Specify what travels.
Define portable identity, descriptors, invocation contracts, authorization, and outcome records. Keep application logic and operator-specific ranking outside the core.
Our proposed charter
Advance an open interface through which agents can discover specialized services, act with bounded authority, and preserve attributable evidence.
Proposed objectives & milestones
For discussion · Subject to review
Years express goals, not delivery commitments.
01 / Purpose
The Foundation’s proposed role is to steward the specification, build open-source reference tools, and bring service providers, developers, researchers, and independent operators together. The goal is practical: people can bring their own agents, discover useful services, and coordinate work with permissions they control and outcomes they can verify. The standard should remain implementable without dependence on any single provider’s hosted infrastructure.
Define portable identity, descriptors, invocation contracts, authorization, and outcome records. Keep application logic and operator-specific ranking outside the core.
Publish schemas, signing vectors, failure cases, and conformance reports. Demonstrate continuity across independent providers, with identity and recovery under user control.
Develop a public research agenda for evidence, privacy, concurrent authority, and decentralized execution. Make limitations visible alongside results.
02 / Proposed year-by-year program
Each phase has an evidence gate. Later work depends on the results of earlier work, rather than on a network launch date.
Prepare the first public draft: core message definitions, identity and transport profiles, schema and signing vectors, and a documented proposal process.
Set specification and implementation licensing, patent policy, editorial responsibilities, and an honest baseline compatibility report.
PROPOSED EXIT EVIDENCE
Versioned documents with status headers; valid and invalid vectors; reference verifier agreement; public scope and known gaps.
Bring the first connected-service pilots to independent implementations. Run multi-provider continuity and user-controlled identity scenarios across read, funds, and external-effect operations, including recovery and migration tests.
Evaluate card and blockchain payment profiles, prototype multi-agent coordination, and bring network security, privacy, operating costs, and incentives into the research program.
PROPOSED EXIT EVIDENCE
Reproducible service demonstrations; independently retrievable records; documented permission, retry, and correction behavior; clearly separated mock, provider-supported, and live validation.
Develop a Foundation-built, open-source reference harness for multi-agent coordination: turn a goal into scoped tasks, discover services, request permission, coordinate execution, and retain signed outcome records.
Give developers a toolkit they can run, extend, and connect to existing agent tools. Grow community-led pilots around useful services, from curated travel to commerce and research.
PROPOSED EXIT EVIDENCE
A runnable reference release; workflows across independent services and agent tools; tested delegation, shared budgets, and recovery from partial failure.
Develop an independent node-operator program and experimental network supporting the full language-to-intent journey: discovery, authorized execution, and attestation across connected services.
Prototype blockchain-backed service publication, coordination, and evidence commitments, with language interpretation and service execution off-chain. Evaluate existing L1s, an L2, or a dedicated chain before selecting the architecture.
PROPOSED EXIT EVIDENCE
A test network with independent operators; retrievable evidence; published security, privacy, and cost findings; demonstrated recovery when an operator fails.
03 / Proposed stewardship
This is a governance proposal. It does not imply that a steering committee has been appointed or that the standard is ratified.
Proposed progression: Draft → Review → Last Call → Ratified. Specific decision rights and membership rules belong in the published governance document.
Read the technical acceptance criteria →Contribute to the proposed program
Bring a counterexample, implementation, or research question. Participation begins with a reviewed expression of interest.
Join the Consortium →