gse-health

name	gse-health
description	Display project health dashboard with up to 8 dimensions (mode-dependent: 8 Full / 3 Lightweight / disabled Micro). Triggered by /gse:health.

GSE-One Health — Health Dashboard

Arguments: $ARGUMENTS

Options

Flag / Sub-command	Description
(no args)	Show full health dashboard
--dimension DIM	Show detail for a specific dimension
--trend	Show health score trend across sprints
--alerts-only	Show only dimensions with risk alerts
--help	Show this command's usage summary

Prerequisites

Before executing, read:

1. .gse/status.yaml — current health scores, sprint state, and engagement counters (consecutive_acceptances — the ai_integrity input)
2. .gse/config.yaml — health.disabled_dimensions (if any)
3. .gse/backlog.yaml — task statuses for computation
4. .gse/profile.yaml — apply P9 Adaptive Communication to all output
5. docs/sprints/sprint-{NN}/review.md — review findings (if exists)

Workflow

Step 1 — Compute Health Dimensions

Read .gse/config.yaml → lifecycle.mode:

• Micro mode — Skip health computation entirely. Display Inform note: "Health dashboard is disabled in Micro mode (minimal persistent state). Run /gse:go to upgrade to Lightweight or Full if you want quality telemetry." Exit the activity.
• Lightweight mode — Compute only 3 of 8 dimensions per spec §13.2: test_pass_rate, review_findings, git_hygiene. The other 5 (requirements_coverage, design_debt, complexity_budget, traceability, ai_integrity) are marked "not computed in Lightweight mode" in the detail view and excluded from the overall score denominator.
• Full mode — Compute all 8 dimensions (default behavior below).

If config.yaml specifies health.disabled_dimensions, also exclude those from the set active for the current mode, and adjust the overall score denominator accordingly.

Calculate each active health dimension on a 0-10 scale (spec §7.1).

1. requirements_coverage (0-10)

• Count requirements (REQ-NNN) that have at least one traced implementation
• Score = (traced requirements / total requirements) * 10
• If no requirements artefact exists: score = N/A (excluded from overall)

2. test_pass_rate (0-10)

• Read latest test campaign results from backlog task entries
• Score = (passing tests / total tests) * 10
• If no tests exist: score = 0 with alert "No tests found"

3. design_debt (0-10)

• Count review findings tagged as architectural/design concerns
• Score = 10 - (design_findings * weight_per_finding)
• Weight: HIGH = 2.0, MEDIUM = 1.0, LOW = 0.5
• Floor at 0

4. review_findings (0-10)

• Count unresolved review findings by severity
• Score = 10 - (HIGH * 1.5 + MEDIUM * 0.8 + LOW * 0.3)
• Floor at 0

5. complexity_budget (0-10)

• Score = (remaining_budget / total_budget) * 10
• Alert if below 2: "Sprint budget nearly exhausted"

6. traceability (0-10)

• Count tasks with proper traces (requirement -> design -> code -> test)
• Score = (fully_traced_tasks / total_tasks) * 10
• Partial traces count proportionally

7. git_hygiene (0-10)

Computed from 6 sub-metrics (each on 0-10, then weighted):

Sub-metric	Weight	Score 10	Score 0
Active branches	20%	≤3	>10
Stale branches	20%	0	>3
Uncommitted changes	20%	0 across all worktrees	>5 files
Merge conflicts	20%	0	>0
Main clean	10%	Clean + tagged	Dirty
Unreviewed branches	10%	0	>2

git_hygiene = sum(sub_metric * weight)

8. ai_integrity (0-10)

Computed from 3 factors (each on 0-10, then weighted):

Factor	Weight	Computation
Unverified assertions	40%	Count of [AI-INTEGRITY] findings from devil's advocate
Hallucination findings	30%	Libraries or APIs flagged as non-existent
User engagement	30%	Based on consecutive_acceptances — high count indicates complaisance risk

Engagement scoring:

• 0-3 consecutive acceptances: 10 (healthy interaction)
• 4-6 consecutive acceptances: 7.5 (mild concern)
• 7-9 consecutive acceptances: 5 (notable pattern)
• 10+ consecutive acceptances: 2.5 (significant concern — user may not be reviewing)

Step 2 — Compute Overall Score

overall = sum(enabled_dimensions) / count(enabled_dimensions)

Dimensions listed in health.disabled_dimensions are excluded from both numerator and denominator.

Step 3 — Risk Alerts

Generate actionable warnings for concerning dimensions:

RISK ALERTS
  [!] test_pass_rate: 4.5/10 — 3 failing tests in TASK-005
      Action: run /gse:fix --task TASK-005
  
  [!] git_hygiene: 5.5/10 — 2 stale branches, 1 worktree with uncommitted changes
      Action: review stale branches, commit or stash uncommitted work
  
  [!] ai_integrity: 6.0/10 — 8 consecutive user acceptances without modification
      Action: review recent outputs critically, consider challenging assumptions

Threshold for alert: dimension score below 7.

Step 4 — Trend Display

If --trend is specified, show health evolution across sprints:

HEALTH TREND
  Sprint | Overall | Req  | Test | Debt | Review | Budget | Trace | Git  | AI
  S01    |   7.2   | 8.0  | 6.5  | 7.0  |  7.5   |  9.0   | 6.0   | 7.0  | 6.5
  S02    |   7.8   | 8.5  | 7.0  | 7.5  |  8.0   |  8.5   | 7.0   | 7.5  | 8.0
  S03    |   6.8   | 8.5  | 4.5  | 7.0  |  6.0   |  4.0   | 7.5   | 7.0  | 6.0
                            ^^                        ^^
                        declining                 declining

Mark dimensions with declining trend (2+ consecutive drops).

Step 5 — Dimension Detail

If --dimension DIM is specified, show detailed breakdown:

DIMENSION: git_hygiene (7.5/10)

Sub-metrics:
  Active branches:     10/10 (weight 20%)  — 3 active (≤ 3 → full score)
  Stale branches:       6/10 (weight 20%)  — 2 stale: gse/sprint-01/feat/old, gse/sprint-01/doc/readme
  Uncommitted changes:  4/10 (weight 20%)  — .worktrees/sprint-02-feat-api has 5 uncommitted files
  Merge conflicts:     10/10 (weight 20%)  — no conflicts detected
  Main clean:          10/10 (weight 10%)  — main matches last delivery
  Unreviewed branches:  5/10 (weight 10%)  — 1 unreviewed: gse/sprint-02/feat/api

Recommendations:
  1. Commit or stash changes in sprint-02-feat-api worktree
  2. Review or delete stale branches from sprint-01
  3. Run /gse:review for unreviewed feature branches

Step 6 — Persist

1. Update status.yaml with computed health scores
2. If trend data does not exist, initialize docs/health-history.yaml with current sprint scores
3. Append current sprint scores to health history for future trend analysis