Why Didn't That Ticket Start? YOLO Mode Now Shows Its Decision
YOLO Mode starts a ticket without waiting for you when Cogny is confident you would approve it. Until now, when a ticket you expected to start just sat in New, there was no way to see why. Was the score too low? Did a rule stop it? Did your policy say no, or did the scorer simply not read it the way you meant?
Now every decision is written down, on the ticket.
What gets recorded
Each time YOLO Mode looks at a ticket it leaves a note in the ticket's history and keeps the latest decision on the ticket itself. A decision is one of three things:
- Started. The score reached your threshold and the rules allowed it.
- Left for review, score below the threshold. With the score, the threshold and the scorer's reasons.
- Left for review, stopped by a rule. The score was high enough, but the ticket uses a tool that changes a platform, has nothing to run with, or there is no repository for the coding agent. The note names the rule.
When you have written a YOLO policy, the first reason says how it applied: which part allowed the ticket, which part said it needs a person, or that the policy doesn't cover it. That line is the one to read before you edit the policy.
You find the decision in three places: the ticket's history, the ticket's "Ids, dates and YOLO Mode" section, and a list of the latest decisions under Settings → YOLO Mode.
A policy change now reaches held tickets
Saving a new policy or threshold used to affect only tickets created afterwards. Now the tickets YOLO Mode held in the last 30 days that are still in New are checked again within a few minutes. If the ticket you had in mind passes under the new policy, it starts.
Two things stay as they were. The rules about what may run on its own don't move with the policy: work that changes a platform still waits for a person. And a ticket YOLO Mode never looked at, one you created by hand or one from before YOLO Mode was on, is not picked up by a policy change. For those there is Check again on the ticket: it gets a fresh score against the current policy and starts if it passes.
A policy that says a kind of work is fine now carries more weight, too. A sound ticket your policy explicitly allows scores 90 or higher, so "SEO content updates are ok" does what it says at the default threshold of 85.
Scoring a ticket uses a small amount of credits. A threshold-only change reuses the score a ticket already has.
From your own agent
Three MCP tools cover the same ground:
get_yolo_historylists the decisions in a period (the last 7 days by default, up to 90), or one ticket's decision withticket_id. Each entry says whether it was made under your current policy and what happened to the ticket since: still waiting, approved by a person anyway, or rejected. Tickets you approved after YOLO Mode held them are the clearest sign the policy is too tight.review_ticket_for_yolochecks one ticket in New. Withdry_run: trueit only reports the score and what YOLO Mode would do.update_yolo_settingsturns YOLO Mode on or off, sets the threshold and replaces the policy. It is for owners and admins, and your agent should show you the old and new policy text before calling it.
A prompt that works: "Look at what YOLO Mode held this month. Which of those did we approve by hand? Suggest a policy change that would have let them through, and show me the text before you save it."
More on tickets, scores and what never runs without you in Tickets and approvals.