Customer Incident Update
During an outage, silence is the second failure. But the customer update is its own craft: say what's affected without guessing at causes, commit to a next-update time you can keep, and never promise a fix-by you don't control. This writes the post for the stage you're in — the messy middle included — so customers feel informed, not managed.
What This Skill Produces
- The update, in the tense of the current stage (investigating / identified / monitoring / resolved)
- The impact line — who and what is affected, in the customer's terms
- The workaround — if one exists, stated plainly
- The next-update commitment — a specific time, always
- Channel variants — a terse status-page version and a fuller email if needed
Required Inputs
Ask for these if not provided:
- Stage — investigating, identified, monitoring, or resolved
- Impact — which product/region/customers, and what they can't do right now
- What's known — only what you're confident of; unknowns stay unknown in the post
- Workaround — any, or none
- Audience — all customers, affected only, or enterprise accounts (tone shifts)
Framework: Honest Status Comms
- Match the tense to the stage. "We're investigating" ≠ "we've identified" ≠ "we're monitoring the fix." Don't skip ahead.
- Impact before cause. Customers care what's broken for them; causes come in the postmortem, not mid-incident.
- Commit to the next update, not the fix. "Next update by 15:00 UTC" is a promise you can keep; "fixed within the hour" often isn't.
- No speculation. If you don't know the cause, say you're investigating — a wrong guess published is worse than an honest unknown.
- Own it plainly. Brief, human, no corporate throat-clearing; apologise once, then inform.
Output Format
Status-page post
[Investigating/Identified/Monitoring/Resolved] — [title] · [timestamp]
[Impact: who/what]. [What we're doing]. [Workaround, if any]. Next update by [time].
Email (if broader comms needed)
- Subject: [clear, non-alarmist]
- Body: impact → status → workaround → next update → apology
- Sign-off
Update ladder (for the incident's life)
- The follow-on posts you'll publish as the stage changes, pre-drafted
Quality Checks
Anti-Patterns
- "Some users may be experiencing issues" when it's a full outage — minimising erodes trust faster than the outage.
- Promising a resolution time you can't control.
- Publishing a guessed cause that turns out wrong.
- No next-update time — leaves customers refreshing in the dark.
- Skipping stages — jumping to "resolved" before monitoring confirms it.
Example Trigger Phrases
- "Write a status page update — we're investigating an outage."
- "Draft customer comms for the API downtime, identified stage."
- "Post an incident notice for enterprise accounts."
- "We've deployed a fix and are monitoring — write the update."
1---2name: customer-incident-update3description: "Write the customer-facing incident update during an outage — status-page post or email — that's honest about impact without over-promising. Use when asked to write a status page update, draft customer comms for an outage, post an incident notice, or tell customers about downtime. Produces the update in the right tense for the incident stage (investigating / identified / monitoring / resolved), with impact scope, any workaround, and a concrete next-update time. Distinct from incident-postmortem (the internal retro)."4disable-model-invocation: true # skill-optimizer5metadata:6 source: https://skillmd.com/skills/slipcast-dev/customer-incident-update7---89# Customer Incident Update1011During an outage, silence is the second failure. But the customer update is its own craft: say what's affected without guessing at causes, commit to a next-update time you can keep, and never promise a fix-by you don't control. This writes the post for the stage you're in — the messy middle included — so customers feel informed, not managed.1213## What This Skill Produces1415- **The update**, in the tense of the current stage (investigating / identified / monitoring / resolved)16- **The impact line** — who and what is affected, in the customer's terms17- **The workaround** — if one exists, stated plainly18- **The next-update commitment** — a specific time, always19- **Channel variants** — a terse status-page version and a fuller email if needed2021## Required Inputs2223Ask for these if not provided:24- **Stage** — investigating, identified, monitoring, or resolved25- **Impact** — which product/region/customers, and what they can't do right now26- **What's known** — only what you're confident of; unknowns stay unknown in the post27- **Workaround** — any, or none28- **Audience** — all customers, affected only, or enterprise accounts (tone shifts)2930## Framework: Honest Status Comms31321. **Match the tense to the stage.** "We're investigating" ≠ "we've identified" ≠ "we're monitoring the fix." Don't skip ahead.332. **Impact before cause.** Customers care what's broken for them; causes come in the postmortem, not mid-incident.343. **Commit to the next update, not the fix.** "Next update by 15:00 UTC" is a promise you can keep; "fixed within the hour" often isn't.354. **No speculation.** If you don't know the cause, say you're investigating — a wrong guess published is worse than an honest unknown.365. **Own it plainly.** Brief, human, no corporate throat-clearing; apologise once, then inform.3738## Output Format3940### Status-page post41> **[Investigating/Identified/Monitoring/Resolved] — [title]** · [timestamp]42> [Impact: who/what]. [What we're doing]. [Workaround, if any]. Next update by [time].4344### Email (if broader comms needed)45- Subject: [clear, non-alarmist]46- Body: impact → status → workaround → next update → apology47- Sign-off4849### Update ladder (for the incident's life)50- The follow-on posts you'll publish as the stage changes, pre-drafted5152## Quality Checks53- [ ] The tense matches the actual stage — no claiming a fix that isn't confirmed54- [ ] Impact is stated in customer terms, up top55- [ ] A specific next-update time is given56- [ ] No cause is speculated when it isn't known57- [ ] No fix-by time is promised that depends on an unknown58- [ ] Tone is human and brief — one apology, no jargon5960## Anti-Patterns61- **"Some users may be experiencing issues"** when it's a full outage — minimising erodes trust faster than the outage.62- **Promising a resolution time** you can't control.63- **Publishing a guessed cause** that turns out wrong.64- **No next-update time** — leaves customers refreshing in the dark.65- **Skipping stages** — jumping to "resolved" before monitoring confirms it.6667## Example Trigger Phrases68- "Write a status page update — we're investigating an outage."69- "Draft customer comms for the API downtime, identified stage."70- "Post an incident notice for enterprise accounts."71- "We've deployed a fix and are monitoring — write the update."