test-status-report

name	test-status-report
description	Run the SDK test suite and produce a structured status report (per-file pass/fail, totals, duration, failure summaries). Use when the user says "test report", "status report", "are tests green", "run the suite and report", "what's the test state", or before/after publishing.

test-status-report

Wraps npm test → parses → emits a clean status report in the BROski badge style. Use before every publish (gate for sdk-publish) and after any non-trivial code change.

The Run

cd "H:\HyperAgent-SDK"
npm test 2>&1 | Tee-Object -FilePath "$env:TEMP\hyper-test-output.log"

npm test runs node --test tests/*.test.js. Output is TAP format. Capture stdout+stderr.

The Report Format

🧪 HyperAgent-SDK Test Report
=============================

📅 Run:        2026-MM-DD HH:MM:SS
⏱️  Duration:  X.XXs
🏷️  Version:   0.3.0
🌿 Branch:    main
🔖 Commit:    <short-sha>

📊 RESULTS
──────────
  ✅ tests/validate.test.js   21/21 pass   (Y.Yys)
  ✅ tests/registry.test.js   15/15 pass   (Y.Yys)
  ✅ tests/init.test.js        7/7  pass   (Y.Yys)
  ✅ tests/client.test.js     14/14 pass   (Y.Yys)
  ──────────────────────────────────────────
  TOTAL                       57/57 pass

🎯 STATUS: ALL GREEN ✅

If failures exist:

❌ STATUS: 2 FAILED

  tests/client.test.js  13/14 pass  ❌
    ✗ awardFromCourse — 401 path
      Expected error code 'BAD_STATUS', got 'TIMEOUT'
      at tests/client.test.js:142:12

  tests/registry.test.js  14/15 pass  ❌
    ✗ computeBadges — Multi-Tool boundary
      Expected ['🔧 Multi-Tool'], got []
      at tests/registry.test.js:88:5

🔧 NEXT ACTIONS
  1. node --test tests/client.test.js   ← isolate the failing file
  2. Read cli/client.js for the 401 path
  3. Use `token-sync-debug` skill if it's a real awardFromCourse regression

How to Build the Report

Order of operations:

1. Capture metadata (run before the test):

   $version = (Get-Content package.json | ConvertFrom-Json).version
   $branch  = git rev-parse --abbrev-ref HEAD
   $commit  = git rev-parse --short HEAD
   $start   = Get-Date
   

2. Run + capture:

   $output = npm test 2>&1
   $exit   = $LASTEXITCODE
   $end    = Get-Date
   $duration = ($end - $start).TotalSeconds
   

3. Parse TAP-ish Node test output. Look for:

   • # tests <N> → total tests
   • # pass <N> → passes
   • # fail <N> → fails
   • not ok <N> - <name> → failing test names
   • Indented at <file>:<line> → failure location
   • Per-file: each test file emits its own block; group by file path

4. Emit the report. If $exit -eq 0 → ALL GREEN. Else → list each failing test with its file, name, expected/actual, location.

5. Save the report for regression comparison:

   $report | Out-File "$env:TEMP\hyper-test-report-latest.txt"
   

   Next run can diff against this to detect new failures vs. pre-existing.

Quick One-Shot (when Bro just wants the badge)

cd "H:\HyperAgent-SDK"
$out = npm test 2>&1
if ($LASTEXITCODE -eq 0) {
  $count = ($out | Select-String '# tests (\d+)').Matches.Groups[1].Value
  Write-Host "✅ $count/$count GREEN" -ForegroundColor Green
} else {
  $fails = ($out | Select-String '# fail (\d+)').Matches.Groups[1].Value
  Write-Host "❌ $fails failing — run full report" -ForegroundColor Red
}

Regression Detection (compare to last run)

$prev = Get-Content "$env:TEMP\hyper-test-report-latest.txt" -ErrorAction SilentlyContinue
# After current run, diff the failing-test sets:
#   - Tests failing now AND in $prev → existing failures (carry over)
#   - Tests failing now but NOT in $prev → 🚨 NEW REGRESSION (called this out loudly)
#   - Tests in $prev but passing now → ✅ NEWLY FIXED (celebrate)

Hard Rules

• Never claim green without running — always actually execute npm test. No shortcuts.
• Never trim failures — show all of them, not just the first
• Always include version + commit + branch — the report is useless without provenance
• If the run itself errors (e.g. node missing, can't find tests dir) → report the run-level failure, don't pretend it was a test failure

Expected Baseline (as of v0.3.0)

tests/validate.test.js   21
tests/registry.test.js   15
tests/init.test.js        7
tests/client.test.js     14
─────────────────────────
TOTAL                    57

If TOTAL drops below 57 → tests were deleted or skipped, that's a regression by itself. Surface it.

If TOTAL is above 57 → great, new tests landed. Update the baseline in this skill + in hypercode-sdk skill.

Pairs With

• sdk-publish — uses this skill as a pre-flight gate (gate 3)
• phase-4-graduation — when adding tests/graduate.test.js, update baseline
• token-sync-debug — when tests/client.test.js regresses, this is the entry point