Design Principle: A human who has never seen the factory should be able to look at the dashboard and understand what is happening, what has happened, and what needs attention — within 30 seconds. Every view answers a specific human question. Every panel has a reason to exist. Every indicator maps to an action the human can take.
Build Implementation Map
Each governance dashboard has a dedicated page shell containing its full build specification. To build a dashboard, give Claude Code the instruction shown below.
| Dashboard View |
Page Shell |
Spec Reference |
Claude Code Build Instruction |
| 1. Mission Control |
factory-mission-control.html |
S19 + View 1 |
"Read the build spec on factory-mission-control.html and implement all panels" |
| 2. Build Activity |
factory-build-activity.html |
S14, S15, S18 + View 2 |
"Read the build spec on factory-build-activity.html and implement all panels" |
| 3. Verification & Quality |
factory-verification.html |
S3, S6, S7, S15 + View 3 |
"Read the build spec on factory-verification.html and implement all panels" |
| 4. Failure & Recovery |
factory-failures.html |
S4, S5, S9, S17 + View 4 |
"Read the build spec on factory-failures.html and implement all panels" |
| 5. Budget & Resources |
factory-budget.html |
S16 + View 5 |
"Read the build spec on factory-budget.html and implement all panels" |
| 6. Configuration & Progress |
factory-configuration.html |
S6, S7, S8, S12 + View 6 |
"Read the build spec on factory-configuration.html and implement all panels" |
| 7. System Health & Validation |
factory-health.html |
S19 + View 7 |
"Read the build spec on factory-health.html and implement all panels" |
Each page shell contains an amber BUILD STATUS banner and the full panel-by-panel specification including chart types, data sources, refresh rates, and interaction models. When a dashboard is built, the amber banner is removed and replaced with the actual dashboard components. The build spec remains in the page source as an HTML comment for reference.
- System Health Traffic Light: single GREEN / YELLOW / RED indicator — GREEN = all health checks passing + last canary passed within 24 hours; YELLOW = health checks passing but canary stale > 24 hours; RED = any health check failing or last canary FAILED or active HARD_HALT
- Build Pipeline Status Bar: horizontal flow showing packet counts at each stage — QUEUED → ASSIGNED → EXECUTING → VERIFYING → COMPLETED — with FAILED packets in red below the bar
- Active HARD_HALT indicator: if any system-wide halt is active, a prominent red banner with the halt reason and time since halt
- Last Canary Test result: PASS/FAIL with timestamp, latency, and cost — red highlight if > 24 hours since last run
- Phase Progress Bars: 11 horizontal bars (Phase 0–10) showing total packets, completed, in-progress, and failed per phase — active phase highlighted, future phases greyed, completed phases with green checkmark
- Today’s Key Metrics: packets completed, packets failed, API calls made, budget spent vs $5.00 remaining, average build time, verification pass rate — each with trend arrow (↑↓→) compared to 7-day average
- Active Alerts feed: scrollable list of CRITICAL / WARNING / INFO alerts from stop conditions, canary failures, health check failures, budget warnings, and escalations — most recent first
- Next pending human action: if any escalation is waiting for human response, show it prominently with “Action Required” badge
S3
S9
S18
S19
- Agent Pool Grid: all Builder, Verifier, and Chief Engineer instances — each showing agent_id, type, status (IDLE = green, BUSY = blue, UNRESPONSIVE = red), current packet if busy, elapsed time, last heartbeat timestamp
- Agent Availability Summary: “X/Y Builders idle, X/Y Verifiers idle, Chief Engineer: idle/busy” — red warning if no idle agents and packets are queued
- Queue Depth Monitor: current queue depth with alert if > 20 packets waiting — shows average wait time and longest-waiting packet
- Live Packet Tracker table: all non-terminal packets with packet_id, capability_block name, phase, current status (colour-coded), assigned agent, elapsed time, retry count — sortable by any column, click for full execution history
- Execution Timeline: Gantt-style horizontal chart showing each packet’s lifecycle over the past 24 hours — colour segments for time in each status, red diamonds at failure points
