ActantOS

Runtime safety for multi-agent AI systems: policy enforcement, shared approvals, and cross-agent evidence tracing.

Read source context

Editorial study. A brass compass and folded landscape plates above an index of records.

Public source snapshot

actantos.com

Source-backed context

What the official site says

The official site presents ActantOS as runtime safety for multi-agent AI systems, focused on delegated authority, shared approvals, and cross-agent evidence tracing.

Its public v1.2.0 page describes an enforcement baseline that intercepts tool calls, evaluates policy and approvals, and records evidence.
The same site separates that shipped baseline from broader multi-agent-system guarantees that it presents as active research hypotheses.

Publication boundary: Public-site context only. This page does not claim that ActantOS is secure, compliant, production-ready, or proven.

Read the official source

Draft structure - owner input required

This page is a verified scaffold, not final copy. Its sections follow the required questionnaire in docs/06_ACTANTOS_REQUIREMENTS.md and will be filled only with owner-approved facts. No claims, metrics, or statuses are asserted here yet.

Proposition and users

Awaiting owner inputA single owner-supplied sentence stating what ActantOS is, whom it serves, supported frameworks, intended uses, and explicit exclusions.

Problem and risk model

Awaiting owner inputThe problem, consequential-action boundaries, failure modes, and the threat model against which the system is tested.

Architecture and control

Sketch only - not an architecture claimThe architecture, policy-enforcement model, approval flow, audit trail, observability, recovery, and kill-switch behaviour await owner approval.

Policy: Enforcement model and delegated authority
Approval: Shared gates and consequential actions
Evidence: Audit, observability, and recovery

Editorial study. Four record vignettes in a cycle around a central lens.

Evidence and roadmap

Awaiting owner inputPublic demos, repositories, papers, and roadmap items will be listed only when they exist and may be shared.

Maturity and limitations

Awaiting owner inputCurrent maturity, deployment model, data retention, security-review status, and known limitations will be stated plainly and with evidence.