What counts as an agent?
An Operon agent is one reusable AI bundle. Under the hood, it is packaged as a SKILL.md plus any files it needs, so the same artifact runs in Claude Code, Codex, Cursor, and ChatGPT. Operon mines agents from sessions, polishes them, and ships them.
Is my session data uploaded?
No. operon mine reads sessions on the laptop they happened on, clusters them locally, and redacts before anything leaves the machine. Only the polished agent you choose to publish is uploaded. Drafts can be staged in the dashboard for review, but the raw session log never moves.
What's a workspace?
A workspace is the unit of access in Operon. Private agents, members, run history, and audit log are all scoped to one workspace. A Team plan is one workspace. Free mode has no private workspace; you publish under your user handle to the public registry.
Can I self‑host?
Yes, under Enterprise. The CLI already supports --api-base for pointing at your own deploy. The control plane is a single Next.js app plus Postgres; we will help you stand it up in your VPC.
When does Team pricing land?
We are running Team in preview with design partners while we finalize seat vs. workspace pricing. If you want in early, email hi@withoperon.com and tell us what your team looks like.
Can I still write agents by hand?
The CLI is free. You can write SKILL.md files in any editor and publish them with operon publish. Where Operon earns its keep is mining: it surfaces the prompts your team is already running so you do not have to start from a blank file.