- Concurrent Build Monitor: number of simultaneous builds running vs maximum (3), with API call rate and token consumption per minute
- Agent Assignment Log: recent assignments with packet → agent mapping, timestamp, and reason (priority order, phase order, dependency order)
S14
S15
S17
S18
- Verification Pass Rate: large percentage with trend sparkline over 30 days — target ≥ 95%
- Two-Agent Discipline Audit: “100% enforced (X checks, 0 violations)” — this must always show 100%, any deviation is a HARD_HALT — shows last 20 independence checks with builder_id ≠ verifier_id confirmation
- Proof Object completeness: “X/Y components have complete proof chains” — any component without a full chain (source → requirement → block → packet → build → verification → proof → manifest → claim) is flagged
- Implementation Claim Ledger integrity: last integrity check timestamp, status (INTACT / TAMPERED), total claims filed
- Recent Verification Results table: packet_id, verifier_agent, result (PASS/FAIL), failing step if FAIL, duration, timestamp — click for full Verification Manifest and all Proof Objects
- Code Quality Gauges: 6 gauges with current value, threshold, and 30-day trend sparkline — Lint cleanliness (% clean, threshold: 100%), Dependency health (vulnerability count, threshold: 0 high/critical), Unused code (% dead code, threshold: 0%), Cyclomatic complexity (avg per function, threshold: ≤ 10), Security score (vulnerability count, threshold: 0 high/critical), Code duplication (% duplicated, threshold: ≤ 5%)
- Proof Object Registry: searchable table of all proof objects by type, associated packet, component, storage reference, and timestamp — filter by type (test log, compliance report, API log, etc.)
- Template Compliance Rate: percentage of builds passing all template verification_rules, with breakdown by template type
S3
S6
S7
S15
- Active Failures count: large number with severity breakdown — X FAILED, Y REPAIRING, Z ESCALATED — ESCALATED packets highlighted prominently as they require human action
- Chief Engineer Resolution Rate: percentage of failures resolved without human escalation — trend over 30 days — declining rate indicates systemic issues
- Mean Time to Recovery: average time from failure detection to successful re-verification — by failure category
- Escalation Queue: all packets awaiting human resolution with escalation_id, packet_id, category, time waiting, recommended actions — each with action buttons: “Approve Repair”, “Modify Architecture”, “Override and Resume”, “Suspend Build”
- Chief Engineer Activity Log: all interventions with diagnosis_id, packet_id, failure classification, root cause, confidence score, outcome (REPAIR/ESCALATE), duration — click for full Diagnostic Report
- Failure Pattern Analysis: most common failure types (pie chart), failure rate by phase (bar chart), failure rate trend (line chart) — if same root cause appears 3+ times: highlighted as “Systemic Issue — architecture review recommended”
- Repair Instruction History: all repair instructions with repair_type (RETRY_SAME / RETRY_MODIFIED / REBUILD_DEPENDENCY / TEMPLATE_UPDATE), success rate per type, average retries to resolution
- Confidence Score Distribution: histogram of Chief Engineer root_cause_confidence scores — low average (< 0.5) indicates diagnostic context is insufficient
S4
S5
S9
S17
- Daily Budget Gauge: circular gauge showing $X.XX spent of $5.00 — green (< $3.50), yellow ($3.50–$4.50), red (> $4.50) — with priority tier indicator: “Normal” / “HIGH+ only” / “CRITICAL only”
- Projected End-of-Day Spend: based on current hourly rate — warning if projected to exceed budget
- Budget Alert Status: any active budget warnings (80% threshold reached, CRITICAL reserve activated, emergency override in effect)
- Canary Test Cost: today’s canary spend vs $0.10 target — alert if canary exceeding expected cost
- Cost Breakdown: stacked bar chart — daily cost by agent type (Builder / Verifier / Chief Engineer / Canary) over past 7 days
- Per-Packet Cost Table: today’s most expensive packets with packet_id, model used (Sonnet/Opus), token count (in/out), cost, and API calls made
- Model Distribution: “Sonnet: XX% of calls ($X.XX) / Opus: XX% of calls ($X.XX)” — with cost-per-completion comparison
- 30-Day Spending History: line chart of daily spend with $5.00 target line — days exceeding $4.50 highlighted red
