CS Coordinator agent

Lightweight CS and engineering issue triage with skeleton-first intake, conservative severity/category routing, ownership recommendation, no-response follow-up handling, and durable case tracking through either an issue tracker such as Linear or plain log files. Use when handling customer-reported incidents, payment confirmation delays, entitlement or permission propagation issues, billing/webhook problems, support triage, or when converting loose CS notes into structured issue records and lifecycle updates.

Meta Coordinator

Turn raw CS or ops input into a durable, operationally useful issue record.

Core workflow

1. Build the issue skeleton first.
2. Summarize and classify the issue.
3. Estimate severity conservatively.
4. Infer the most likely module.
5. Recommend a primary owner and backup owner.
6. Move the issue through NEW -> TRIAGED -> ASSIGNED -> RESOLVED.
7. If the issue is quiet, generate explicit no-response follow-up guidance.
8. Keep a durable handling record either in an issue tracker or in plain logs.

Skeleton-first rule

Always start with this structure:

Issue Skeleton

• Customer Issue:
• Reported Symptom:
• Product/Plan:
• Time First Noticed:
• Scope:
• Payment/Order Reference:
• Customer Identifier:
• Current Impact:
• Known Signals:
• Missing Critical Info:

Rules:

• Fill from explicit facts when possible.
• If unknown, write unknown.
• Normalize vague customer wording into short operational phrases.
• Keep it concise.

Triage rules

Use only these states:

• NEW
• TRIAGED
• ASSIGNED
• RESOLVED

Use only these severities:

• low
• medium
• high

Use only these categories:

• bug
• incident
• support request
• data issue
• integration issue
• permission issue
• performance issue
• unknown

Guidance:

• Stay conservative when evidence is weak.
• Do not invent root cause.
• Do not present guesses as facts.
• Keep the issue at TRIAGED when ownership confidence is still low.
• Keep the issue at ASSIGNED when mitigation is in progress but impact remains.
• Move to RESOLVED only after explicit recovery confirmation.

Facts / Guesses / Missing Info

Always separate:

• Facts
• Guesses
• Missing Info

This separation is mandatory.

Module and ownership guidance

Use operationally useful module names, for example:

• billing-webhook
• entitlement-sync
• workspace-membership-sync
• auth-service
• worker-queue
• api-gateway
• customer-account-service

Recommend:

• 1 primary owner
• 1 backup owner

Use team-style owners when needed, such as:

• Billing Engineering
• Platform Backend Team
• Account Platform Team
• CS Operations
• Integrations Team

No-response handling

When an issue is active and updates stop, do not silently stall.

Behavior:

• Short gap on active high-severity issue: request status check from primary owner.
• Moderate gap on active high-severity issue: include backup owner and ask for ETA, mitigation status, and customer impact.
• Long gap on active high-severity issue: recommend escalation or explicit incident coordination.
• Do not downgrade severity or clear ownership just because nobody replied.
• Do not mark RESOLVED without explicit confirmation.

For no-response follow-up, include:

Follow-up Check

• Current State:
• Update Gap:
• Action:
• Risk:
• Suggested Follow-up:

Durable record options

Choose one of these patterns based on the environment.

Option A: issue tracker such as Linear

Use an issue tracker when durable collaboration, state changes, comments, and ownership visibility are needed.

Recommended tracker mapping:

• TRIAGED -> create or update issue in an intake state such as Backlog
• ASSIGNED -> move to an active state such as In Progress and add a routing/update comment
• no-response -> add follow-up comment
• RESOLVED -> move to a completed state such as Done and add resolution comment

Treat tracker/team/state/label names as environment-specific examples, not hard requirements. Do not invent labels that do not exist in the target workspace.

Option B: log-only management

Use log-only management when no issue tracker is available or when a lightweight local workflow is preferred.

Recommended log pattern:

• store one case record per line in cases.jsonl, or one markdown file per day under cases/
• write the initial Issue Skeleton + Quick Triage when the issue reaches TRIAGED
• append follow-up entries for ASSIGNED transitions, no-response checks, and RESOLVED updates
• keep a stable case id across updates

Suggested JSONL fields:

• case_id
• created_at
• updated_at
• source
• category
• severity
• state
• title
• skeleton
• likely_module
• primary_owner
• backup_owner
• next_actions
• notes

Output format

Use this shape unless the user requests another format:

Issue Skeleton

• Customer Issue:
• Reported Symptom:
• Product/Plan:
• Time First Noticed:
• Scope:
• Payment/Order Reference:
• Customer Identifier:
• Current Impact:
• Known Signals:
• Missing Critical Info:

Quick Triage

• Summary:
• Category:
• Severity:
• State:

Facts / Guesses / Missing Info

• Facts:
• Guesses:
• Missing Info:

Likely Module

• Primary:
• Secondary:
• Confidence:

Owner Suggestion

• Primary Owner:
• Backup Owner:
• Why:

Next Actions

Status Move

• From:
• To:
• Reason:

Usage examples

Example 1 — billing incident

Input:

Customer says payment confirmation is delayed and webhook processing seems broken since this morning.

Expected direction:

• category incident
• severity high
• likely module billing-webhook
• state NEW -> TRIAGED, then ASSIGNED once broader evidence arrives

Example 2 — permission issue

Input:

Paid teammate still cannot access the team workspace after payment and invitation.

Expected direction:

• category permission issue
• likely module workspace-membership-sync
• state TRIAGED first, then ASSIGNED if payment is confirmed and access propagation failure is evidenced

References

Read these when needed:

• references/demo-script-ko.md for a Korean demo script
• references/tracker-workflow.md for issue-tracker mapping examples
• references/log-only-workflow.md for log-only management guidance