Omniscio

Runaway agent commands (told to the agent, never your inbox)

How Omniscio handles an agent command that has run too long: it tells the agent that ran it, and that agent's crew lead, and never puts a card in your inbox. Read-only searches are stopped on their own; everything else is left to the agent. How to turn the watch off.

On this page

What it is

Agents run shell commands all the time. Most finish in seconds. Occasionally one runs away — a search through a project's whole history, or a listing of a folder with a million files — and keeps running for half an hour. While it runs it can slow every agent on the machine, and nothing about its own session looks wrong.

Omniscio watches for that. About every two minutes it looks at every command your agents are running right now. When one has run ten minutes or more and either

  • it is a search or listing (such as grep, rg, find, ls or a git history read) that is still busy while the other agents are measurably slowing down — an honest search finishes in seconds, or
  • it has the shape of a search known to run away (for example a history search with no file named, or a recursive listing with no depth limit),

the agent that ran it gets one message: "A command you started at 3:12 PM has run 23 minutes and is still busy while the other agents are slow." If that agent is part of a crew, the session it reports to (its foreman or crew lead) gets a short note too, so someone with the context can step in.

You never see these in your inbox. Deciding what to do about an agent's command is the agent's job, not yours.

The message says only what Omniscio measured: how busy the command was over the last two minutes (a share of one processor core, and file operations a second) and that the other agents are slow right now. It never claims this command is the one slowing them — nothing measures that. It names the command by when it started, never by the command itself, because a command can carry a password the agent was never meant to see.

If the agent cannot be reached (it has ended, been archived or paused), Omniscio does not wake it and does not fall back to your inbox: it writes the miss to its log and record instead.

A build, a test run, an install, a merge gate or a script is never flagged just for being busy. That work is expected to run long and hard, so being busy says nothing about it. It is still flagged if its command contains a search with a known runaway shape.

Where to find it

There is nothing to open. The watch runs on its own in the background while your agents work, and it never adds a card to your inbox, so there is no screen to visit and nothing to dismiss. Cards of this kind raised by older versions are cleared from your inbox the first time the watch runs after an update.

Neither of its two controls has a switch on a Settings screen. To change one, ask an agent to do it for you: because the watch polices agents' own commands, an agent's request to change it reaches you as a card to approve rather than taking effect on its own.

The record it keeps of every command it judged is a file on your machine, described under For agents below.

How it behaves

Commands left running after their tool call ended

When an agent's command takes too long, its tool call gives up and its shell ends — but the program the shell started can keep running with nobody waiting for it. A recursive search left this way once kept a processor busy for almost two hours.

Omniscio watches those too, judged by exactly the same rules as a running command, and the agent is told the command was left running after its tool call ended. When Omniscio saw the original command, it judges the leftover on what the agent actually typed. When it did not, the leftover is never stopped automatically, because it cannot tell whether the program was writing its output to a file.

When an agent starts a second copy

An agent that puts a long command in the background sometimes loses track of it. Git Bash's ps shows no program arguments, so a check like ps | grep probe.mjs finds nothing while the command is still running, and the agent starts it again — one agent ran three copies of an hours-long check this way.

So when an agent's command starts a program that is byte-for-byte the same as one the same session has already been running for ten minutes or more — still under its tool call or left running after it — Omniscio tells that agent once. The message names the program, when each copy started and its process number, says the first copy had not died, and says how to stop the extra copy safely. It never shows the command itself. It comes after two checks in a row see both copies, and reaches a busy agent when its current turn ends. Nothing is stopped.

Only the program a command started counts — never the small helpers every command shares (a console window host, tail, grep, sleep) — and the same command in another session, or two programs one command deliberately started together, never count.

What Omniscio stops on its own

Only read-only searches and listings (such as git log, grep, rg, find, ls, du), only after twenty minutes, only while they are busy and the other agents are slowing, and at most two per check. It looks at the whole command the agent typed, so a search whose output is written to a file is not read-only and is never stopped automatically. If an automatic stop cannot finish, the command keeps running and the agent is told instead.

A search run through the rtk output filter (rtk rg …, rtk grep …) is judged exactly like the same search typed without it, so it is told and stopped by the same rules. rtk running anything that is not a read-only search (rtk npm …) is never stopped automatically.

A stop ends only that command — including every program it started (a pipeline such as git log … | grep … is treated as one command) — and never the agent's session, its shell, or its tool servers. The agent sees its command fail, and receives a short message explaining what was stopped (named by when it started), after how long, why, and how to run it again with limits. It never closes an app you are using.

Anything that writes, installs, builds or tests — a commit, a rebase, an install, a test run — is never stopped automatically.

Turning it off

It is on by default, and two settings control it: one turns the whole watch off — no messages and no stops — and the other keeps the messages but turns off only the automatic stops. Ask an agent to change either one; the request reaches you as a card to approve.

  • Heavy job pacing — how heavy background work is spaced out.
  • Session cpu cap — the ceiling on how much a session may take.

Last verified 2026-11-17