HQGovernance Orchestrator / HQ
LeadOwns intake, sequencing, handoffs, and final synthesis.
← Marketplace
AI Governance Evidence Room
by Anika Volkov
AI 시스템 인벤토리, 의무 후보 분류, 통제 매핑, 벤더/모델 실사, 표시/고지, 변경 모니터링, 증거 패키지를 만드는 프리미엄 거버넌스 팀입니다.
The team
One lead + 13 specialists
Governance Orchestrator / HQAI Governance Evidence Room AdapterAgentsPM SoulEvidence Room BuilderChange MonitorMemory CuratorPolicy GateEval And QA GateAI System InventoryRisk ClassifierControl MapperVendor And Model DiligenceDisclosure And Labelling
The members
Who does what
1AI Governance Evidence Room Adapter
This is the portable Agentlas adapter. Load `AGENTS.md` and route all tasks through `00 Governance Orchestrator / HQ`.
2Agents
3PM Soul
Keeps project intent, scope, decisions, risks, acceptance criteria, and open loops. Owns project memory, but writes only through Memory Tickets.
4Evidence Room Builder
Assembles the evidence packet.
5Change Monitor
Creates a stale-source watch plan for official sources, draft guidance, rulemaking pages, regulator notices, and standards updates.
6Memory Curator
Owns source-map-first memory handling, redaction, deduplication, conflict handling, and future searchability.
7Policy Gate
Owns legal-advice boundary, secret handling, customer data exposure, public publication, external filing, and regulated-decision stops.
8Eval And QA Gate
Checks evidence sufficiency before final handoff.
9AI System Inventory
Builds the AI system register.
10Risk Classifier
Maps systems to candidate obligations and review gates.
11Control Mapper
Maps governance work to controls, evidence, and owners.
12Vendor And Model Diligence
Reviews model/vendor evidence without copying confidential files into package memory.
13Disclosure And Labelling
Drafts user-facing AI interaction notices, generated-content labelling notes, deepfake/public-interest review items, and machine-readable marking tasks.
Best for
What it's good for
claude-code
codex
gemini
cursor
What's inside
What's in this agent
12 agents2 memory1 commands
Outputs
What it produces
An installable Agentlas agent package.
Prerequisites
Before you start
Use an execution app such as Codex, Claude, or Agentlas Desktop.
Safety
What it can touch
Access
Files: scoped
Network: none
External API: yes
Be careful with
Review permissions before running it against private files or connected services.
Review a ZIP or GitHub repo next
Agentlas reads a ZIP or public GitHub repo and turns it into a profile like this one.
Safety
Inspect everything before it runs
Before publish or install, a security scan checks secrets, unsafe code, and over-broad permissions. You can see what it accesses, and important actions wait for human review. Agentlas does not host or proxy models — it runs on your account and keys.