Workspace & BillingToolsXpo Code Reference

Settings reference

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

Updated September 2, 2026

Settings live under the ToolsXpo panel's gear icon, and the ones below are also real VS Code settings — so they sync with Settings Sync, can be overridden per workspace or profile, and can be pushed by policy.

Edit them in either place; the two stay in step.

Agent

Setting	Default	What it does
`toolsxpo.agent.autoCompact`	on	Summarise older turns as a task nears the context window. Without it, long tasks fail outright.
`toolsxpo.agent.subagents`	on	Delegate exploration to cheap scouts. Usually lowers what a task costs.
`toolsxpo.agent.preferredLanguage`	English	Language the agent explains in. Code is unaffected.

Chat

Readability, not behaviour. A transcript after an hour is a different artefact from a single message.

Setting	Default	What it does
Collapse code blocks	off	Show edits as a summary line
Collapse terminal output	on	Show the command and status, fold the log
Collapse MCP responses	off	Fold large tool payloads
Summarise changed files	on	List every file touched when a task ends
Show cost per request	off	Print tokens and credits under each request
Compact layout	off	Tighter spacing, more of the session on screen

Editor

Setting	Default	What it does
`toolsxpo.editor.backgroundEdit`	on	Apply edits without stealing focus
`toolsxpo.editor.checkpoints`	on	Snapshot before each edit. Costs disk on a large monorepo.
`toolsxpo.editor.worktrees`	off	Allow real git worktrees for parallel tasks

Terminal

Setting	Default	What it does
Execution mode	Background exec	How commands run. See Terminal.
`toolsxpo.terminal.shellIntegrationTimeout`	4000 ms	How long to wait for shell integration
`toolsxpo.terminal.defaultProfile`	default	Which terminal profile to use

Tools & MCP

Setting	Default	What it does
`toolsxpo.mcp.displayMode`	rich	How MCP responses render
`toolsxpo.mcp.responsesCollapsed`	off	Fold responses by default

Privacy

There is no telemetry setting, because there is no telemetry. See What we collect.

First-run preset

After signing in you are asked one question — how autonomous the agent should be — and the answer writes a coherent set of the above. You can change any of it afterwards; nothing there is permanent.