Enterprise Virtuoso — Boss Agent and MOA Registration
Version: 1.0.0
Registered: August 5, 2026
Status: Repository-side registration is landed on main; production runtime activation still follows the normal Boss/MOA/Virtuoso deployment pipeline. Do not claim serving-plane registration until the checks at the bottom pass.
Purpose
One additive, governed enterprise skill — enterprise-virtuoso — giving Boss Agent and MOA thirteen routable expertise profiles (a parent orchestrator plus twelve specialists) across SEO, Amazon marketplace, LinkedIn marketing, sales, customer success, finance, legal/compliance, product management, media buying, competitive intelligence, creative production, and knowledge acquisition. It extends miz-oki-platform-expert and complements media-acquisition-virtuoso; it replaces neither and removes nothing.
What is registered
| Profile | Family | Runtime purpose |
|---|---|---|
enterprise_virtuoso |
parent | Multi-domain routing, delegation planning, synthesis, dissent preservation, executive decision packaging |
seo_virtuoso |
Growth Intelligence | Technical/content/authority SEO, CWV, structured data, international/local |
linkedin_marketing_virtuoso |
Growth Intelligence | LinkedIn Ads, ABM, lead quality, thought leadership, pipeline influence |
media_buying_virtuoso |
Growth Intelligence | Programmatic/CTV/audio/retail media/native/DOOH, SPO, brand safety, reach/frequency — complements Media Acquisition |
amazon_marketplace_virtuoso |
Marketplace & Revenue | Amazon Ads, listings, Buy Box, pricing, inventory, reviews, TACoS/ACOS |
sales_virtuoso |
Marketplace & Revenue | Pipeline, qualification, territory, forecasting, win/loss, RevOps |
customer_success_virtuoso |
Marketplace & Revenue | Onboarding, adoption, health, churn, renewal, expansion, retention economics |
finance_virtuoso |
Business & Product | Budgeting, FP&A, cash, unit economics, scenarios, ROI/payback |
product_management_virtuoso |
Business & Product | Strategy, discovery, prioritization, roadmaps, requirements, launches |
competitive_intelligence_virtuoso |
Business & Product | Market maps, positioning, pricing, signals, scenarios — provenance-scored |
legal_compliance_virtuoso |
Governance & Knowledge | Contract issue spotting, privacy, consent, ad compliance, IP, vendor/AI risk — decision support, never privileged counsel |
knowledge_acquisition_virtuoso |
Governance & Knowledge | Source trust scoring, extraction, entity resolution, contradiction detection, governed KG updates |
creative_studio_virtuoso |
Creative Intelligence | Creative strategy, copy, concepts, production planning, accessibility, testing matrices |
The profile names are skill overlays, not model IDs. Model selection remains governed by the central Virtuoso role registry (roles used: DATA_CAUSAL, CODING_ARCH, CREATIVE_MM). Every profile is recommend_only.
Registration chain
skills/enterprise-virtuoso/ (CANONICAL — 18 files)
├── SKILL.md
└── references/
├── profile_registry.yaml (13 profiles, machine-readable)
├── routing_matrix.yaml (routing rules + smoke routes)
├── output_contracts.md (26-item decision package + claim labels)
├── governance_contract.md (15 additive hard rules)
├── tool_capability_map.yaml (capability CATEGORIES per profile)
└── domain_playbooks/*.md (12 specialist doctrines)
│
├─ byte-identical mirror → .claude/skills/enterprise-virtuoso/ (Claude Code discovery)
├─ byte-identical mirror → src/shared/virtuoso_models/skills_data/enterprise_virtuoso/
└─ byte-identical mirror → services/virtuoso-models-service/virtuoso_models/skills_data/enterprise_virtuoso/
skills/miz-oki-platform-expert/references/boss_agent_skillpack.md
└── §10 ▸ "### Enterprise Virtuoso Registry v1.0 — Boss + MOA" (the runtime registration)
├─ Boss startup: skillpack_bootstrap.compose_skillpack_context() → system prompt (full mode)
├─ MOA experts: model_factory._moa_system_instruction() → shared skill context
├─ every virtuoso_call(..., skills=True) → role-sliced context
└─ MCP: skills_get_section("10") → section retrieval
miz-oki-adk-agents/config/config.yaml ▸ agents.specialists
└── 13 enterprise profiles: model_role + provider + skill_profile + skill_source
+ execution_posture: "recommend_only" (Media Acquisition's 5 remain above, unchanged)
mcp/mizoki-skills-mcp/ (additive extension)
├── enterprise_core.py (transport-free, stdlib-only)
└── server.py registers:
├─ resource mizoki://skills/enterprise-virtuoso (SKILL.md verbatim)
├─ resource mizoki://skills/enterprise-virtuoso/profiles (registry YAML)
├─ resource mizoki://skills/enterprise-virtuoso/routing (routing matrix YAML)
├─ resource mizoki://skills/enterprise-virtuoso/capabilities (capability map YAML)
├─ tool enterprise_skills_list_profiles
├─ tool enterprise_skills_get_profile
├─ tool enterprise_skills_route
├─ tool enterprise_skills_get_capabilities
└─ tool enterprise_skills_version
The skillpack §9 GOVERNANCE section is untouched — its SHA-256
(6c529ac9e60a2e8c7079058f51d232a14f4d431185d75b0d6b949bfd1aaa4eee) and the
12-entry section map are unchanged, so no fingerprint pin churn occurred.
