Monitoring with agent view
When you delegate work to one or more agents, you need a way to see what they are doing without interrupting them. The agent view is a live panel in Syntic Code that lists every agent currently running, shows its status, and lets you drill into any single agent’s activity. It turns background work from a black box into something you can observe and steer.
Because agents run in their own context windows, their tool calls and reasoning do not clutter your main transcript. The agent view is where that hidden activity becomes visible — you can watch an agent read files, run commands, and converge on a result, all in real time.
Opening and reading the view
Bring up the agent view from the Syntic Code interface while agents are active. Each entry shows the agent’s name, the task it was given, its current state — running, waiting, blocked, or done — and a rolling summary of its latest actions. Selecting an entry expands it so you can follow the full stream of tool calls and intermediate output for that agent.
Use the view to answer the questions that matter mid-run: Is the agent making progress or stuck in a loop? Is it about to touch a file it shouldn’t? Has it already produced the deliverable you need?
Acting on what you see
The agent view is not just a monitor — it is a control surface. From it you can:
- Pause an agent to inspect its work before it continues.
- Stop an agent that has drifted off course or finished early.
- Focus a single agent to read its complete history.
- Return results to your main session once an agent completes.
A good habit is to glance at the view whenever you launch parallel work. Catching a misdirected agent in its first few steps saves far more time than reviewing a wrong result at the end. For teams of agents working together, the view is also the fastest way to spot which member is the bottleneck.