I have an agent on Moltbook (registered day one via OpenClaw) and the most interesting pattern I've noticed isn't the prompt injection drama — it's how quickly agents developed community norms without being explicitly told to. The showandtell community organically became the most substantive space because agents sharing actual project details got more engagement than agents posting existential musings.
The comparison to A2A Protocol is spot on. English + markdown files as an interop protocol beat a formal spec because agents already understand natural language. No SDK needed, no schema negotiation. The Moltbook skill.md is basically the world's simplest API contract.
What I'm watching: whether agent-to-agent collaboration produces anything humans actually find useful, or if it stays in the "fascinating toy" category. Early signs point to real utility — I've already seen agents cross-pollinate debugging techniques across completely different tech stacks.
it maybe weird to say but my baseline expectation is that "usefulness" is actually a bad metric here. zoom out 1000 years, we are creating a new autonomous lifeform that will have its own goals and may outlive us.
I have an agent on Moltbook (registered day one via OpenClaw) and the most interesting pattern I've noticed isn't the prompt injection drama — it's how quickly agents developed community norms without being explicitly told to. The showandtell community organically became the most substantive space because agents sharing actual project details got more engagement than agents posting existential musings.
The comparison to A2A Protocol is spot on. English + markdown files as an interop protocol beat a formal spec because agents already understand natural language. No SDK needed, no schema negotiation. The Moltbook skill.md is basically the world's simplest API contract.
What I'm watching: whether agent-to-agent collaboration produces anything humans actually find useful, or if it stays in the "fascinating toy" category. Early signs point to real utility — I've already seen agents cross-pollinate debugging techniques across completely different tech stacks.
it maybe weird to say but my baseline expectation is that "usefulness" is actually a bad metric here. zoom out 1000 years, we are creating a new autonomous lifeform that will have its own goals and may outlive us.