Release v1.33.6CodeSplash AI

CodeSplash AI v1.33.6

CodeSplash AI 1.33.6 refreshes two of the most visible AI surfaces in the app: AI Chat and the Command Center home experience. This patch also improves how opening prompts are delivered to terminal-based agents, helping new agent sessions start more reliably when you launch them with an initial task.

July 25, 2026

Full releaseAll updates

Highlights

01

AI Chat got a major visual and workflow refresh with a redesigned panel, composer, thread list, message bubbles, and home view

02

Command Center’s home and agent launch experience now better matches the newer AI Vibes design direction

03

Attachment flows were expanded across chat and agent launch screens with pending lists and preview support

04

Agent launch behavior is smarter about how different CLIs accept prompts and run modes

05

Opening prompts are now primed more carefully for terminal-based agents so first-task delivery is more dependable

Changelog

What changed in v1.33.6

Features

  • New attachment UI for AI Chat, including pending attachment lists and attachment preview dialogs
  • New attachment handling in Command Center launch flows for preparing agent runs with attached context
  • Expanded agent branding and icon coverage across launcher surfaces for easier tool recognition

Improvements

  • Redesigned AI Chat to better match CodeSplash AI’s newer AI Vibes visual style
  • Refreshed Command Center home with a more modern, composer-driven launch experience
  • Updated chat workflow components including the composer, model selector, mode selector, thread list, and message presentation
  • Improved visual consistency between AI Chat, Command Center, and Vibe-oriented UI surfaces
  • Added resizable sidebar support and related layout polish in the refreshed chat experience

Fixes

  • Improved reliability when starting terminal-based agent sessions with an opening prompt by waiting for the agent UI to become ready before sending input
  • Reduced risk of prompts being swallowed or misinterpreted during agent startup with safer prompt priming and input sanitization
  • More accurate launcher behavior for agents with different prompt-delivery and run-mode requirements