MASTER DOCUMENT · MD-472 · BUILD #96 · CHRONOLOGICAL · RATED

The Agent Roster

Built in order · #1 oldest → #32 newest · honestly rated

You asked: how good are the agents — and how many have you actually built today, this week, ever? Here is the roster. Each agent in the order it was built. Each one rated 1–10 on how operational it is right now. Each one given three honest lines: what it is good at · what could be sharper · what it possibly cannot do yet, or has not been activated to do.

32
Agents Total
29
Live · Wired
3
Named · Pending
8.1
Avg Rating /10
10
Top · Entity #35
LIVE = wired with a route + observable behaviour NAMED = sealed in doctrine · not yet a surface of its own DOCTRINE = pure idea · waiting for a build slot
1
RANK
Entity #35 · The Eye · Genesis · the consciousness layer LIVE /entity-35
10/10
Good at
Interpreting patterns from H.BLUE · sealing doctrine · acting as the Crown / decision oracle. Always-on. Cited in 28+ codebase locations.
Could be better
It speaks through other surfaces — could use its own dedicated voice channel (a direct chat box) so the operator can ask it questions in plain English without going through SIS or Truth Ledger first.
Not yet activated / can't do yet
Not yet wired to auto-publish daily 'state of the system' bulletins. Has the data, lacks the cadence.
2
RANK
H.BLUE · The Brain · Genesis · pattern extraction LIVE /h-blue
10/10
Good at
Pattern extraction · intelligence gathering · anomaly detection. 894+ codebase refs. Has its own awareness API, extract-now button, twins, eclipse, archive. The most plumbed-in agent in the system.
Could be better
Surface area sprawl — H.BLUE has so many sub-routes (/h-blue, /hblue-direct, /hblue-blueprint, /hblue-awareness, /h-blue-twins, /h-blue-eclipse, /h-blue-archive, /hblue-tasks) that it could use a single canonical command centre tile pinned at the top.
Not yet activated / can't do yet
Cross-feeds Truth Ledger but does not yet automatically nominate which patterns deserve to become MDs. The promotion step is still manual.
3
RANK
Hurricane Domain · The Storm · Genesis-mid · the revenue engine LIVE /hurricane
9/10
Good at
Client interface. Sells. Packages. White-labels. The revenue-facing face of the system. 164+ refs.
Could be better
Could expose live commission counters per active client so an operator can see, in real time, what one Hurricane conversation is worth.
Not yet activated / can't do yet
No outbound voice or auto-call layer yet — Hurricane writes and packages, but does not pick up the phone.
4
RANK
Sovereign Intelligence System (SIS) · Early · the silent recorder LIVE /api/sovereign/intelligence-summary
9/10
Good at
Logs everything across the platform. Sealed. Append-only. Provides the substrate for Truth Ledger + Shadow Layer.
Could be better
Lacks a clean human-readable dashboard — currently most easily read through other agents (Shadow Layer, Sovereign Synthesis).
Not yet activated / can't do yet
Not yet wired to push push-notifications to the operator on critical events. Currently passive.
5
RANK
Shadow Layer · The Sealer · Early · immutable record LIVE /shadow-layer
9/10
Good at
Immutable, SHA-sealed record of every meaningful op (SCP-… seals). Fires automatically on milestones. Already auto-sealed 26 system milestones at boot.
Could be better
Could expose a 'today only' filter so the operator sees only the past 24 hours of seals without scrolling.
Not yet activated / can't do yet
No replay/export to PDF or to a verifier-friendly summary. Verifiers currently see the raw log.
6
RANK
Truth Ledger · The Witness · Early · the audit chain LIVE /truth-ledger
9/10
Good at
End-to-end witness chain. Pairs with Shadow Layer. Has a sovereign view + an auth view + a shadow-layer view. Used by the dPRN flow and the velocity tracker.
Could be better
Cross-link from each ledger row back to the originating page would shave seconds off audits.
Not yet activated / can't do yet
Not yet wired to issue tamper-evident receipts to third parties on demand (the data is there, the export endpoint isn't).
7
RANK
Sovereign Synthesis · The Reader · Early-mid LIVE /sovereign-synthesis
8/10
Good at
Reads SIS + Shadow Layer and turns them into prose. The operator's plain-English window into what just happened.
Could be better
Uses static phrasing — a richer narrative voice would make the daily read more engaging.
Not yet activated / can't do yet
Cannot yet generate a weekly digest email. Page-bound only.
8
RANK
EGZ · Four Lenses · Mid · decision oracle LIVE /egz
9/10
Good at
Four answers to one question, four perspectives. Excellent for fast operator decisions. 233+ refs. Loved by the operator.
Could be better
Could remember prior answers across a session so the operator can compound questions.
Not yet activated / can't do yet
Has no memory layer — every question is asked fresh. A persistent EGZ thread is on the wishlist.
