S●PRO / Project Assurance Talk to us
Live snapshot · Reporting period: 2026-05-04 — 2026-05-05

Maintaining The Hub
across 0 codebases
and 0 lines of code.

A real, in-production case from S·PRO — a multi-year engagement covering mobile, web, admin and backend. AI is woven through code review, documentation, regression analysis and reporting. Below is what that actually produces.

See AI in the workflow Jump to evidence
0
Vulnerabilities
across all 4 repositories
A/A/A/A
Maintainability
SonarQube rating, all repos
0%
Regression pass rate
excl. blocked, last 3 runs
0
Test cases mapped
in Testomatio · eBoard
The Hub stack React Native + Expo SDK Angular Webapp Node Backend Admin Panel Telegraph (Chat)
01 · The spine of how we work

AI-driven engineering, end-to-end

Every artifact you see below — code review notes, gap analysis, regression summaries — was either produced or accelerated by AI. Not as a gimmick: as the spine of a long-term maintenance contract.

1
Plan & Spec
Confluence + AI doc-gap analysis
Feature specs validated against the test plan; missing pages flagged before code is written.
2
Implement
Claude Code · in-IDE pair
Engineers ship features with an AI pair: skeleton generation, refactor suggestions, security hints.
3
Review
SonarQube + AI explainers
Every MR is scanned. Each issue is opened with a code excerpt, plain-language explanation and a suggested fix.
4
QA & Regression
Claude QA runs · Testomatio
Reports are aggregated and triaged by AI: pass / fail / blocked, with risk-ranked sub-suites.
5
Report
Executive summaries
Stakeholders get a one-page picture every week — generated, not hand-typed.
⟢

Reports as a primitive

The aggregated regression summary on this page comes from three locally-saved HTML test reports from prior Claude QA runs — the AI does the bookkeeping so QA can stay on judgment.

Source: eBoard Summary, 2026-05-04
⌬

Issues with explanations

SonarQube Community 25.9 is wired into CI (runs on every MR and push to develop/master). Each finding lands with a fix recipe — not a stack trace.

Source: HUB SonarQube Report, 2026-05-05
◈

Gap analysis you can act on

AI reads the Confluence spec and the Testomatio test set side-by-side and tells you exactly where the spec lags the tests, or vice-versa.

Source: eBoard Gap Analysis, 2026-05-05
0
AI-generated reports
feeding this dashboard
0
Test cases automatically
reconciled against spec
0
Cross-cutting topics surfaced
by AI doc-gap analysis
0
Code-smell findings, each
with an AI-suggested fix
02 · Documentation state

We document the spec the way we run the tests — page by page.

For the eBoard feature alone, 7 Confluence pages cover 287 Testomatio tests. AI reconciles both sides every reporting cycle and flags drift before it ships.

0
Total tests
(documented + cross-cutting)
0
Confluence pages
in the eBoard set
0/0
v1.0 / v1.1 distribution
of mapped tests
0%
Undocumented tests
flagged for spec follow-up

Coverage of documented requirements per page

eBoard · v1.0 + v1.1
Confluence page Ver Spec items Tests Coverage Verdict
Create a Custom Boardv1.02267
64%
 MEDIUM
View Board – Webv1.0105
50%
 MEDIUM
View Board – TV / Fullscreenv1.0917
33%
 HIGH
View Board – Mobilev1.070
0%
 CRITICAL
Date-Based Viewv1.120103
90%
 LOW
Locked Columnsv1.120143
90%
 LOW
Fullscreen Table Formattingv1.1041
n/a
 CRITICAL · spec missing
Coverage = test items directly verifying a documented requirement. Two pages flagged CRITICAL by AI gap-analysis are now in the action plan with named owners.

