> ## Documentation Index
> Fetch the complete documentation index at: https://docs.claboard.dev/llms.txt
> Use this file to discover all available pages before exploring further.
# Project Terminal
> Watch every running task in a project from a single live terminal tab with per-task color-coded badges
The **Project Terminal** tab aggregates log output from every task in the current project into one live view. Instead of opening a separate terminal per task, you get a single feed where each line is prefixed with a color-coded `[TASK-KEY]` badge so you can tell which CLI wrote what at a glance.
## Opening the tab
Open any project, then click the **Terminal** icon in the top view tabs (next to Roadmap). The tab streams logs in real time via the same `task:log` event used by the per-task Live Terminal, so there's nothing to start — the moment an agent emits a line, it appears here.
## Per-task color-coded prefix
Every line carries a `[TASK-KEY]` badge styled with the color mapped to the task's `task_type`:
| Type | Color |
| ---------- | ------ |
| `feature` | Blue |
| `bugfix` | Red |
| `refactor` | Purple |
| `docs` | Green |
| `test` | Yellow |
| `chore` | Gray |
Hover the badge to see the task title. The timestamp is shown in a fixed-width column so columns align cleanly even when messages span multiple lines.
## Unified vs Split view
The header has a toggle between two modes:
A single auto-scrolling log where all active tasks share one feed ordered by event arrival time. This is the default — it's the fastest way to follow a swarm of parallel agents without losing context.
Each active task gets its own pane in a responsive grid. The grid adapts to the number of active tasks:
* 2 tasks → 1 × 2
* 3–4 tasks → 2 × 2
* 5–9 tasks → 3 × 3
* 10+ tasks → 4 columns
Each pane auto-scrolls independently, so pausing your cursor in one pane doesn't pin the others.
## Active filter
The **Active only** toggle (default ON) limits the view to tasks that are actively running or under review. A task counts as active when any of the following is true:
* `is_running` is `true`
* `status` is `in_progress`
* `status` is `review` (the task is in testing)
* `status` is `verifying`
Toggle to **All tasks** when you want to see logs from completed or paused tasks as well — useful for scrolling back through history or comparing outputs across states.
## Stream controls
Freezes the view. New log events are queued (a counter shows how many). Press Resume to flush them in order.
Dumps all queued events into the view and re-enables live streaming. If the buffer exceeds the cap, older entries are trimmed.
Empties the current view (and any paused queue). Back-end logs are not touched — the per-task Live Terminal still has the full history.
## Auto-scroll behavior
The view sticks to the bottom while you're at the bottom. Scroll up and it pauses auto-scroll so you can read earlier output; an **Arrow-down** button appears in the bottom-right — click it to jump back to the live tail.
## When to use this
* **Watching parallel orchestrations** — multiple agents running via the DAG / queue at once; Split mode gives every agent its own viewport.
* **Debugging a chain of tasks** — Unified mode with Active only off lets you see the full timeline of cause-and-effect across a plan's execution.
* **Triage without context-switching** — stay on the Terminal tab while agents work; jump to the board only when an agent finishes or needs review.
Subscription to the event bus is registered once when the tab mounts and never re-subscribes on task-list updates. That means log events are never dropped between a React re-render and the new listener attaching.
## Limits
* The view keeps the most recent 3000 lines in memory (trimmed to \~70% when the cap is hit). Older lines scroll out of view but remain in the per-task database if you need them later.
* Split mode fits up to 4 columns; beyond \~16 simultaneous active tasks the panes become quite narrow — Unified mode is a better fit at that scale.