Engineering
Check production when someone asks if it's down
Ask Merlin in Slack when someone wonders if production is down. It checks error rates in Datadog, Vercel or Cloudflare and what merged last, then answers in the thread.
@Merlin from now on, when someone in #eng-help asks if production is down or broken, check the last hour of error rates in Datadog, Vercel or Cloudflare (wherever ours are) and the pull requests merged most recently to acme/web. Reply in the thread with what you found and whether it looks like an outage. Only read: don't restart, roll back or page anyone, and say so if you can't reach them.
- Starts whenSomeone asks
@Merlinin#eng-helpwhether production is down Merlin worksClaude Code or Codex checks error rates in Datadog, Vercel or Cloudflare (wherever yours are) and the pull requests merged last to acme/web
- You getA reply in the thread with the current error rates, the latest merged change and whether anything looks wrong
Tools
- Slack
- Datadog
- Vercel
- Cloudflare
- GitHub
Capabilities
- Integrations
- Memory
- Your agent's MCP
How it works
From an is-it-down question to the numbers in the thread.
Anyone who hears that something is broken mentions
@Merlinin#eng-helpand asks.Claude Code or Codex reads the last hour of error rates in Datadog, Vercel or Cloudflare, wherever yours are, and the pull requests merged most recently to
acme/web. It only reads.The reply gives the current numbers, the latest merged change and whether it looks like an outage. Nobody gets paged and nothing is rolled back.
Anyone who hears that something is broken mentions @Merlin in #eng-help and asks.
Claude Code or Codex reads the last hour of error rates in Datadog, Vercel or Cloudflare, wherever yours are, and the pull requests merged most recently to acme/web. It only reads.
The reply gives the current numbers, the latest merged change and whether it looks like an outage. Nobody gets paged and nothing is rolled back.
Acme
Channels
general
eng
product
support-eng
data
eng-help
Direct messages
Merlin
Grace
Ben
eng-help
24
Today
Grace16:20
@Merlin a few customers say the dashboard won't load. Is production down?
Message #eng-help
Before and after
Is-it-down questions get answered with numbers, not guesses.
Without Merlin
Support pings around until an engineer stops what they're doing to open the dashboards.
- 01Ask who's on call
- 02Wait for an engineer to reply
- 03Open the service dashboard
- 04Compare error rates by region
- 05Check p95 latency
- 06Look up the latest merges
- 07Line up the times
- 08Reply in the thread
With Merlin
One mention gets the current numbers and the latest change in the thread.
@Merlinin#eng-helpAsked by anyone who hears of a problemReads Datadog and GitHub on Ben's Mac miniClaude Code, read-only
- What it found, in the threadError rates, latency and the latest merge
Setup
Add Merlin to #eng-help and give your agent Datadog, Vercel or Cloudflare.
Set these up once and anyone in the channel can ask whether production is down. Memory is optional.
- 01Connect a machineRuntimeInstall the Merlin desktop app or CLI on a computer running Claude Code or Codex. You can also use a machine a teammate shares with your workspace.
- 02Add Merlin to SlackIntegrationsInstall the Slack app and add Merlin to the channel you work in. A private channel needs
/invite @Merlin. - 03Add Datadog or Vercel or Cloudflare to your agentYour agent's MCPConnect the Datadog or Vercel or Cloudflare MCP server to Claude Code or Codex on the machine that runs this. Its credentials stay on that machine.
- 04Install the GitHub appIntegrationsPick the repositories Merlin can access, so it can read and link the pull requests there. Pushes use the machine owner's own Git access.
- 05Share a machine that has your monitoringRuntime
The people who ask are often in support and don't run an agent. An engineer can share a machine with Datadog, Vercel or Cloudflare connected. Merlin picks the asker's own machine first and a shared one after that.
- 06Use read-only accessYour agent's MCP
Sign the Datadog MCP in with a read-only role, or grant only read permissions when you connect Cloudflare, so a check can read but can't change monitors, mute alerts or edit settings.
- 07Turn on memoryOptionalMemoryMemory is off by default. When an owner or admin turns it on, every run gets the facts your team saves in public Slack channels. A private channel also keeps its own store for its members.
- 08Paste the setup promptGuidanceAn owner or admin sends the prompt below to @Merlin once, and it becomes workspace guidance for every future run.
Prompt
Tell Merlin how to answer is-it-down questions.
@Merlin from now on, when someone in #eng-help asks if production is down or broken, check the last hour of error rates in Datadog, Vercel or Cloudflare (wherever ours are) and the pull requests merged most recently to acme/web. Reply in the thread with what you found and whether it looks like an outage. Only read: don't restart, roll back or page anyone, and say so if you can't reach them.
Once, when you paste it
- PromptYou paste the prompt to @Merlin
- GuidanceMerlin saves your rules as workspace guidance
Every time someone mentions Merlin in Slack
- IntegrationThe Slack app spots the mention
- RuntimeMerlin picks an available machine
- ContextThe run gathers context
- WorkClaude Code or Codex reads your error rates and GitHub
- SteeringYour team steers the work in the thread
- ResultMerlin replies with what it found
Related use cases
More engineering tasks you can hand to Merlin.
- AutomationsIntegrationsYour agent's MCP
Engineering
Investigate an alert when it fires
When incident.io raises an alert, Merlin checks the service in Datadog and Sentry and what merged recently. You get the likely cause in #oncall.
incident.io reports a new alertThe likely cause in#oncall - IntegrationsMemoryYour agent's MCP
Engineering
Find what changed since the last deploy
Ask Merlin in Slack what changed since the last deploy. You get the pull requests that went out, what each touched and which look risky.
@Merlinin#engWhat went out, risky changes flagged - IntegrationsMemoryYour agent's MCP
Engineering
Find why an endpoint got slow
Ask Merlin in Slack why an endpoint got slower. You get when it started, the change that likely caused it and what to fix.
@Merlinin#engThe likely change and what to fix, in the thread
Bring your coding agents into the tools your team already uses.
Let your whole team hand work to Claude Code and Codex today. Connect one machine and add Merlin to Slack, Linear or GitHub. 14-day free trial, no API keys, no card.