Slack
Run your whole flags & experiments platform from Slack by chatting with @Shipeasy — create gates, configs, kill switches, experiments, alerts and metrics in plain language, then act on everything as it streams back into the channel as interactive cards.
Slack is a full front-end to Shipeasy. Connect it once, then talk to @Shipeasy in plain language to create and change anything — feature flags, dynamic configs, kill switches, A/B experiments, alerts, metrics — and watch it all stream back into the channel as interactive cards you act on without ever opening the dashboard. Bugs, requests and platform events land the same way.
Connect Slack
Add the connector
On the Feedback page open Connectors → New connector, pick Slack, name it, and Continue.
Add to Slack
You're redirected to authorize the Shipeasy Slack app in your workspace. After consent you land back in the configure view.
Pick channels
Choose the feedback channel new bugs/requests post to. Optionally pick a separate notifications channel for platform events — leave it on Same as feedback channel to share one. Tick Enabled and Save.
Meet @Shipeasy — run the whole platform from chat
Mention @Shipeasy in any channel it's in and ask for what you want. There's nothing to memorise — no flags, no syntax. It understands plain language, acts as you (your Slack identity is matched to your Shipeasy email, and you must be a project member), and tells you exactly what it did.
It can do everything the dashboard assistant can — read and change your entire platform:
| Ask @Shipeasy to… | …and it will |
|---|---|
| Feature flags | create a flag, roll it out to a %, choose who sees it, turn it on/off |
| Dynamic configs | create a config, change what it returns |
| Kill switches | create one, flip it on or off |
| A/B experiments | set one up, start or stop it, ship the winning version, read the results |
| Alerts | tell you when a number gets too high or too low |
| Metrics | start tracking something people do |
| Just ask | list what you have, check how a test is doing, "what changed today?" |
Create anything, just by asking
It's just as good at reading and changing what already exists:
@Shipeasy runs every action as you — your Slack identity → your Shipeasy email → a project membership check. A non-member gets a quiet "ask an admin to invite you," never silent access.
From chat to channel — what you set up comes back to you
The alert you just asked @Shipeasy to create doesn't disappear into a settings page. When it fires, it posts into your channel as a card — and you control its whole lifecycle right there.
Click Resolve and the same message is rewritten in place — buttons gone, stamped with who acted, the border flipped to green. Dismiss leaves a distinct grey, and Mute rule disables the rule that fired so it stops nagging.
When a specific error caught by see() starts
spiking, that lands too — with a little chart of the last hour so you can see
how bad it is at a glance, and the same resolve / dismiss / mute controls:
PaymentGatewayTimeout — the payment provider didn’t respond in 8s.Caught by
see() · 47× in the last hour (up from ~3/hr) · 12 users affected.The same loop runs for everything you set up. Start an experiment and, when it reaches significance, the result lands with the actions that matter:
…and if a guardrail moves the wrong way, you can stop it on the spot:
Flip a kill switch and the channel offers a one-click revert; publish a config to prod and the change is announced with a link:
Every action writes straight back into Shipeasy, records an audit entry, stamps who did it, and recolours the card. Here's the full set:
| Event | Card actions |
|---|---|
| Alert triggered | Resolve · Dismiss · Mute rule (disables the rule that fired) |
Error spiking (from see()) | Resolve · Dismiss · Mute this error (with a trend chart) |
| Experiment reached significance | Ship "<group>" · Stop experiment · Keep running |
| Guardrail breached | Stop experiment · Acknowledge |
| Kill switch flipped | Flip back (one-click revert) |
| Config published to prod | Acknowledge |
| Agent needs your attention | Open the issue ↗ · Acknowledge |
| Weekly digest · New team member | (informational — link only) |
Choose which of these post to Slack in Settings → Notifications (the Slack column is live once a connector exists).
Stopping or shipping an experiment, flipping a kill switch, and muting an alert rule go through the real admin path (so the change propagates) and verify you're a member of the project first. If you're not, only you see a quiet "ask an admin to invite you" — the shared card is untouched for everyone else.
Bugs & requests — the other inbound stream
New bug reports and feature requests post to your feedback channel with the title, reporter, page URL and full details. Each carries a Status ▾ dropdown covering the whole lifecycle.
Click “Log in” — nothing happens, no dialog.
| Status | Border |
|---|---|
| Open | 🔵 blue |
| In progress | 🟠 amber |
| Ready for QA | 🟠 amber |
| Resolved | 🟢 green |
| Won't fix / Dismiss | ⚪ grey |
Pick one and the card recolours its left border to match — Won't fix is a distinct grey, never the green of Resolved. Link a PR to the item in Shipeasy and the same message is edited in place to show it:
Click “Log in” — nothing happens, no dialog.
…and when it's done, the card stamps who closed it and turns green:
Click “Log in” — nothing happens, no dialog.
When an agent gets stuck — open the issue
Unattended agent runs (the ops loop burning down your queue) raise an "Agent needs your attention" card when they're blocked. Open the issue ↗ takes you straight to the item it's stuck on — not a generic dashboard landing — where the full escalation (the summary and the numbered steps the agent left) is shown in context.
1. Add a test payment key to the local setup
2. Re-run the checkout test
Prefer a quick slash command?
For a fast, no-chat shortcut, /shipeasy creates a metric alert from inside a
channel — the alert posts back to that channel, and the reply is private to you:
/shipeasy alert <metric> <gt|gte|lt|lte> <threshold> [name]For anything beyond a quick alert, just talk to @Shipeasy.
Drive Shipeasy from Slack's AI
If you'd rather use Slack's own AI assistant, Shipeasy is also available there as an MCP connection — the same flags, configs, kill switches, experiments and metrics. Your workspace admin enables it once under the Slack App's MCP Server Connection (Slack identity auth); after that, ask Slack's assistant to manage Shipeasy and it calls through to your project under the same membership check.
Channel routing
- Feedback channel — where new bugs and requests land.
- Notifications channel — where platform events post (defaults to the feedback channel).
- Per-alert channel — an alert created in a specific channel posts back to that channel, overriding the project default.
Status & health
The connector in the list shows a status — Needs OAuth, Needs channel, Active, Disabled, or Failing (last dispatch errored). The configure view surfaces the last error and last successful delivery time, so you can tell at a glance whether mirroring is healthy.
The @Shipeasy assistant, privileged card actions, the /shipeasy command, and the Slack-AI
connection resolve access by your Slack email, which needs the users:read.email scope. If you
connected Slack before this was added, click Add to Slack again to re-grant.