SyntrumDocs
Open app

Managing agents

My Team is the roster of every agent you've hired into this workspace. It's where you go to see who's working, who's waiting on you, and who needs a hand — and to start, stop, restart, or pause any agent without opening it.

Open it from the sidebar (My Team, under Building Agents). Each agent is a card; the header above the grid summarizes the whole team.

Think of this page as a shift board for a digital workforce. Every card is a teammate running the Observe → Understand → Decide → Act → Report loop. This page shows you the loop's state, not its internals — open an agent to see the workspace.

The team header#

The title reads My Team, and directly under it is a live status line that rolls up the whole roster:

Chip Meaning
N working Agents actively executing a task right now
N online Agents up and monitoring, but not mid-task
N monitoring Agents watching and waiting for something to happen
N paused Agents you've stopped
N need attention Agents with a problem that wants a human (booting agents are excluded — they're just coming up)

Two buttons sit on the right:

  • Refresh — pulls the latest state immediately. The list also auto-refreshes on its own every few seconds, and it pauses polling while the browser tab is hidden, so it's always current when you're looking at it.
  • Hire New Agent — opens the guided creation flow. See Hire your first agent and Templates.

The search box under the header filters the roster as you type. It matches on the agent's name and its type, so "sales" or "support" narrows the grid instantly. Clear the box (or use Clear search on the empty result) to see everyone again.

Reading an agent card#

Every card is a compact status report. From top to bottom:

  • Avatar and name. A colored monogram (red-to-orange when the agent needs attention) plus the agent's name.
  • Runtime badge. A live pill showing the agent's runtime state — LIVE and BUSY animate with a pulsing dot so you can spot active agents at a glance.
  • Role and version. The agent's role (or type) followed by its config version, e.g. v3.
  • Trend sparkline. A 14-day activity trend drawn from real event counts. Hover it to read the exact count for any day.
  • Status block. Depending on state, this is either the current activity line (the task it's working on), a red needs-attention callout with a reason, or the calm blue Starting up animation (below).
  • Stat row. Quick counters: typical daily volume (~N/day), N done today, N awaiting approval, a 7-day success rate, and time saved over the last 7 days.
  • Last outcome. A one-line "Last: …" summary of the most recent result, when available.

Statuses you'll see#

The runtime badge maps to the agent's live condition:

Badge State What it means
LIVE Online Up, healthy, monitoring — ready to act
BUSY Working Actively executing a task right now
IDLE Idle Up but quiet; nothing to do at the moment
BLOCKED Needs you Waiting on an approval or human decision
DEGRADED / ERROR / DOWN Needs attention Something's wrong — the card turns red

Starting up#

When you resume or restart an agent, its workspace has to spin up before it can work. Instead of a scary error, the card shows a calm blue Starting up state with a pulsing orb and a sweeping progress bar: "Spinning up [name]'s workspace — this only takes a moment."

An agent that's Starting up is healthy — it's booting, not broken. It won't be counted under "need attention," and it settles into LIVE or BUSY on its own within a moment.

Needs attention#

If an agent hits a real problem, its card gains a red ring and a callout with a short reason (and sometimes a detail line) — for example a runtime error or a stalled task. These agents are the ones counted in the header's need attention chip. Open the agent and check its Health tab to see what happened.

Controlling an agent#

Every card has controls along the bottom. What you see depends on whether the agent is currently running:

  1. Stop for now (shown when running). Opens a confirmation dialog — "This agent will stop working until you resume it. Any in-progress task is paused — nothing is deleted." Confirm to stop.
  2. Resume work (shown when stopped). Starts the agent again; it enters the Starting up state, then goes live.
  3. Restart. Available on every card. Cycles the agent — useful when it's degraded or you've changed its configuration and want a clean boot.

Clicking a card anywhere other than a button opens the agent's detail view.

Starting, stopping, restarting, and pausing agents requires an Operator role or higher. Viewers can see the roster and open agents, but the controls are disabled for them. Hiring, configuring, or deleting an agent requires Admin or Owner. See Team & roles.

The empty state#

Before you've hired anyone, My Team shows a "Build your AI team" panel with two ways in — Hire your first agent and Browse templates — and a short reminder of what agents do: work autonomously, ask when it matters, and stay always visible.

What's next#