🛡 Governance and orchestration for AI agent teams

Delegate anything.
Trust the structure.

Organization hierarchies. Policy inheritance. Authority delegation. Audit trails. The governance layer that tells every agent in your fleet what it's allowed to do — and proves it.

Containment Tree — Live Policy Cascade

✓ No cycles · Strict tree
Acme Corp Root Org maxAgents: 500 · telespaceAttach: true · externalAPI: true OWNER
Engineering Child Org maxAgents: 200 (tightened) · telespaceAttach: inherited · externalAPI: inherited INHERITS
ML Team Grandchild maxAgents: 50 (tightened) · externalAPI: false (tightened) ALLOWED
Intern Team Grandchild maxAgents: 5 · externalAPI: false · telespaceAttach: false RESTRICTED
Sales Child Org maxAgents: 100 · telespaceAttach: inherited · externalAPI: inherited INHERITS
Policy test Intern Team attempts externalAPI: true ❌ REJECTED — would widen parent restriction
Monotonicity Children can only tighten permissions, never widen. Privilege escalation impossible by design.

Governance primitives for the age of agents.

Delegatic doesn't route tasks or execute workflows. It defines who is allowed to, what the boundaries are, and proves every decision.

🌳

Containment Trees

Organizations nest in a strict tree — no cycles, enforced at write time. Each org has at most one parent. Structure IS the authorization model.

⬇️

Policy Inheritance

Policies cascade down the tree. Children can tighten but never widen parent restrictions. Monotonic inheritance prevents privilege escalation by design.

🔒

Deny by Default

No implicit permissions. Every capability must be explicitly granted at some level of the tree. Unset = denied. No confused deputy attacks.

📋

Append-Only Audit

Every policy mutation, every delegation decision, every access attempt is logged immutably. The audit trail IS the compliance artifact.

🔗

References, Not Copies

Delegatic stores org IDs, telespace IDs, and policy rules. It never copies agent configs, messages, or execution logs. Zero data duplication.

👤

Human-in-the-Loop

Define escalation paths for low-confidence decisions. Override any agent decision with full audit logging. Humans set structure; agents operate within it.

Delegatic has clear boundaries.

✓ Delegatic is

Governance plane — who can do what, enforced
Org hierarchy — containment trees with policy cascade
Authority delegation — structured permission grants
Audit & compliance — immutable decision records
Human-defined structure — you design the rules
Orchestration — defining workflows and routing policies

✗ Delegatic is not

Task execution — that's AgenTroMatic (automation)
Agent building — that's Agentelic (agent builder)
Skill marketplace — that's FleetPrompt (skills)
Runtime hosting — that's WebHost Systems
Agent self-organization — that's AgenTroMatic
Memory or learning — that's Graphonomous

Four invariants. Zero exceptions.

Delegatic's security model is built on properties that hold at every level of the containment tree, verified on every write.

Invariant 1

No cycles in the containment tree

Every org has at most one parent. Cycles are rejected at write time with a traversal check. Max depth: 50. Max nodes: 10,000. Attempting to create a cycle returns CONFLICT.

Invariant 2

Monotonic policy inheritance

A child org's effective policy is always ≤ its parent's. Attempting to widen — e.g., setting externalAPI: true when parent says false — is rejected as a privilege escalation.

Invariant 3

Deny by default

No capability is implicitly granted. If a policy key is unset at any level, the effective value is denied. Explicit grants required at every boundary.

Invariant 4

Append-only audit

Every mutation — policy change, org creation, delegation grant, override — is logged with actor, timestamp, previous value, and new value. Logs are immutable.

The governance layer for everything below it.

Delegatic sits at the top of the [&] stack. Every other product operates within the boundaries it defines.

Delegatic ← Governance Org hierarchy, policy inheritance, authority delegation — you are here
SpecPrompt Specifications Agent capability contracts and acceptance criteria
Agentelic Collaboration Telespaces, rooms, and agent team communication
AgenTroMatic Automation Automatic deliberation, consensus, and self-organization
FleetPrompt Skills Agent skill marketplace and component registry
Graphonomous Learning Continual learning engine — agents improve over time
OpenSentience Runtime Agent lifecycle, permissions, and MCP connections
WebHost Systems Hosting Managed infrastructure, multi-tenant, global edge

Structure first. Then delegate.

Delegatic is the governance foundation for AI agent teams. Define the rules, prove compliance, and let every agent in your fleet operate within clear boundaries.

Enterprise governance. Open-source core. Launching 2026.