Launched this week
ShipOS
The feature flag platform built for the agent era
4 followers
The feature flag platform built for the agent era
4 followers
ShipOS is a feature flag platform where the MCP server and REST API are the primary interface. Claude Code or Cursor can create flags, stage percentage rollouts, target beta users, and pull kill switches, with agent-scoped keys, an agent-attributed audit trail, and approval guardrails for prod. One simple meter, from $9.99/mo. Private alpha: 50% off for life.
Health2Notion
Hey Product Hunt 👋
I'm Mehdi, and I built ShipOS because of a moment that kept repeating: my
coding agent would finish a feature in minutes, code written, PR open
and then everything stopped so a human (me) could log into a flag
dashboard and do data entry. Create the flag the agent already named. Wire
up three environments. Click the rollout slider. The agent did the
engineering; I did the clicking.
That's backwards. So ShipOS treats the MCP server and the API as the
product, and the dashboard as the observation layer. From Claude Code or
Cursor you can create a flag, turn it on in staging, stage a 10% production
rollout, pin it on for your beta cohort, watch evaluation stats, and kill
it instantly if something's wrong. Every action is attributed to an
agent-scoped key in the audit log, and you can require human approval for
the scary stuff (like prod kill switches) while everything else runs
autonomously.
Pricing is deliberately boring: one meter (flag evaluations), public
prices, from $9.99/mo, no "Contact sales". That's the whole model.
Today we're opening the private alpha. Everyone who joins from this
launch locks in 50% off, for life, and gets a direct line to me while
we shape the roadmap.
I'd love the community's take on one question in particular: where should
the line between agent autonomy and human approval sit by default? I have
opinions, but I'd rather hear yours.
Ask me anything, I'll be here all day. 🚢