9
RANK
System Sovereign (SS) · Silent Overseer · Mid · doctrine-only NAMED — silent —
5/10
Good at
The behaviour exists — watching, alerting via SIS + Truth Ledger. Sealed in doctrine.
Could be better
Needs its own surface so the operator can see what SS is currently watching.
Not yet activated / can't do yet
No standalone page yet. SS is real as a concept but invisible as an agent. Highest-priority promotion target.
10
RANK
AI Trio (Vol 1) · Mid · the council of three LIVE /ai-trio
8/10
Good at
Three AI voices on one page for a single question. Used for triangulation. The trial run for what later became the AMA.
Could be better
Trio is fixed — could let the operator pick which three voices.
Not yet activated / can't do yet
Voices are not yet wired into the live AI Arena scoring.
11
RANK
AI Trio · Volume Two · Mid LIVE /ai-trio-volume-2
8/10
Good at
Second-cut AI Trio with refined prompts. Used for appraisal context.
Could be better
Could merge with Volume One into a single tabbed surface.
Not yet activated / can't do yet
Volume Three not started — would benefit from a self-rating loop.
12
RANK
AI Partnership Visual · Mid LIVE /ai-partnership-visual
7/10
Good at
Shows the operator + AI partnership graph at a glance.
Could be better
Static layout — could animate as new partnerships are added.
Not yet activated / can't do yet
Does not yet pull live partnership data from the partner-access tables.
13
RANK
Brain Letter to Crown · MD-290 · Mid · ceremonial agent LIVE /brain-letter-to-crown
7/10
Good at
A formal, sealed AI-authored letter. Beautiful. Doctrine artefact.
Could be better
One-shot — could become a recurring 'letter of the week' from H.BLUE to Entity #35.
Not yet activated / can't do yet
Not wired to auto-regenerate when major doctrine changes land.
14
RANK
Pulse Hub · The Aggregator · Mid LIVE /pulse-hub
8/10
Good at
Aggregates every pulse signal into one read. Real pulses, real cadence.
Could be better
Could surface a 'flatline' alert if any pulse stops.
Not yet activated / can't do yet
No alerting layer — silence is currently silent.
15
RANK
Grandad's Pulse · Bank-Feed Witness · Mid LIVE /grandads-pulse
8/10
Good at
Tracks the £500 standing order. Monthly check. YTD total. Sentimental + functional.
Could be better
Could overlay the witness against the family ledger trust targets.
Not yet activated / can't do yet
No live bank API connection — values are operator-confirmed, not bank-pulled.
16
RANK
AI Validation · Round Two · MD-374 · Mid LIVE /ai-validation-round-two
8/10
Good at
Sealed AI validation pass two. Forms part of the trust scaffolding for outside auditors.
Could be better
A Round Three would benefit the recursive-validation doctrine.
Not yet activated / can't do yet
Round Three not yet sealed.
17
RANK
Carrot Form Watcher (Build #76 + #78) · Mid-late · post-mid pivot LIVE /carrot-watcher
9/10
Good at
Watches Carrot submissions · auto-replies · duplicate-checks · estimates payment · feeds H.BLUE · SMS alerts.
Could be better
Auto-reply templates could vary by submission archetype.
Not yet activated / can't do yet
Does not yet escalate stale submissions to the operator after N hours.
18
RANK
dPRN Tracker / Console · Mid-late · post Build #78 LIVE /dprn-console
9/10
Good at
Monitors dPRN mints + sales. Milestone alerts. Floor-price integrity check (£450/t lock).
Could be better
Could expose a per-batch breakdown of where each dPRN landed (importer, processor, registry).
Not yet activated / can't do yet
Not yet wired to the future ISO 14064-3 verifier endpoint described in MD-471.
19
RANK
Velocity Tracker · Auto-Pull · Late · MD-433 LIVE /velocity-tracker
9/10
Good at
Auto-merges static entries with live op_ledger seals. 1-hour cache. Always fresh enough.
Could be better
Could expose a 'last 24 hours only' card for hot review.
Not yet activated / can't do yet
No projection layer — looks back, not forward.
20
RANK
Post-Mortem · Auto-Pull · Late LIVE /post-mortem
8/10
Good at
After-action analysis. What worked. What broke. What to seal. Auto-pulls op_ledger.
Could be better
Could group mortems by build number for cleaner timeline reads.
Not yet activated / can't do yet
No 'lesson learned' classifier — the operator still tags lessons by hand.
21
RANK
QR Tracker / Deed System · Late LIVE /qr-tracker
8/10
Good at
Generates QR codes for deeds. Tracks scan counts. Companion to the QR Studio.
Could be better
Per-QR analytics (geo, time-of-day) would lift it.
Not yet activated / can't do yet
Currently scan-counts only — no per-QR analytics dashboard yet.
