portfolio-work-review

name	portfolio-work-review
description	Compute portfolio-mix, SLA-aging, and release-readiness reviews from a shared work-item environment. Use when a task asks for a closed-work portfolio mix vs target, an SLA breach/aging audit, or a release ship-readiness assessment, and points at a shared environment with work items, mix targets, SLA policy, releases, milestones, dependencies, and blockers.

Portfolio Work Review

This skill produces a single JSON answer for three kinds of engineering-portfolio review tasks against a shared work-item environment:

1. Portfolio mix — closed work in a scope, classified into portfolio categories, compared to a target mix, with a rebalance recommendation.
2. SLA aging / breach audit — reliability & security work tracked against SLA due dates, with overdue items, aging, duplicate clusters, and breach rate.
3. Release readiness — a release's milestone completion, gating work, blockers, dependency chains, and a ship decision.

The environment, the available endpoints, and any access token are supplied separately at solve time (e.g. an environment_access.md file). Read that file to learn how to reach the environment. Do not call any judge/evaluation endpoint while solving — produce the best answer directly from the environment data.

Shared method (all task types)

1. Pull the raw data from the environment: work items, mix targets, SLA policy, releases, milestones, dependencies, blockers. A restricted SQL endpoint (SELECT only) may also be available and is useful for cross-checks.
2. Use authoritative fields, not stale mirrors. Each work item has an authoritative status plus a stale mirror_status, and an authoritative work_type/labels/title plus a stale legacy_category. Always use the authoritative fields; ignore mirror_status and legacy_category.
3. Classify each work item into one of four portfolio categories (NewFeature, TechDebt, Reliability, Security) using the convention in classification.md. This convention is shared across all task types.
4. Filter to scope and exclude duplicates and cancelled records consistently (see scope-and-exclusions.md).
5. Compute the task-specific metrics following the per-type reference below.
6. Order and round exactly as each task's template specifies — ordering and precision are scored.

Per-type references

• Portfolio mix → portfolio-mix.md
• SLA aging / breach audit → sla-aging.md
• Release readiness → release-readiness.md

Status vocabulary (authoritative status)

• Closed/done set (counts as complete/closed): Closed, Done, Verified, Deployed.
• Open set (still in flight): Backlog, In Progress, Review, Reopened.
• Duplicate: status == "Duplicate" (regardless of whether duplicate_of is set).
• Cancelled: status == "Cancelled".

A record is a duplicate when status == "Duplicate" or duplicate_of is not null. A record is primary when it is neither a duplicate nor cancelled. (A Duplicate status with a null duplicate_of is still a duplicate — exclude it; it is not primary.)

Ordering and precision (universal)

• Work-item ID lists: sort lexicographically as strings unless the task says otherwise (digits sort before letters, so an id ending in 099 precedes one ending in S001).
• "closed_at ascending, then id ascending" is the default ordering for portfolio work.
• Percentages: percentage points rounded to 1 decimal.
• Rates (breach, readiness): rounded to 3 decimals.
• Dates are ISO YYYY-MM-DD. Date comparisons are calendar-day comparisons.

Worked discipline

• Re-derive every value from the environment data; do not hand-wave.
• After computing, re-check that every required field from the task's answer_template.json is present with the right type and key names, and that no extra fields are added (additionalProperties: false is common).
• If the task gives a scope_id, look up the matching row in mix_targets to recover the intended team_group and product_area for the scope.