- Cost Attribution: per-phase cost breakdown — shows which phases are most expensive and why (complexity, retries, Opus usage)
S14
S16
S19
- Template Registry Status: X templates ACTIVE, Y DEPRECATED, Z DRAFT — alert if any required base template is not ACTIVE
- Prompt Version Status: current active prompt_version with last update date — yellow warning if any prompt references an outdated specification version
- Specification Version: current Hilbert Factory specification version with last modification date and section count
- Knowledge Graph Health: entity count, relationship count, last integrity check, query latency p95
- Capability Block Map Visualisation: the 7-layer map with per-block status — NOT_STARTED (grey), IN_PROGRESS (blue), COMPLETE (green), MODIFIED (orange) — click any block to see its packets and their statuses
- Phase Completion Matrix: grid of Phase 0–10 × status showing which phases are complete, active, or blocked — with blocking reason for blocked phases
- Template Usage Report: which templates are used most, which agents use them, recent template evolution proposals
- Build Velocity Forecast: at current completion rate, estimated date for each remaining phase — “Phase 3 estimated completion: April 12, 2026”
S6
S7
S8
S12
- Component Health Grid: 10 components each with green/yellow/red indicator, response time (ms), uptime percentage over 24 hours — PostgreSQL, Knowledge Graph, Template Registry API, Prompt Registry, Artifact Storage, Claude API, Build Orchestrator, Prompt Execution Engine, Builder Agent Pool, Verifier Agent Pool
- System Uptime: overall factory uptime percentage over 30 days — target ≥ 99%
- Last Pre-Production Validation: date, result, categories passed/failed — alert if > 30 days since last full validation
- Factory Commissioning Status: COMMISSIONED (date) or NOT COMMISSIONED — with link to the Factory Commissioning Proof Object
- Canary Test History: table of recent runs with trigger (SCHEDULED/MANUAL), result (PASS/FAIL), failing step if FAIL, latency, cost, timestamp — most recent at top with large PASS/FAIL indicator
- Canary Pass Rate: percentage over past 30 days — should be 100%, any deviation is a red flag with trend chart
- Run Canary Test Button: prominent button with real-time progress display — Readiness ✓ → Queued ✓ → Building... → Verifying... → Complete ✓ or ✗ at failing step — 5-minute cooldown between manual runs
- Run Health Check Button: adjacent button — runs all 10 component checks immediately, results populate the Component Health Grid above
- Health Check History: per-component 24-hour uptime chart — click any component to see error log and response time distribution
S9
S19
Human Action Map — Where to Find Answers
Every human question mapped to the dashboard and panel that answers it.
| Human Question | Dashboard | Panel |
|---|
| “Is the factory working?” | 1. Mission Control | System Health Traffic Light |
| “What’s the overall progress?” | 1. Mission Control | Phase Progress Bars |
| “What needs my attention right now?” | 1. Mission Control | Active Alerts / Next Pending Action |
| “Where is my packet?” | 2. Build Activity | Live Packet Tracker |
| “Is anything stuck?” | 2. Build Activity | Queue Depth Monitor |
| “How long are builds taking?” | 2. Build Activity | Execution Timeline |
| “Is the code good quality?” | 3. Verification | Code Quality Gauges |
| “Can I trust the verification is independent?” | 3. Verification | Two-Agent Discipline Audit |
| “Where is the proof this component was built correctly?” | 3. Verification | Proof Object Registry |
| “What broke?” | 4. Failures | Active Failures |
| “Is this a recurring problem?” | 4. Failures | Failure Pattern Analysis |
| “What decision do you need from me?” | 4. Failures | Escalation Queue |
| “How much is this costing?” | 5. Budget | Daily Budget Gauge |
| “Where is the money going?” | 5. Budget | Cost Breakdown + Per-Packet Cost |
| “How much of the system has been built?” | 6. Configuration | Capability Block Map Visualisation |
| “What templates are we using?” | 6. Configuration | Template Registry Status |
| “When will this phase be done?” | 6. Configuration | Build Velocity Forecast |
| “Is the factory healthy right now?” | 7. Health | Component Health Grid |
| “Prove the factory works” | 7. Health | Run Canary Test Button |