Twg Operational Health

twg-operational-health

Use together with the root twg skill. Exact command grammar must come from live twg help, twg help <terms>, or twg help describe <path>.

Use When

• "I'm taking over on-call"
• "Reliability/HOT/incidents/PIR readout"
• "Laptop refresh candidates"
• "Windows laptops for testing"
• "Team capacity or staffing candidates"
• "Meeting recordings weekly summary"
• "Open risks, blockers, overloaded people, operational health"

First Move

Resolve the operational scope and time window:

• Service, component, team, customer, project, person, org, or asset domain.
• Canonical anchors such as service/component records, HOT/incident/PIR issues,

runbooks, projects/goals, meeting recordings, or Assets object types.

• Current owner, escalation path, status, recency, and known follow-up work.

Do not start with a broad cross-product inventory. Find the operational anchor first, then join only surfaces that answer the question.

Route Selection

• On-call handoff: incidents/HOTs/PIRs, recent Jira follow-up, runbooks/docs,

linked projects, owners, and escalation paths.

• Reliability review: HOT/incident/PIR records, linked follow-up, service or

component context, owners, comments, and current status.

• Assets/laptop refresh: build the contributor list from work evidence first,

then inspect Assets schema/type metadata before AQL.

• Capacity/staffing: combine related recent work, ownership, review influence,

docs, project/goal involvement, assigned open work, review queue, blockers, and current risks.

• Meeting summaries: query recordings for the requested people/time scope, fetch

transcript preview first, and fetch full transcript only for central meetings.

Evidence Policy

• Rank by impact, urgency, owner clarity, recurrence risk, and actionability.
• Group incidents, HOTs, PIRs, blockers, and follow-up work into patterns rather

than listing every artifact independently.

• Separate live risks from historical mentions.
• For Assets, discover schema and object type first. Join people to devices using

discovered user-like attributes such as Calculated user before broad display name filters.

• For staffing, do not recommend people solely from recent activity counts; blend

contribution centrality, ownership, expertise, availability/load signals, and project risk.

Recipe Cards

On-Call Handoff

Resolve team/service/component. Pull incidents/HOTs/PIRs, recent Jira follow-up, runbooks/docs, linked projects, and owners. Cluster symptoms into patterns and produce a first-hour checklist plus escalation map.

HOT / Reliability Review

Search/query HOTs, incidents, PIRs, linked follow-up work, service/component context, owners, comments, and status. Cluster by failure theme. Connect each theme to concrete follow-up and missing ownership.

Assets / Laptop Refresh

Build the contributor list from project/goal/Jira/PR/doc/activity evidence. Query Assets schema/type metadata before AQL. Join people to laptops using discovered user-like attributes, then rank by contribution centrality plus asset risk.

Windows Asset Candidates

Use Assets schema-first discovery. Confirm site, schema, object type, and owner or contact fields. Report confidence and gaps for each candidate.

Capacity / Staffing Candidates

Resolve project/topic/org. Identify people with related recent work, ownership, review influence, docs, and project/goal involvement. Check current load signals before recommending candidates.

Meeting Recording Summary

Query meetings/videos for the requested time window and people scope. Fetch transcript preview first. Fetch full transcript only for central recordings and write it to a file when needed. Summarize decisions, action items, themes, and gaps.

Output Shape

• Lead with severity, urgency, or recommendation summary.
• Inventory table with owner, status, recency, impact, confidence, and evidence.
• Patterns/themes grouped across artifacts.
• Ranked next actions and suggested owner for each.
• Data gaps: missing transcripts, no asset match, stale update, ACL/auth gaps, or

weak ownership evidence.

Anti-Patterns

• Do not fetch every transcript or page body.
• Do not treat every incident mention as a live risk.
• Do not join Assets by display-name guesses before inspecting schema/type fields.
• Do not recommend staffing solely from recent activity counts.