Nexus Hub

USER MANUAL · v1.8.9
01

OVERVIEW

The Nexus Hub extension analyzes Azure DevOps Work Items in real time, evaluating story quality based on the I.N.V.E.S.T framework and producing an actionable quality score.

The extension has three main modules:

Module 1 — Work Item Form: Analysis panel embedded inside every Work Item. Evaluates the open item in real time.
Module 2 — Backlog IQ (Backlog Intelligence): Strategic dashboard that consolidates quality across all teams in the project, with ranking and drill-down.
Module 3 — Predictive Analytics: Monte Carlo simulator that forecasts delivery dates and team capacity from real throughput history, with NLP outlier detection and AI semantic adjustment.
02

WORK ITEM FLOW

When you open any Work Item (User Story, Feature, or Epic) in Azure DevOps, the "Nexus Hub" tab appears automatically inside the form.

  1. 1Open a Work Item under Boards › Work Items.
  2. 2Click the "Nexus Hub" tab in the form's side panel.
  3. 3Click "Analyze" to start the analysis.
  4. 4Wait for processing — the system interprets the title, description, and acceptance criteria.
  5. 5Review the Score, the I.N.V.E.S.T dimensions, and any Alerts generated.
Important: the minimum input for analysis is the title field — if description and acceptance criteria are empty, the system runs the evaluation with reduced confidence.

SCORE EXAMPLES

03

SCORE & CLASSIFICATION

The score ranges from 0 to 100 and is the weighted average of the I.N.V.E.S.T criteria.

RANGECLASSIFICATIONMEANINGRECOMMENDED ACTION
90 – 100 Excellent Item is well written, clear, and actionable. Ready for review and sprint intake.
80 – 89 Good Acceptable, may need minor adjustments. Address minor points before refinement.
60 – 79 Warning Has structural gaps; refinement is needed. Schedule a refinement session with the team.
< 60 Critical Item is missing essential information. Block sprint intake · rewrite the story.
04

I.N.V.E.S.T CRITERIA

Six dimensions are evaluated independently and combined into the final score. Each receives a grade from 0 to 15 (max. 90), normalized to 100.

I· Independent 15%

The item can be delivered without depending on other stories.

N· Negotiable 10%

Not a contract — leaves room for conversation.

V· Valuable 25%

Delivers measurable value to a real persona.

E· Estimable 15%

The team can size the effort with reasonable confidence.

S· Small 15%

Fits in a single sprint without needing to be split.

T· Testable 20%

Has explicit, verifiable acceptance criteria.

05

BACKLOG IQ

The Backlog IQ dashboard ranks teams by average backlog quality and highlights recommended refinement actions per squad.

  1. 1Open Nexus Hub › Backlog IQ in the side bar.
  2. 2Filter by Sprint, Area Path, or Item Type using the toolbar.
  3. 3Click any team row to drill down into per-item scores.
  4. 4Use the Schedule action on red items to book a refinement session.
Tip: teams below 70 points for two consecutive weeks trigger an automatic alert in the Overview module.
06

PREDICTIVE ANALYTICS

The Monte Carlo engine runs 10,000 simulations against your team's throughput history to project delivery confidence intervals (p50/p85/p95).

p50 OPTIMISTIC

Best case — only the fastest 50% of simulations made it.

p85 TARGET

Recommended commitment — 85% of simulations delivered by this date.

p95 CONSERVATIVE

Worst case — used for buffer planning and stakeholder commitments.

AI Semantic Adjustment: when descriptions contain risk markers such as legacy, refactor, spike, or migration, the model applies a multiplier to historical variance — your forecast confidence reflects this automatically.
07

BEST PRACTICES

  1. 01.
    Run analysis early

    Trigger Nexus Hub during ideation, not at sprint planning. Catching gaps days ahead saves hours of refinement.

  2. 02.
    Don't force the score

    A 100/100 story nobody wants is worse than a valuable 68/100. Use the score as a signal, not as a target.

  3. 03.
    Pin a target date

    In Predictive Analytics, pin a date per release. Tracking confidence drift over time is more useful than a single snapshot.

  4. 04.
    Review NLP findings

    When the AI flags semantic risk, read the rationale (the "REASON" link) — that's where the most actionable refactors live.

Nexus Hub User Manual · v1.8.9 · 2026 by TRX22