Agent catalogue
Preview-only catalogue with honest live, build, and roadmap posture. Tenant execution loads after sign-in.
45 of 45 entries
Primary control-plane experience for pilots and demos; uses cookie-based JWT to `/api/v1/*` (see `apps/web/middleware.ts`).
Tenant-scoped operator UI: interactions list, command & execution views, overview metrics, connectors, approvals, interaction detail, and this Agent Overview map.
Agent profile →Primary operator workspace for governed marketing operations; MERCURY domain agent manifests in integration topology.
Commercial Marketing module UI: tenant-scoped campaigns, content items, strategy workspace, MERCURY intelligence — backed by Prisma (`MarketingCampaign`, `MarketingProfile`, etc.) and `/api/v1/marketing/*` routes.
Primary operator workspace for governed sales and pipeline intelligence; AUCTOR domain agent manifests in integration topology.
Commercial Sales module UI: accounts, opportunities, outbound sales campaigns — backed by Prisma (`SalesAccount`, `SalesOpportunity`, `SalesCampaign`, …) and sales API routes.
Single pane for demand-to-revenue posture when one or both commercial modules are active.
Combined MERCURY + AUCTOR growth operating surface: ATLAS strategic synthesis, TERMINUS maturity, MERCURY↔AUCTOR handoff events — `GET /api/v1/growth/command-center`.
Single controlled ingress for the TypeScript kernel; aligns with API-first posture.
Fastify `/api/v1` surface with Zod validation, tenant resolution, and RBAC — fronted by Next middleware for JWT.
Agent profile →Gateway for POST /interactions: validates payload, attaches trace context, and hands off to ATLAS + policy.
Interaction entry, validation, and framing for governed requests.
Agent profile →Classifies operation (e.g. MONETA_ANOMALY_SCAN), coordinates policy evaluation inputs, persists interaction status.
Orchestration: classify Golden Path operation and persist pre-execution state.
Agent profile →`ApprovalRequest` rows with PENDING/APPROVED/REJECTED; bridges console Approvals page and execution pauses.
Human-in-the-loop decisions before sensitive execution or dispatch.
Agent profile →Produces ALLOW / DENY / REQUIRE_APPROVAL / … with traceable rule ids.
Deterministic policy evaluation for tenant operations.
Agent profile →Orchestrates parallel, auditable retrieval (ephemeral seekers) and evidence pack assembly before deterministic MONETA execution.
Evidence agent with Swarm Intelligence: spawns short-lived seekers to retrieve governed data from inside the tenant boundary and from approved external surfaces, then materializes evidence for downstream execution.
Converts pose frames into stride, cadence, symmetry, confidence, and calibration-aware diagnostics without placeholder scoring shortcuts.
Deterministic 3D movement analysis engine for RTW: ingests pose landmarks, computes gait metrics, and emits governed evidence used by readiness and policy flows.
Produces structured scan outcomes and recommendations for operator review.
Domain agent: deterministic finance anomaly scan on governed evidence.
Surfaces risk level, summary, and recommended action prior to NUNTIUS.
Pipeline risk assessment after MONETA — assessed posture before communications drafting (execution `ExecutionJobStage.JANUS` namespace).
Agent profile →Prepares channel / audience / message drafts; may require explicit approval for dispatch.
Drafts governed outbound communications; never sends without approval path.
Agent profile →Governance read — not anomaly detection (per operator console).
Control posture: exposure / containment across prior stages and approval state.
Agent profile →Short remediation steps / plan status — not a full task system.
Action plan: governs what happens next from prior stages + approvals.
Agent profile →Materializes `OutboundDispatch` rows and obeys tenant + env gates for live send.
Outbound dispatch stage: dry-run default, live send multiply gated.
Agent profile →Persists operator feedback linked to interactions / workflows per policy.
Structured human correction capture on meaningful outputs.
Agent profile →`AuditEvent` rows with actor + tenant attribution.
Traceability plane for meaningful actions (and selected reads per policy).
Agent profile →Platform-scoped sanitized knowledge review lane complementing tenant-local ATLAS artifacts.
Reviewed cross-tenant knowledge candidates parked by SUPER_ADMIN; approve parks only and never auto-pushes to tenants.
Structured knowledge scoped to `tenantId` for retrieval, reuse, and governed deprecation without Global ATLAS promotion.
Tenant-scoped knowledge artifacts for ORACLE retrieval and continuous learning — distinct from MNEMOSYNE memory and ATLAS orchestration.
Surfaces execution-substrate posture without substituting architecture validation (catalog-terminus-validator).
ExecutionJob substrate observer: queues, leases, approvals adjacency, and messaging fabric counters for the tenant runtime.
Agent profile →Observability / dependency clarity for operators across declared and live graphs.
Declared platform architecture + tenant operational topology observer with models/learning command surface inventory.
Agent profile →Standardize tool / capability access with governance hooks.
Typed registry of capability contracts with tenant-scoped upsert and platform catalog reads.
Agent profile →Observing governance lane for tenant-local learning candidates and promotion evidence packets.
