jira-fields-and-flags

Jira fields and flags

Push deadlines, priority, labels and impediment flags into the Jira ticket itself, where the team can actually filter and sort on them. No comments.

Inputs

{
  "ticket_key": "LEGAL-4321",
  "priority": "Highest" | "High" | "Medium" | "Low" | "Lowest",
  "sla_days": 1-90,
  "matter_type": "nda" | "contract_review" | ...,
  "risk_flags": ["regulator", "claim", "tight_deadline", "inspection"],
  "human_review_required": true | false,
  "today_iso": "2026-06-02"
}

Step 1: Compute the due date

due_date = today_iso + sla_days calendar days formatted YYYY-MM-DD.

If the resulting date falls on Saturday or Sunday, roll back to the previous Friday -- internal deadlines should land on workdays so they trigger morning sweeps.

Step 2: Compute labels

Build a stable, alphabetically sorted label set:

• matter:{matter_type} (e.g., matter:nda)
• auto-triaged-{today_iso} (e.g., auto-triaged-2026-06-02)
• For each risk flag: risk:{flag} (e.g., risk:regulator, risk:claim)
• If human_review_required: add needs-lawyer-review

Never replace existing labels. Always merge: read the current label set via mcp__atlassian__getJiraIssue (fields: ["labels"]) and combine.

Step 3: Compute the flag field

myPOS Jira uses the system Flagged field (customfield_10021, value Impediment) to mark a ticket as blocked / needs attention. Set it ON if any of:

• risk_flags includes regulator, inspection, or claim
• human_review_required == true
• sla_days <= 2

Otherwise clear it.

Field discovery. If the customfield_10021 ID is wrong on a particular Jira instance, look up the right ID once via mcp__atlassian__getJiraIssueTypeMetaWithFields and cache it in knowledge/config.md under "Jira field IDs". The skill always reads from there first; the customfield ID above is the documented fallback.

Step 4: Push the edit

A single editJiraIssue call with all fields. Reasons:

• Atomic: either all fields land or none do; partial state is the worst outcome.
• Cheap: one HTTP round trip.

mcp__atlassian__editJiraIssue(
  cloudId = "fb47470f-f5c2-44bc-8182-f2a22f059adb",
  issueIdOrKey = ticket_key,
  fields = {
    "priority":  {"name": priority},
    "duedate":   due_date,
    "labels":    merged_labels,
    "customfield_10021": flagged_value   // [{"value": "Impediment"}] OR []
  }
)

Where flagged_value is [{"value": "Impediment"}] when on, and [] when off (Jira represents "no flag" as an empty array, not null).

If the edit returns an error mentioning an unknown field, peel off that field and retry. Order of importance: priority > duedate > flagged > labels. Never silently drop priority or duedate. If those two cannot be set, return:

{"applied": false, "reason": "priority/duedate edit rejected: {error}", "fields_attempted": {...}}

Step 5: Return

{
  "applied": true,
  "priority": "High",
  "due_date": "2026-06-09",
  "labels_added": ["matter:nda", "auto-triaged-2026-06-02", "risk:claim"],
  "flagged": true,
  "fields_skipped": []
}

The calling command surfaces these in the AI Triage comment under "Fields set" -- not as a duplicate of what's already on the ticket, but as a one-line confirmation:

Fields set: priority=High, due=2026-06-09, flagged=yes, labels=[matter:nda, risk:claim, ...]

---

Hard rules

• NEVER set priority via a comment. The priority field exists -- use it.
• NEVER set deadlines via a comment. The duedate field exists -- use it.
• NEVER blow away existing labels. Always merge.
• NEVER set a duedate in the past. If sla_days computes to a past date (clock skew, retroactive triage of an old ticket), set duedate to today_iso instead and add the label sla-already-elapsed.
• NEVER turn the Flagged field on without a real reason. Spurious flags train the team to ignore them.