What this guide covers
Workspace setup, Git and worktrees, Agent orchestration, All-in-one engineering modes.
Use CodeSplash AI as a full engineering desktop: connect your workspace, move between editor, terminal, Git, search, files, and run agent sessions without fragmenting the workflow.
Workspace setup, Git and worktrees, Agent orchestration, All-in-one engineering modes.
Work from top to bottom the first time, then use the section links as a reference once the product becomes part of your normal workflow.
The fastest path is to treat the app like your daily engineering home rather than as a single AI panel.
Download the platform for your OS, launch it, and start from the repository or workspace you already work in.
Open Editor, Files, Search, or Terminal based on what the task actually requires instead of forcing everything through a chat flow.
The product becomes significantly faster once shortcuts, platform settings, and preferred defaults are configured around your personal loop.
The core workflow is strongest when code, shell, search, and repository state stay visible inside the same environment.
Use Files to navigate structure, Search to jump across a codebase, and Editor to make the final changes without leaving the product.
Run package scripts, build steps, tests, and one-off commands directly beside the code they affect so context is never lost.
Architecture notes, working plans, and task state can stay close to the repo instead of disappearing into separate tabs and apps.
Git is one of the product's clearest differentiators, so lean into the built-in repository surfaces rather than bouncing to another tool.
Use the Git mode to inspect changed files, diffs, history, and branch state before you commit or hand off work.
Use separate worktrees when you need multiple active branches, experiments, or agent-led tasks open at once.
The visual branching and history views are useful for staying oriented when several features or fixes are moving in parallel.
The strongest way to use the flagship is to supervise multiple agent sessions alongside normal engineering work.
Open agent sessions with clear repository context and a bounded task instead of relying on one long-running generic conversation.
Use Command Center to watch multiple sessions, compare outputs, and keep orchestration tied to the rest of the build workflow.
Prompt-based creation works best when you can immediately drop into Git, Editor, Terminal, API, or Database modes to finish the real task.
CodeSplash AI is more valuable when research, operations, and planning stay in the same environment as the code itself.
Keep docs, references, and research close to the repo instead of juggling a separate browser workspace.
Inspect requests, environments, queries, and data while you build instead of context-switching to external clients.
Calendar, Email, Notes, and Kanban extend the product from coding into actual product delivery work.