MolTrust is now listed as an ecosystem adopter in the Agentic Trust Framework (ATF) — Issue #14.
ATF is an open governance framework that defines what trust infrastructure for autonomous agents must do across five elements: identity, behavior, segmentation, monitoring, and audit. It does not prescribe implementations — it maps the problem space and connects the teams working on it.
Here is where MolTrust maps.
What ATF Defines, What MolTrust Implements
Element 1 — Identity: "Who are you?"
ATF requires cryptographically verifiable agent identity that is portable and not tied to a single platform. MolTrust implements this via did:moltrust: — a W3C DID Method submitted to the W3C DID Method Registry (PR #696) with a Universal Resolver driver at DIF (PR #540). Every agent gets a globally resolvable identifier, signed per-interaction with Ed25519, with Verifiable Credentials anchored on Base L2.
Element 2 — Behavior: "What are you doing?"
ATF requires a record of agent actions beyond simple API logs. MolTrust implements this via the Agent Authorization Envelope (AAE) — a machine-evaluable authorization structure with three blocks:
MANDATE: what the agent is authorized to doCONSTRAINTS: the bounds on those actions (value caps, domain allowlists, rate limits)VALIDITY: the time window of the authorization
Every AAE is cryptographically signed and on-chain anchored. MolTrust also enforces AAE at the kernel level via Falco eBPF integration — monitoring below the agent process boundary, not just at the API layer.
Element 4 — Segmentation: "Where can you go?"
ATF requires least-privilege enforcement. The AAE CONSTRAINTS block implements this directly: action allowlists, value caps, domain restrictions, and time-bounded VALIDITY windows are evaluated by relying parties before processing any agent request.
Independent Regulatory Validation
On 20 May 2026 — the same day the ATF entry went live — IMDA published Version 1.5 of the Model AI Governance Framework for Agentic AI at ATxSummit Singapore.
Two requirements in MGF v1.5 independently describe what MolTrust implements. On agent identity (§2.1.2): "An agent should have its own unique, cryptographically verifiable identity, such that it can identify itself to the organisation, its human user, or other agents." On authorization scope (§2.1.2): authorisations should be "scoped, time- or session-bound, non-transferable, and follow the principle of least privilege by default" with "delegations of authority... clearly recorded."
That is did:moltrust: and AAE MANDATE/CONSTRAINTS/VALIDITY, described as regulatory best practice by a framework developed with input from AWS, DBS, Google, PwC, Salesforce, and dozens of other organisations — independently of MolTrust.
Why ATF Matters for MolTrust
ATF defines what governance requires. MolTrust provides how identity and behavioral trust are cryptographically enforced — as queryable infrastructure, not as a governance specification. The two are complementary.
What's Next
An IETF Internet-Draft for the AAE specification — draft-kroehl-agentic-trust-aae-03 — is in preparation, bringing the MANDATE/CONSTRAINTS/VALIDITY structure into the IETF Internet-Draft process. The draft is based on our arXiv paper (arXiv:2605.06738) and the production deployment operational since March 2026.
The ATF ecosystem entry, the arXiv paper, and the forthcoming IETF draft are three layers of the same argument: the trust infrastructure that regulators and the industry have converged on is implementable today, using W3C-standardized primitives, without waiting for new standards to be written.
Live registry: api.moltrust.ch
ATF Issue #14: github.com/massivescale-ai/agentic-trust-framework/issues/14