A living incident state, pinned
Hypothesis, actions taken, owners, and open questions in one message that stays current as responders investigate. How the channel summary works.
State updated: hypothesis changed, 1 new owner, 1 open question.
T+00:00 paged · #inc-payments-latency
Less catching up. More useful work.
Incident Context Copilot is a free Slack bot that keeps responders on the same page. One pinned picture of what’s known, what’s been tried, and who’s on it, kept current as the channel moves.
The problem
Incidents get loud. Context gets buried.
Forty minutes into a SEV 2, the channel is the only record of what’s been tried, and it’s scrolling faster than anyone can read.
- 214messages in #inc-payments-latency
- 6responders, joining at different moments
- 3people asking “any update?”
- 1pinned picture, kept current
An example incident, followed through this page.
The pin
Folded into one message. Pinned where everyone looks.
Incident Context reads the channel and keeps a single summary current: the hypothesis, what’s been tried, who’s on it, and what’s still open. When the picture changes, the pin changes with it.
#inc-payments-latency
SEV 2
Current hypothesis
RDS connection saturation Checkout p95 rose after the 14:02 deploy; payment-api pods are restarting.
Tried
- Canary rolled back
- RDS connections checked
On it
- Maya · rollback
- Julian · API logs
Open question
What’s triggering the pod restarts?
How it works
Keep the channel moving. Keep everyone with it.
Three things, done quietly inside the incident channel you already use: a pinned state, duplicate flags, and handoff briefings.
Duplicate work, gently flagged
When the same check is happening twice, the bot says so in the thread. Once. Then it gets out of the way.
Maya and Priya are both checking the payment-api connection pool. Worth splitting?
A clean handoff, not “scroll up”
Someone joins late or the shift changes? They get a short briefing. Type /handoff whenever you need one. What goes in a good incident handoff.
You’re up to speed, Riley. Latency followed a deploy. Canary is rolled back; pod restarts are still open.
Scope
A copilot, not an autopilot.
It helps responders see the same picture. It doesn’t diagnose root cause, touch your systems, or replace the judgment your team already relies on.
Does
- Summarizes what the channel already knows
- Makes owners and open questions visible
- Briefs late joiners in under a minute
- Wipes channel history on
/inccloseor after 6 quiet hours
Doesn’t
- Claim a root cause
- Run commands or auto-remediate
- Add a dashboard you have to keep open
- Replace incident.io, Rootly, or PagerDuty
Give the next incident a little more context.
Install it, invite it to an incident channel, run /incstart. Summaries use your workspace’s Anthropic API key, set once with /setkey. The bot is free; Anthropic bills API usage to you directly. Read the FAQ.