Release v4.16.0 · deer-flow/llm-space

What's Changed

A spawn_agent built-in Tool requests a manually created subtask. Calls pause
automatic execution and ReAct loops; creating a subtask opens its tab and
expands its folder without starting a run or filling in the tool response.
Subtasks inherit prompt templates, variables, model settings, and tools suited
to general work, research, or code review. Readable task names become unique
filenames, and missing subtask files can be created again from the tool card.
An exec_code built-in Tool runs Python, JavaScript, and TypeScript in
persistent sessions, retaining variables and imports between calls.
calculator and date_difference built-in Tools handle mathematical
expressions and date or time intervals, with matching Python exports.

The General Agent template now includes exec_code and replaces its previous
agent() function with spawn_agent.
Python exports explicitly report that manual subtask creation requires the
desktop app; generated tools and project plans explain this limitation.
Scrollbars use consistent thumb styling throughout the app.
The remote runtime protocol is now v5. Update the server alongside the desktop
app to use this version.

Long conversations in the Performance edition keep consistent message spacing
when the page is zoomed.

Two editions. Install either, or both — they share the same ~/.llm-space data, so switching keeps your threads and settings.

LLM Space — uses the system WebView. Small download (~27 MB), light on memory and battery.
LLM Space Performance — embeds its own rendering engine (~130 MB). Rendering stays consistent across macOS versions, and usually performs better.

Download a DMG below. Existing installs update in place.

Full changelog: https://github.com/deer-flow/llm-space/blob/v4.16.0/CHANGELOG.md