health-pulse

name	health-pulse
description	Health Pulse data model, status rules, scoring algorithms, and thresholds — the canonical source the extension consumes
tier	standard
applyTo	**/*health*,**/*pulse*,**/*status*,**/*scoring*
disable-model-invocation	true
currency	2026-04-22T00:00:00.000Z

Health Pulse Skill

Canonical source of truth for the Health Pulse feature. The extension reads these rules; the brain defines them.

Data Model

type HealthStatus = 'healthy' | 'attention' | 'critical';

interface HealthPulse {
  status: HealthStatus;

  // Connections
  connectionCount: number;
  connectionHealth: number; // percentage 0-100

  // Rituals
  lastDreamDate: Date | null;
  dreamNeeded: boolean;
  lastMeditationDate: Date | null;
  meditationCount: number;

  // Architecture
  trifectaCompleteness: number; // percentage 0-100
  architectureVersion: string;
  extensionVersion: string;
  syncStale: boolean;

  // Chat memory
  chatMemoryLines: number;
  chatMemoryWarning: boolean;
}

B1: Health Status Rules

Three states, evaluated in priority order. First matching rule wins.

Critical

Rule	Condition	Rationale
Broken connections	connectionHealth < 90	Architecture integrity compromised
Sync failure	syncStale && daysSinceSyncCheck > 3	Heir running stale architecture
Dream overdue + issues	dreamNeeded && daysSinceLastDream > 14	Two weeks without diagnostics while issues exist

Attention

Rule	Condition	Rationale
Dream recommended	dreamNeeded	Issues detected but not urgent
Dream stale	daysSinceLastDream > 7	Weekly cadence missed
Memory bloat	chatMemoryWarning	Chat memory exceeds threshold
Sync stale	syncStale	Heir behind master
Meditation gap	daysSinceLastMeditation > 7	No consolidation in a week
Trifecta gaps	trifectaCompleteness < 80	High-use skills missing components

Healthy

All other cases. No conditions matched above.

Status Computation

function computeHealthStatus(pulse: HealthPulse): HealthStatus {
  const daysSinceDream = pulse.lastDreamDate
    ? daysBetween(pulse.lastDreamDate, now())
    : Infinity;
  const daysSinceMeditation = pulse.lastMeditationDate
    ? daysBetween(pulse.lastMeditationDate, now())
    : Infinity;

  // Critical rules
  if (pulse.connectionHealth < 90) return 'critical';
  if (pulse.syncStale && daysSinceDream > 3) return 'critical';
  if (pulse.dreamNeeded && daysSinceDream > 14) return 'critical';

  // Attention rules
  if (pulse.dreamNeeded) return 'attention';
  if (daysSinceDream > 7) return 'attention';
  if (pulse.chatMemoryWarning) return 'attention';
  if (pulse.syncStale) return 'attention';
   (daysSinceMeditation > )  ;
   (pulse. < )  ;

   ;
}

B2: Meditation History Parsing

Meditation history is tracked through file artifacts, not a dedicated log.

Sources

Source	What it tells you	How to read
/memories/repo/ entries	Repo facts created during meditation	File modification dates
/memories/ user memory	Cross-project patterns	File modification dates
.github/quality/dream-report.json	Last dream run timestamp	JSON timestamp field
Git log	Files changed in meditation commits	git log --grep="meditate|consolidat"

Metrics

Metric	Computation	Use
meditationCount	Count of meditation artifacts this month	Activity level
lastMeditationDate	Most recent meditation artifact date	Staleness check
daysSinceLastMeditation	now() - lastMeditationDate	Attention trigger (>7 days)

Heuristic

Since meditation doesn't write to a central log, detect it by:

1. Check git log for commits with meditation keywords in the message
2. Check modification dates on /memories/repo/ files
3. Check .github/quality/ for recent reports
4. Fall back to workspace state stored by the extension

B3: Dream State Markers

Dream produces diagnostic artifacts. These markers tell the extension when dream last ran and what it found.

Primary Marker

File: .github/quality/dream-report.json

{
  "timestamp": "2026-04-16T03:00:00Z",
  "skills": 173,
  "instructions": 133,
  "prompts": 81,
  "agents": 17,
  "trifectaIssues": 12,
  "brokenRefs": 3,
  "status": "ATTENTION_REQUIRED"
}

Derived Flags

Flag	Derivation	Threshold
dreamNeeded	brokenRefs > 0 || trifectaIssues > 20	Any broken refs or excessive trifecta gaps
daysSinceLastDream	now() - report.timestamp	>7 = attention, >14 = critical (if dreamNeeded)
connectionHealth	1 - (brokenRefs / totalConnections) * 100	<90% = critical

No Report Found

If .github/quality/dream-report.json doesn't exist:

• dreamNeeded = true
• lastDreamDate = null
• daysSinceLastDream = Infinity
• Nudge: "Run /dream to establish a health baseline"

B4: Connection Health Scoring

Connection health measures the integrity of applyTo patterns and file references across the architecture.

Algorithm

function computeConnectionHealth(dreamReport: DreamReport): number {
  const total = dreamReport.skills + dreamReport.instructions + dreamReport.prompts;
  if (total === 0) return 100;

  const brokenRatio = dreamReport.brokenRefs / total;
  const health = Math.round((1 - brokenRatio) * 100);
  return Math.max(0, Math.min(100, health));
}

Thresholds

Health %	Status	Meaning
95-100	Healthy	Normal — minor issues from recent edits
90-94	Attention	Several broken references — run meditation to fix
<90	Critical	Architecture integrity at risk — fix immediately

What Counts as a Broken Connection

Type	Detection	Severity
applyTo pattern matches no files	Glob returns empty	Low (pattern may be aspirational)
Skill references nonexistent instruction	Cross-reference check	High
Instruction references moved file	Path validation	High
Prompt references nonexistent agent	Agent name check	Medium

B5: Chat Memory Audit Thresholds

VS Code Copilot Chat memory (/memories/) accumulates over time. Large memories waste context tokens.

Thresholds

Lines	Status	Action
0-150	Healthy	No action
151-200	Attention	Review for consolidation opportunities
>200	Warning	chatMemoryWarning = true — prune redundant entries

Audit Procedure

1. Count total lines across all files in /memories/ (excluding /memories/session/ and /memories/repo/)
2. Flag entries that are:
   • Duplicated across files
   • Outdated (references old versions, removed features)
   • Too verbose (>5 lines for a single fact)
3. Suggest consolidation: merge related entries, remove stale facts

Memory Efficiency Score

function computeMemoryEfficiency(lines: number): number {
  if (lines <= 150) return 100;
  if (lines <= 200) return Math.round(100 - ((lines - 150) / 50) * 30);
  return Math.max(0, Math.round(70 - ((lines - 200) / 100) * 70));
}

Score	Meaning
80-100	Efficient — good information density
50-79	Adequate — some bloat
<50	Needs curation — significant waste

Visual Indicators

Status	Icon	Color	CSS Variable
Healthy	✅	Green	--vscode-testing-iconPassed
Attention	⚠️	Amber	--vscode-editorWarning-foreground
Critical	🔴	Red	--vscode-editorError-foreground

Extension Integration

The extension reads these rules by:

1. Running dream-cli.cjs to produce the dream report
2. Parsing .github/quality/dream-report.json for markers
3. Applying the status computation algorithm above
4. Rendering the Health Pulse card with the computed status

The skill defines the rules; the extension implements the computation.