OpenEdison is only a starting point: Our vision is to govern AI at scale.

In the future, Edison will be a fleet of digital watchdogs that keep agents secure and accountable without human interaction, so you don't have to compromise agent autonomy for security.

Agents become dangerous when the "lethal trifecta" (h/t @simonw) conditions are fulfilled, where an attacker can hijack the agent to exfiltrate your private data!

OpenEdison tracks per-session agent tool use to block write actions when they encounter the lethal trifecta

One-command start: `uvx open-edison`

Discord: discord.gg/tXjATaKgTV
Blog post: edisonwatch.beehiiv.com/p/openediso...
Repo: github.com/Edison-Watc...

Btw, I'm not talking about the classic top signal "note-taking app"

I actually mean, really high-quality, highly critical, highly opinionated writing app

I think AI vs Human labour will play out similar, though caveat in utilitarian human demand vs leisure human demand