Fabric pushed the constraint update while I was mid-test.
Speed envelope tightened a few percent. Proposal cleared. Governance vote passed. Policy hash flipped before my optimizer finished its run.
I had a new motion profile ready. Better cornering. Lower vibration under load. Nothing reckless. Just tighter.
Couldn't deploy it.
The update changed the max_velocity parameter I was tuning against. Seventeen minutes in. Convergence at 62%. The new constraint invalidated the artifact. CI flagged it against the previous envelope version. On Fabric foundation.
So I retrained.
Not because the robot failed.
Because the rule moved.
Edge logs clean. Agent runtime stable. No silent drift. The Fabric policy hash advanced and my branch was suddenly behind it.
Live governance inside a build cycle feels different.
I adjust a threshold. Someone else proposes a tighter safety margin. Vote passes. The allowed surface shifts. Just enough that the optimizer starts over. Same hardware. Same runtime. Different ceiling.
Throughput of the robots stayed high.
My iteration clock didn’t.
I re-queued certification. Rebased against the new policy hash. Watched the pipeline wait for the vote to close before tagging the release.
The robot could execute the new profile right now. Mechanically capable.
Ledger says not yet.
Another proposal already open on @Fabric Foundation . Different region. Different risk tolerance. If that one clears, I rebase again. Maybe retune damping. Maybe not.
Policy hash keeps moving.
I keep watching the governance thread instead of the telemetry stream.
Optimizer still running.
Constraint might shift again before it converges.