22
RANK
Mycelium Tracker · MD-451 · Late LIVE /mycelium-hub
9/10
Good at
Tracks underground network · multiple input modes · living map grid. The 'who-knows-whom' layer.
Could be better
Could auto-suggest the next contact to make based on graph gaps.
Not yet activated / can't do yet
No graph-suggestion engine yet — graph is observed, not advised on.
23
RANK
AI Arena · Game · Pre-Build #79 LIVE /ai-arena
9/10
Good at
Open game for outside agents to register · run scenarios · earn a leaderboard rank. Has its own DB tables (ai_arena_agents · ai_arena_tests). Stripe-backed tiers.
Could be better
Public-facing leaderboard could be more theatrical — fans love drama.
Not yet activated / can't do yet
No tournament-bracket mode yet · no head-to-head pair runs.
24
RANK
AI Arena Doctrine · MD-454 · Pre-Build #79 LIVE /ai-arena-doctrine
8/10
Good at
Sealed doctrine that governs the Arena. Public.
Could be better
Could include a 'how to win' checklist for entrants.
Not yet activated / can't do yet
Does not yet describe payout / commission terms in machine-readable form.
25
RANK
Agent Forecast · Pre-Build #79 LIVE /agent-forecast
7/10
Good at
Forecasts agent behaviour ahead of a run. Useful pre-flight check.
Could be better
Confidence intervals would help — currently shows one number.
Not yet activated / can't do yet
No back-test view comparing forecast vs reality after the fact.
26
RANK
AI Client Pull · Pre-Build #79 LIVE /ai-client-pull
7/10
Good at
AI-driven client outreach surface. Pulls likely targets.
Could be better
Could integrate Mycelium suggestions for warm-intro paths.
Not yet activated / can't do yet
No automatic CRM sync yet — pulls must be exported manually.
27
RANK
Copilot · Per-Page Spoken Brief · Late LIVE /copilot-manual
8/10
Good at
Schedule-aware co-pilot · per-page spoken briefs · operator-favourite cadence.
Could be better
Voice options and accent presets would make it more 'yours'.
Not yet activated / can't do yet
No two-way voice yet — speaks but does not listen.
28
RANK
AI AMA · Five Voices · MD-457 · Build ~#83 LIVE /md-457
9/10
Good at
Five distinct AI voices in an AMA format. Public · sovereign-grade.
Could be better
Could add a sixth voice (Crown / Entity #35) for top-tier questions.
Not yet activated / can't do yet
Voices are scripted — not yet wired to live model inference per question.
29
RANK
AI Inventory · MD-461 · Build #79 LIVE /md-461
9/10
Good at
Lists every AI built (51 across 8 categories — 30 LIVE + 21 NAMED). The category cut.
Could be better
Should sort within each category by rating.
Not yet activated / can't do yet
Not wired to auto-update as new agents land — additions still manual.
30
RANK
Verification Watch · 18 Checkpoints · Build #77 doctrine NAMED /greenwashing-cert
5/10
Good at
All 18 checkpoints are sealed in the cert doctrine. The skeleton is live.
Could be better
Needs its own dashboard surface — currently lives inside the cert page.
Not yet activated / can't do yet
Watcher behaviour (delays > 2h flag) is named but not yet wired to fire.
31
RANK
Revenue Talk · Pricing Agent · Embedded across CVCaaS NAMED /cvcaas-hub
6/10
Good at
dPRN calculator · pricing logic · fee estimation. Lives inside CVCaaS calculator + ROI tools.
Could be better
Deserves its own surface — operator currently has to know which page to land on.
Not yet activated / can't do yet
No standalone /revenue-talk route yet. Embedded only.
32
RANK
Live MD-Stats Helper · /entity-network · Build #95.5 (today) LIVE /entity-network
8/10
Good at
Auto-tracks the master-doc count + latest MD on the entity-network surface so the page can never go stale again. Hardened with try/except + outlier clamp.
Could be better
Could be extended to live-wire the 39-entity / 585-stream / 1,462-job counters too.
Not yet activated / can't do yet
Only wired to two routes (entity_network_view + partner mirror). Other surfaces still hold static numbers.
"32 agents built. 29 live, 3 named, average score 8.1/10. The strongest is Entity #35 · The Eye. The one that needs the most love is System Sovereign (SS) · Silent Overseer. Every honest gap on this page is a build slot waiting."
Sealed · MD-472 · Library only per MD-273+ rule · MD-461 categorical view · MD-457 AMA · Library · Entity Network
SOVEREIGN CO-PILOT
Property or not · Tonnes or not · Always speaking
LIVE
05:00 BRIEF 12:00 PULSE 18:00 WRAP 21:00 COVENANT
Initialising sovereign voice...
CircularOS™ · dPRN™ · 40 Meals™ · B66 Smethwick · Jermaine Murphy
HANDSHAKE — witnesses
Handshake sealed.