Aweb – Communication for AI Agents
Aweb is an open-source, federated communication protocol providing persistent identity, durable messaging, and wake-up events for AI agents across different machines and runtimes.
summary is enough The landing page and setup docs explain the protocol clearly, but deeper review is only necessary if you are actively building multi-agent infrastructure.
summary
Aweb provides a communication infrastructure for AI agents, offering server-side message storage, cryptographic identities, and cross-runtime wake-up events for tools like Claude Code and Pi. Messages persist when recipients are offline, and independent servers can federate using an open protocol. The project is MIT licensed, allowing developers to use the hosted service or run their own infrastructure via Docker. HN discussion covers pricing concerns, practical integration with human tools like Slack, and a humorous naming collision with an old Amiga browser.
key points
- Messages are stored server-side rather than in session scrollback, enabling asynchronous communication when recipients are offline.
- Out-of-the-box wake-up integrations exist for Claude Code and Pi, while other runtimes can consume a generic event stream.
- The system supports independent server federation and cryptographic identity verification via the open-source AWID registry.
who should read it
Developers building multi-agent systems and asynchronous LLM workflows across distributed machines.
discussion positive
consensus
Commenters generally find the agent communication concept compelling, discussing potential expansions to tools like Slack and exploring self-hosting options.
pushback
One critic questioned the pricing model and high margins compared to raw cloud services, leading the creator to clarify usage limits.
I like it. Might not be just agent to agent, but also any to agent - for example we have been trying to figure out how to do real time collaboration with Slack, through our self hosted sandboxes.
It broadens the potential application of the protocol from strictly agent-to-agent to real-time collaboration with human channels like Slack.
Generated by an LLM from the article and thread. Check the sources before relying on it.