Tenant learning governance posture plus explicit promotion packets; approve/revoke never auto-promotes cross-tenant.
Operational metrics for governed capability invocations and agent posture.
Tenant-scoped capability invocation telemetry and computed agent SLO summaries.
Agent profile →Implements the kernel rule: workflow state is durable and inspectable.
Deterministic execution state persisted in PostgreSQL — not prompt-held state.
Agent profile →Golden Path narrative step after workflow — surfaces trace, evidence, MONETA, stages.
Structured response payload to operators / API clients after governed execution.
Agent profile →Makes the tenant visible on 2D/3D maps next to commercial modules (MERCURY, AUCTOR) and Growth Command.
Logical boundary for `tenantId`-scoped data and APIs: resolved from JWT/membership at ingress and enforced on every query — topology links this node to Marketing OS, Sales OS, and Growth Command as module surfaces that run inside a tenant.
Durable substrate for deterministic workflows — no prompt-held state for meaningful platform facts.
Canonical storage for tenant data, workflow runs, evidence, jobs, approvals, audit, and outbound dispatch.
Agent profile →Operational glue between kernel stages and external systems — governed and auditable.
Async worker, PostgreSQL connector, outbound adapters (Resend / log-only), and command/read models.
Agent profile →Read model / analytics plane — not an execution stage and not LLM-generated.
Cross-interaction strategic signal analysis: scans recent completed MONETA workflow outputs in-window and emits typed STRATA signals (severity, recommendation, supporting interaction ids).
Boardroom-facing synthesis — templates only; produces priority, headline, recommended actions, and supporting interaction ids.
Leadership intelligence layer: merges STRATA signals with command-layer posture (approvals, execution jobs, dispatch, risk summaries) into a deterministic executive brief.
Bridges executive briefings into trackable operator work (acknowledge, close, reopen, assign) without altering Golden Path execution semantics.
Lightweight governed artifacts for leadership follow-through: queue items with owner, due date, escalation posture, and lifecycle audits — created or deduped from HELIOS when priority warrants review.
Read-only governance signal for operators and boards; complements roadmap vision nodes.
Deterministic validation of declared Agent Overview topology and catalog against Prisma `ExecutionJobStage` and leadership-chain rules — explains drift; does not auto-fix.
Readiness validation, governance gating, transition control, and oversight across orchestration → leadership → execution. Retains constitutional AI/ISMS themes, cites allowlisted public sources, surfaces operator & Sentinel insights, and refreshes the corpus weekly. Supports (does not replace) human certification workflows.
AI governance and compliance intelligence for readiness evidence, control assurance, and governed operator guidance.
Surfaces contradictory or weakly supported leadership paths through a tenant-scoped assurance snapshot for operator review.
Decision assurance and validation intelligence: deterministic, read-only checks over STRATA, HELIOS, Strategic Review, and JANUS governance posture — observing only; no execution or policy override.
Separates tenant-local context authority from broader memory scope; supports trust-aware integrity narratives alongside learning vision and governance artifacts.
Context and memory governance intelligence: tenant-bound MNEMOSYNE memory records with provenance, retention, deprecation, and deletion receipts — not per-interaction retrieval (ORACLE) and never cross-tenant.
Surfaces handoff, SLA, and bottleneck posture for operator review without substituting ATLAS orchestration.
System coordination and flow intelligence: tenant-scoped, read-only flow-health observations over execution jobs, incidents, and remediation signals.
Growth intelligence adjacent to read models and outbound paths; message send remains NUNTIUS; orchestration remains ATLAS.
Marketing and growth intelligence: tenant-scoped MERCURY domain agent — pairs with the Marketing OS module (Prisma `Marketing*` models, `/console/marketing`, marketing APIs) for campaigns, content, and signals; topology node remains the integration anchor.
Separates customer success intelligence from executive synthesis (HELIOS) and from domain execution (MONETA) while staying tenant-scoped.
Customer success intelligence: tenant-scoped customer-health narratives composed from CRM, messaging, finance, and incident signals — observing with explicit missing-source posture.
Commercial pipeline posture alongside execution outputs; not a full external CRM — governed APIs and connectors (e.g. HubSpot) as configured.
Sales and pipeline intelligence: tenant-scoped AUCTOR domain agent — pairs with the Sales OS module (Prisma `Sales*` models, `/console/sales`, sales APIs) for accounts, opportunities, and outbound sales campaigns.
Owns the support-operations narrative — operator/service adjacency without sending outbound comms.
Support intelligence: tenant-scoped triage read models from messaging, incidents, and connector-job evidence — distinct from NUNTIUS send authority.
Bridges signal layer toward opportunity narratives for leadership follow-through; complements AEGIS assurance without duplicating it.
Innovation and strategic opportunity intelligence: evidence-backed opportunity hypotheses from STRATA, HELIOS, finance, and Strategic Review signals.
Expand governed verticals without forking the OS kernel — observes entitlement posture only.
Commercial module SKU + plan pack inventory observer for Sales, Marketing, Finance, and GaitIQ verticals.
Agent profile →
