When to Use
- Investigate a customer bug or "it doesn't work" report
- Reproduce issues across environments (browser, API, mobile)
- Analyze logs, request IDs, and HTTP/API error payloads
- Draft technical response with steps, workaround, or ETA
- Escalate to engineering with minimal repro and impact
- Author or update KB articles after resolved issues
When NOT to Use
- Support queue SLAs, CS metrics, renewals, dunning →
customer-ops-specialist - Company-wide outage incident commander role →
incident-management-engineer - Implement product code fixes in repo →
fullstack-software-engineer - Public API reference documentation →
tech-writer-researcher - Security incident or SOC alert investigation →
defensive-security-analyst
Related skills
| Need | Skill |
|---|---|
| Support ops, health scores, billing tickets | customer-ops-specialist |
| App code fixes and PRs | fullstack-software-engineer |
| Browser auth, CORS, session issues | web-application-developer |
| Production outage process | incident-management-engineer |
| Customer-facing docs and API guides | tech-writer-researcher |
| Infra/logs on platform side | devops |
| Company-wide incident or launch messaging | communication-lead |
| Exec/VIP and public community escalation program | community-executive-escalations-program-manager |
| How-to replies, KB gaps, feature-request capture | product-support-specialist |
Core Workflows
1. Intake and triage
Capture before deep dive:
- Customer ID, environment (prod/sandbox), region
- Timestamp with timezone
- User role and exact URL or API endpoint
- Error message verbatim; request/correlation ID
- Recent changes (release, config, integration)
- Severity: blocked vs degraded vs question
Route billing/access-only issues to customer-ops-specialist if non-technical.
See references/ticket_triage.md.
2. Reproduce
- Follow customer steps on comparable environment
- Reduce to minimal repro (fewest steps, clean account if possible)
- Note deterministic vs intermittent
- Capture HAR, server logs, or curl with redacted secrets
See references/reproduction_debugging.md.
3. Isolate root cause
| Layer | Checks |
|---|---|
| Client | Browser, cache, extensions, version |
| API | Status code, body, rate limits, auth token |
| Config | Feature flags, tenant settings, quotas |
| Data | Specific record IDs, permissions |
| Platform | Status page, known incidents |
Document hypothesis and evidence; avoid guessing in customer reply.
4. Resolve or escalate
If known issue: link KB, workaround, linked engineering ticket + ETA.
If new bug: file engineering ticket with:
- Title:
[component] short failure description - Impact: customer count, blocked workflows
- Repro steps, expected vs actual
- Logs/IDs attached
See references/escalation_engineering.md.
5. Customer communication
- Acknowledge impact; no blame
- State what is known, unknown, and next update time
- Provide workaround before permanent fix when possible
- After fix: confirm with customer; close loop
See references/customer_comms.md.
6. Knowledge base
After fix verified:
- Symptom → cause → resolution → prevention
- Searchable title; tags for product area
- Link to public API doc if applicable
See references/kb_article.md.
When to load references
- Priority and intake →
references/ticket_triage.md - Repro and logs →
references/reproduction_debugging.md - Engineering handoff →
references/escalation_engineering.md - Customer updates →
references/customer_comms.md - KB template →
references/kb_article.md