Features
A closer look at what Floex does
Six pillars that make running multiple coding agents feel like using one well-designed tool instead of six loosely related ones.
Multi-agent management
Every agent CLI you use — Claude, Codex, Cursor, OpenCode — runs in its own terminal inside the same window, with the same keyboard shortcuts and the same history. Switch between them as fast as switching browser tabs.
- One adapter interface behind every agent, so new ones slot in cleanly
- Per-session scrollback that survives reattaching or reconnecting
- Bring your own credentials — Floex never sees your API keys in plaintext
Projects & workspaces
A workspace is a project's home base: its repo, its running sessions, its knowledge, and its board, all in one place you can open and close as a unit.
- Per-project default shell and working directory
- Pick up exactly where you left off, days later
- Keep client work cleanly separated from everything else
Swarms
Some tasks are better split up than serialized. Swarms let you fan a job out across several agents at once, each in its own pane, so you review results instead of waiting on them one at a time.
- Launch a swarm from a single task description
- Watch every pane update live as agents work
- Merge or compare outputs before committing to one
Kanban task tracking
Agent output turns into real, trackable work the moment it matters — drag it onto a board with the same columns your team already uses.
- Create a task directly from a terminal session
- Link a task back to the session that produced it
- See what's queued, in progress, and shipped at a glance
Open Knowledge
A shared knowledge base scoped to a project, so decisions, conventions, and context don't get lost every time a session ends or a teammate picks up the work.
- Every agent in a project can read from the same knowledge base
- Version-controlled alongside your code, not a separate wiki
- Searchable from inside any terminal session
Extensions
Floex's adapter interface is intentionally simple, so adding support for a new CLI or a custom internal tool doesn't require waiting on a roadmap.
- A single trait to implement for a new agent adapter
- Custom actions you can trigger from the command palette
- Share extensions across your team or keep them private