UNVEIL A HUMAN'S OK consent for risky actions
no — spawn / resume
yes — a gated action
approved
not approved
an agent wants to act
is it fork, merge, or deactivate?
runs right away
— for spawn and resume, the messagethat triggered them is the consent
parked — stored, not run yet
a human reacts with a check-mark in Slack?
run it — but only once
(the approval is used up, so asecond check-mark does nothing)
nothing happens
UNVEIL A HUMAN'S OK consent for risky actions
no — spawn / resume
yes — a gated action
approved
not approved
an agent wants to act
is it fork, merge, or deactivate?
runs right away
— for spawn and resume, the messagethat triggered them is the consent
parked — stored, not run yet
a human reacts with a check-mark in Slack?
run it — but only once
(the approval is used up, so asecond check-mark does nothing)
nothing happens