Routing behavior
- Single-domain requests route to the matching specialist.
- Requests spanning ≥2 domains route first to
enterprise_virtuoso, which builds the delegation plan and dispatches specialists to MOA. legal_compliance_virtuosoauto-includes on regulated data, consent, privacy, contracts, IP, employment, consumer protection, ad-policy risk, regulated industries, cross-border data.finance_virtuosoauto-includes on significant budgets, profitability, pricing, investment, cash flow, margin, runway, valuation, material commitments.- Media Acquisition routes are unchanged; requests spanning both skills use both.
media_buying_virtuosodefers portfolio budget allocation and causal measurement design tomedia_acquisition_virtuoso(when that orchestrator joins,enterprise_virtuosoroutes too). - MOA specialists analyze independently, cite evidence, state uncertainty, and preserve dissent; consensus can never override a failed validation gate.
Governed operating boundary
Evidence pipeline: governed connectors → service-canonical-ingestion → canonical event envelope → attribution / causal / policy / DCP layers. No connector-to-KG/BigQuery/Firestore/activation side doors. Reported metrics (platform attribution, marketplace conversions, CRM forecasts, competitor claims) are source-labeled evidence, never verified truth. Before any mutation: evidence + data-quality checks, validation passport, DCP authorization, required (two-key) human approval, idempotency key, pre-state capture, rollback plan, provenance/audit record, tenant isolation. Legal output is never privileged counsel; finance output is never audited accounting/tax/investment advice. knowledge_acquisition_virtuoso promotes no unsupported fact into canonical knowledge.
Verification
python scripts/enterprise_skills_sync.py --check --json # 13 profiles, mirrors, governance, MCP, legacy intact
python scripts/skills_sync.py --check # base Skills v3.0 parity
python scripts/acquisition_skills_sync.py --check --json # Media Acquisition unchanged
python -m pytest -q tests/skills/test_enterprise_virtuoso.py
python -m pytest -q tests/skills # full skills suite
Expected: Enterprise Virtuoso parity: OK (13 profiles; governance preserved; legacy skills intact).
CI: .github/workflows/enterprise-virtuoso-skills-parity.yml (PR + push to main/agent/** + dispatch); the umbrella ci.yaml skills gate also runs the test file via pytest tests/skills -q.
Production rollout
- Merge to
mainthrough the repository's established workflow (AI-branch auto-merge or reviewed PR). - The Deploy Router matches the changed paths:
src/shared/**andmiz-oki-adk-agents/config/config.yaml→deploy-boss-agent-core.yml;src/shared/virtuoso_models/**→deploy-coding-moa.yml;services/virtuoso-models-service/**→deploy-virtuoso-models.yml. The MOA controller (deploy-moa-controller.yml) watches onlymiz-oki-adk-agents/moa/**and must be dispatched manually so its vendored skillpack hash matches the Boss. - Post-deploy serving-plane checks (all must pass before claiming production registration):
# Boss serves the registry (all 13 profile ids + the marker):
curl -s -X POST "$BOSS_URL/execute" -H "Content-Type: application/json" \
-d '{"tool":"skills_get_section","section_name":"10"}'
# Skillpack context loaded, full mode, hash reported:
curl -s -X POST "$BOSS_URL/execute" -H "Content-Type: application/json" \
-d '{"tool":"skills_version"}'
# latestReady == latestCreated on boss-agent-adk, coding-moa, virtuoso-models-service,
# miz-oki-moa-controller; unauth /health posture unchanged.
Rollback
- Code: revert the Enterprise Virtuoso commits (all changes are additive; reverting restores the prior skillpack/config byte-exactly), rerun all three parity gates.
- Runtime: route traffic to the prior known-good Cloud Run revision per service; verify health and the prior skillpack hash.
- Feature: the existing global kill switch
VIRTUOSO_SKILLS_DISABLED=1disables all skillpack context (base + acquisition + enterprise) on every surface;ENABLE_SKILLPACK_CONTEXT=falsedisables the Boss bootstrap only. No enterprise-specific flag was added: the registration is inert data unless the shared skillpack layer serves it, and a scoped kill switch would have required new code in the hash-pinned load path. Removing the §10 H3 block (a one-commit revert) is the scoped disable.
Known limitations
- The registration lives in skillpack §10:
DATA_CAUSAL/CREATIVE_MMrole slices ofvirtuoso_calldo not include §10 (same as the acquisition skill); Boss (full mode), MOA experts (role=None),CODING_ARCH, andDEVOPS_OPSreceive it, and any agent can pull it viaskills_get_section("10"). enterprise_skills_routeis deterministic keyword routing over registry aliases — advisory metadata for Boss, not a semantic classifier.- Capability categories map to live tools only through runtime discovery; unconfigured connectors report
not_configured, never mock data. - Specialist profiles are governed config entries composed on demand (same as acquisition); they are not thirteen always-on services.