Top 5 gaps · severity-ranked

AI-prioritised
  1. 1
    "Fullscreen Table Formatting" page does not exist
    41 tests · 0 pages · Doc gap
  2. 2
    View Board – Mobile: 0 tests
    0 / 7 reqs (0%) · Test gap
  3. 3
    v1.1 cross-cutting topics undocumented
    11 topics · audit, perms, network, a11y… · Doc gap
  4. 4
    View Board – TV: missing negative tests
    ~5 missing reqs · Test gap
  5. 5
    Create Board: Shared Mode + UI/UX boundaries
    ~6 missing reqs · Test gap

Action plan · in flight

Owners assigned
  • CRITICALBACreate "Fullscreen Table Formatting" Confluence page
  • CRITICALQAWrite View Board – Mobile suite (~25–30 cases)
  • HIGHBAAdd cross-cutting sections to v1.1 pages (11 topics)
  • HIGHQATV/Fullscreen negative tests + View Mode Toggle (~6)
  • MEDIUMBAUnify "All" filter spec across pages
  • MEDIUMQAv1.0 boundary + Shared Mode + step indicator (~10)
  • LOWQAWeb View: lazy-load, last-updated, collapse (~5)
03 · Code health · SonarQube

0 lines of code, scanned on every push.

SonarQube Community 25.9 runs on every merge request and push to develop/master across all four repositories. Zero open vulnerabilities. All four rated A on Security and Maintainability.

hub-v2-telegraph Chat
0 LoC
A E A
MaintRelSec
  • Bugs2
  • Vulnerabilities0
  • Code smells26
hub-v2-backend Node
0 LoC
A D A
MaintRelSec
  • Bugs43
  • Vulnerabilities0
  • Code smells278
hub-v2-admin Angular
0 LoC
A E A
MaintRelSec
  • Bugs26
  • Vulnerabilities0
  • Code smells490
hub-v3-webapp Angular
0 LoC
A C A
MaintRelSec
  • Bugs167
  • Vulnerabilities0
  • Code smells1,589

Issues by repository

Bugs · Code smells
Vulnerabilities are intentionally absent — the count is 0 across the board. Bugs and code smells are concentrated in the legacy webapp; new-code mode is on, so the trend is one-way down.

Lines of code distribution

119,675 total
Four codebases, one product. The dashboard you're reading collapses all four into a single picture every time SonarQube finishes a run.

Continuous integration

Source: sonarqube.s-pro.io
0%
of MRs scanned
0
open vulnerabilities
A
on Maintainability · all 4 repos
A
on Security · all 4 repos
25.9
SonarQube Community edition
04 · Regression coverage

Three Claude QA sessions across six weeks — reconciled into one report.

59 acceptance criteria executed across three AI-assisted sessions on the eBoard feature. Every result reconciled against the master plan in Testomatio (287 cases, 8 sub-suites).

Run outcomes

59 ACs · 3 runs
76.3%Pass rate (raw)
95.7%Pass rate (excl. blocked)
3.4%Defect density

Sub-suite execution

5 of 8 touched · 62.5%
Date-Based View
100%
Editing & Filtering
100%
Lock Column Data
60%
Editing Locked Column
blocked
Auto-Pop & Read-Only
40%
Interaction · DBV
blocked
Fullscreen Mode
100%
Fullscreen Table Formatting
not run
0
Passed ACs
0
Failed · isolated to one sub-suite
0
Blocked · pre-conditions queued
0
Bugs filed and tracked

What this means in practice

Translated for stakeholders
When tests can run
95.7% pass on the first attempt — a strong signal that recent code changes didn't regress core flows.
When tests are blocked
20.3% block rate flags missing fixtures, not bad code. AI summary names the exact pre-conditions to fix.
What's still untested
3 sub-suites untouched in the last cycle — already scheduled, with named owners and ETAs.
05 · The S·PRO approach

Three commitments we make on every long-term project.

01

Visibility, by default

Code health, doc state and regression coverage are first-class stakeholder data — generated continuously, never on request.

02

Automation as the floor

Every MR is scanned. Every report is regenerated. Quality rituals don't depend on a single engineer remembering to run them.

03

AI as a teammate

AI explains issues, drafts gap analyses, summarises runs and ranks risk — so humans spend their time on judgment, not bookkeeping.

Considering a long-term partner?

Let's run your project assurance the same way.

Fixed cadence, real numbers, AI throughout. The picture above is what your status review can look like — not a once-a-quarter PDF.

Start a conversation Visit s-pro.io