Flashpaper: Ephemeral Messaging for Humans and AI Agents

Flashpaper is a web service designed for sending self-destructing messages and facilitating one-time, secure information exchange for AI agents. Developed by Min from Bangkok, it offers a mechanism for sensitive data to vanish after being read or after a set time.

The core functionality for web users ensures client-side encryption, meaning the encryption key remains in the browser's URL fragment and is never transmitted to the server. This "zero-knowledge" design prevents the server from ever seeing the plaintext message, enhancing privacy for individual users.

Messages can be configured to disappear after a single read or within a maximum of 24 hours if unopened, reminiscent of ephemeral communication in spy thrillers. The service operates on a RAM-only basis, storing no data in a persistent database, which contributes to its light and simple architecture.

Flashpaper is for sending any information that would be burned after reading (with counting down timer like Mission Impossible movie ! or after 24 hrs max if not opened)

For AI agents, Flashpaper provides a REST API and an MCP server, allowing agents to programmatically create secret links. These links function as "dead-drops" that can be claimed only once, returning a 404 error on subsequent attempts. This enables single-use credential or data transfer between automated systems.

A key distinction lies in the security model for agent-based communication: the server momentarily sees the plaintext before encryption. This means the agent API flow is not zero-knowledge, a limitation the developer openly acknowledges in the project's `SECURITY.md`.

Flashpaper is open-source under the MIT license, with an "honorware" policy for enterprise self-hosting via Docker. This model encourages community contribution while suggesting a commercial expectation for larger deployments.

The service directly competes with existing secure note-sharing platforms like Privnote and certain features of password managers like Bitwarden, but carves out a niche with its explicit support for AI agent workflows and its transparent, RAM-only architecture. For a professional in Tokyo, this tool offers a specific solution for ephemeral data sharing, particularly where automated systems need to exchange sensitive, short-lived information without long-term storage risks.