A Slackbot use case · IT support

Same help desk.
Fewer tickets.

Slackbot can be the first line of support. Built on a customer’s existing help desk workflow, it checks the employee’s laptop, fixes what it can, and hands off to an agent only when it needs to.

Employee submits a request Slackbot checks the device Fixed in Slackor Handed off with context

The idea

Nothing new for employees to learn

A slow laptop or a full disk usually waits in the same queue as an outage. Agents start each one from scratch, even when the fix takes two minutes.

  • Same help channel
  • Same request button and form
  • After Submit, Slackbot works the request before anyone opens a ticket

Path 1 · Fixed in Slack

It finds the problem, asks, then fixes it

Example: an employee is running low on storage and doesn’t know where the space went.

Two examples.Storage and a slow laptop are just the two cases walked through here. There’s no fixed list of what it can do. See what else it can handle.

  1. The request comes in

    Submitted through the usual form, low urgency.

    “I am running into storage issues. I deleted some things from my Downloads folder, but not sure what to delete next.”

  2. Slackbot checks the Mac

    It reports free space, the biggest folders, and the exact files worth removing. It leaves anything it isn’t sure about alone and asks before changing anything.

  3. The user picks a fix

    They reply “1”. Slackbot moves the installers to the Trash, so the change can be undone, and says how much space they’ll get back.

Done. No ticket, no queue, no agent time.

Path 2 · Handed to a person

When an agent steps in, they start with context

Example: an employee’s Mac is running slow. They already restarted twice and marked it high urgency.

  1. Slackbot diagnoses it

    Load average, the busiest background apps, memory, disk and startup items. It recognizes Zscaler and Cisco as company security tools and says to leave them alone.

  2. The user asks for a person

    Slackbot confirms first, every time, so nothing is escalated by accident. Then the request goes to the help channel like it always has.

    This time the agent gets a summary of the diagnostics, what was tried, and that no fix was applied. High or “Blocking my work” urgency pings the help desk team right away.

What else it can do

Two examples. No fixed list.

The device tool combines commands and writes scripts to fit the problem in front of it, so there’s no set menu of tasks. Here’s a sample of nearly 100, grouped by area.

Desktop and focus

  • Hide desktop icons before screen sharing
  • Enable Do Not Disturb
  • Fix an unexpectedly hidden Dock
7 more
  • Restore hidden desktop icons
  • Switch to dark mode
  • Switch to light mode
  • Disable Do Not Disturb
  • Adjust Dock size and position
  • Configure Finder to show file extensions
  • Make screenshots save to a chosen folder

Files and storage

  • Identify folders consuming the most storage
  • Locate duplicate files by content
  • Organize a cluttered Downloads folder
7 more
  • Find unusually large files
  • Batch-rename files consistently
  • Find files by text inside them
  • Identify old installers you could remove
  • Find empty folders
  • Compare two folders for differences
  • Create a searchable inventory of project files

Performance and stability

  • Diagnose an application that repeatedly crashes
  • Investigate excessive battery consumption
  • Collect diagnostics for a support ticket
7 more
  • Investigate high CPU usage
  • Identify processes using excessive memory
  • Investigate storage pressure
  • Review crash reports for likely causes
  • Investigate slow application startup
  • Identify unnecessary login items
  • Find background processes causing slowdowns

Networking

  • Diagnose a website connection failure
  • Troubleshoot VPN-related connectivity
  • Investigate TLS certificate errors
7 more
  • Investigate DNS lookup failures
  • Check whether a service port is reachable
  • Identify which process owns a listening port
  • Diagnose local development connection errors
  • Investigate proxy configuration problems
  • Diagnose SSH connection failures
  • Compare connectivity across several endpoints

Applications and configuration

  • Diagnose an application that will not launch
  • Investigate application permission errors
  • Repair malformed configuration files
7 more
  • Identify conflicting application versions
  • Find an application’s configuration files
  • Back up settings before changing them
  • Compare settings between working and broken setups
  • Troubleshoot file association problems
  • Investigate a missing command-line application
  • Create a repeatable application setup script

Terminal and developer tools

  • Fix a broken shell PATH
  • Diagnose Homebrew installation problems
  • Troubleshoot Docker startup or container errors
7 more
  • Resolve shell startup errors
  • Clean up conflicting shell aliases
  • Resolve incompatible Node.js versions
  • Repair a Python virtual environment
  • Diagnose npm dependency conflicts
  • Fix package registry authentication configuration
  • Resolve “address already in use” errors

Git and code

  • Diagnose Git authentication failures
  • Recover commits using Git’s reflog
  • Fix a broken project build
7 more
  • Resolve merge conflicts
  • Repair incorrect Git remote URLs
  • Identify why files are unexpectedly ignored
  • Remove accidentally tracked build artifacts
  • Investigate failing automated tests
  • Diagnose local environment differences causing bugs
  • Add a setup check for required development tools

Documents and data

  • Merge several PDFs
  • Extract text from scanned PDFs
  • Clean inconsistent CSV data
7 more
  • Split a PDF into separate documents
  • Convert documents between supported formats
  • Repair spreadsheet formulas
  • Combine related spreadsheets
  • Find duplicate spreadsheet records
  • Generate charts from local data
  • Batch-resize images for sharing

Backup and recovery

  • Check whether selected folders have backups
  • Investigate Time Machine backup failures
  • Restore selected files from an accessible backup
7 more
  • Create a dated backup of a project
  • Verify copied files using checksums
  • Find differences between a backup and current files
  • Create an archive excluding generated files
  • Recover overwritten work from version history
  • Create a restore script for application settings
  • Test a backup by restoring into a temporary folder

Automation helpers

  • Build a reusable local health-check script
  • Automate sorting files into folders
  • Create a one-command development startup routine
5 more
  • Automate consistent screenshot naming
  • Create a shortcut for opening a workspace
  • Automate converting incoming data files
  • Create a log collector for recurring errors
  • Build a tool that checks required project settings

How I built it

Mostly existing pieces, connected in a new way

The white cards are what I added. Everything else was already in place. The workflow, the skill and the design decisions are on their own page.

Existing

Slack Workflow

The help request form, now routed to Slackbot instead of straight to the channel.

New

Slackbot skill

Tells Slackbot to always run diagnostics first and confirm before acting.

New

Custom app + MCP server

Gives Slackbot access to the existing Codex plugin that checks and fixes devices.

Existing

Help channel

Where handoffs land, now with a summary and an urgency-based ping.

See how it’s built

Why it matters

What it changes for the help desk

Fewer tickets

Common device issues get fixed in Slack before a ticket exists.

Faster fixes

Minutes in a conversation instead of a wait in the queue.

Better handoffs

Agents start with diagnostics and a summary, not a blank ticket.

Same process. Slackbot does the rest.

The pattern works anywhere support already runs through Slack: keep the request people know, let Slackbot work it first, and hand off with context when a person is needed.