Meet Me

Meet-Me is a notes-stage project that explores how an AI agent can coordinate a meeting with another person's agent without human involvement. The principal publishes a hub-and-spoke set of Markdown skill manifests (agents.md, with sibling skill files for meeting coordination, press inquiries, introduction requests, and support routing) at a stable URL, alongside a structured JSON twin. Email signatures, business cards, and profile pages link to the manifest, and a receiving agent parses the relevant skill to identify availability, booking links, and working-hour constraints before executing a booking and reporting completion to its principal. The repository includes an HTML email-signature wireframe, a worked example manifest, and diagram assets, and is released under the MIT license.

The design is deliberately public and unauthenticated, and the README documents its threat model directly. Identified risks include reconnaissance leakage of working hours and travel windows, absence of requester authentication, and lack of a signed audit trail for disputed bookings. Proposed mitigations include WAF and bot management on the manifest endpoint, reduced-detail variants for low-trust requesters, and a human-in-the-loop confirmation step that holds booking writes as pending until the principal approves them. Stronger guarantees such as verified identity and signed transcripts are reserved for the companion project.

The project may be of interest to developers and researchers working on agent interoperability, calendar automation, or machine-readable contact and scheduling conventions, as well as those evaluating how AI agents might discover and act on published user capabilities. It is presented as a sketch rather than a formal specification, and the author invites prior-art references.