read and search but never write. Why this lowers cost without lowering quality.

Terminal and commands

How the agent runs commands, the two execution modes, approving what runs, and fixing a terminal that appears to hang.

The tool catalogue

230+ tools the agent can call directly — what is in there, how it decides to use one, and what each call costs.

ToolsXpo Code CLI

The same agent in your terminal — install, sign in, and run tasks non-interactively. Shares one account with the editor.

What an agent task costs

How a ToolsXpo Code request is priced, what makes one task cheaper than another, and where to see what you spent.

What we collect

ToolsXpo Code sends no analytics, no crash reports and no usage data. What that means, and why there is no setting for it.

Where your code goes

Your code goes to the model you chose, per request, so it can answer — and nowhere else. What that means in practice.