Recite the no-progress streak and spawn count to the decider#1263
Merged
Conversation
Both bounds existed and killed runs (SupervisorBounds force-stops on the no-progress cap and refuses waves past the total-spawn cap), but the decider saw neither — it could only infer the streak from a post-hoc tier-escalation note. That blindness is the shape of the observed evidence-less spawn/retry loops: the model burns its runway without knowing a runway exists. SupervisorBoundsRecitation renders both counters at the prompt tail beside the budget recitation: the streak line teaches what SETTLED EVIDENCE means (accepted unit / merge integrating new work / answered ask), counts down the remaining decisions explicitly, and names the honest exits (stop or ask) over burning turns; the spawn line shows count against the effective cap (lane default when the context carries none). Null while both counters are zero — a fresh run's prompt stays byte-identical.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
NoProgressDecisions/MaxNoProgressDecisions) and the total-spawn count (TotalSpawnedAgents/effective cap) — now render at the decider's prompt tail (SupervisorBoundsRecitation, beside the P3.5 budget recitation). Previously the model could only infer the streak from a post-hoc A2 escalation note; the evidence-less spawn/retry loops in the decision-eval reds are exactly a model burning a runway it cannot see.Test plan
DefaultMaxTotalSpawns) · pinned header · prompt wiring (block present once counters move, absent on a fresh run)