FA-99866 / Agent workflow termination / Open access

Workflow: a productive turn fails to reset the stall counter · case 01

The workflow stops after separated idle turns even though progress reset the streak.

Verified by executionVariant 1 · 7 checks per implementationDownload source bundle ↓JSON ↗

ROOT CAUSE

The implementation counts all idle turns cumulatively.

THE FAILURE

The implementation counts all idle turns cumulatively.

Unsuccessful approach: Checking one fewer event than the limit can stop early on a shorter idle streak.

Case contract

Return true when the number of consecutive false progress flags at the end reaches limit.

Why this case matters

This toy decision rule isolates one bounded agent-loop completion contract.

1 / The failure

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(progress, limit):
    return sum(not x for x in progress) >= limit
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('progress breaks streak', solve([False,False,True,False],2),False)
check('trailing stall', solve([True]+[False]*N,N),True)
check('short idle streak', solve([False],2),False)
check('empty history', solve([],1),False)
check('exact threshold', solve([False]*N,N),True)
check('productive tail', solve([False]*N+[True],N),False)
check('limit one', solve([True,False],1),True)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
progress breaks streak	`True`	`False`	Failed
trailing stall	`True`	`True`	Passed
short idle streak	`False`	`False`	Passed
empty history	`False`	`False`	Passed
exact threshold	`True`	`True`	Passed
productive tail	`True`	`False`	Failed
limit one	`True`	`True`	Passed

SHA-256 / 70f16b7f5e082f9fc711a656b554e8f07368c8f7ab11ccdeff7e777edd4abdbb

2 / The unsuccessful fix

Exit 1

"""Failure Map reference implementation. Python standard library only."""
import json

N = 1
observations = []
def solve(progress, limit):
    return len(progress) and all(not x for x in progress[-limit+1:])
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('progress breaks streak', solve([False,False,True,False],2),False)
check('trailing stall', solve([True]+[False]*N,N),True)
check('short idle streak', solve([False],2),False)
check('empty history', solve([],1),False)
check('exact threshold', solve([False]*N,N),True)
check('productive tail', solve([False]*N+[True],N),False)
check('limit one', solve([True,False],1),True)
print(json.dumps({"observations": observations, "passed": all(x["passed"] for x in observations)}, ensure_ascii=False))
raise SystemExit(0 if all(x["passed"] for x in observations) else 1)

Boundary fixture	Actual	Expected	Outcome
progress breaks streak	`True`	`False`	Failed
trailing stall	`False`	`True`	Failed
short idle streak	`True`	`False`	Failed
empty history	`0`	`False`	Passed
exact threshold	`True`	`True`	Passed
productive tail	`False`	`False`	Passed
limit one	`False`	`True`	Failed

SHA-256 / 473afc589bf5eac9b092656aca7449a685a46af3d38e46321f12b281bdacb5b1

HELD IN THE MEMBER ARCHIVE

The verified repair and its recorded checks are member-only.

This mechanism has 7 recorded checks per implementation. The open-access tier publishes the failure and the unsuccessful fix; the repaired source that passes every check, and the observations that prove it, are available to members.

Every case sharing this mechanism uses the same contract and the same repair, so this one record is held back for all of them.

Member access is invitation-based. Sign in with your invited account to inspect the repair.

Verification & scope

Progress flags are booleans; limit is a positive integer. This reproducer isolates one failure mechanism. Results cover the supplied fixtures. Variants within a family share a test contract and should remain grouped when constructing evaluation splits. Related mechanisms with a shared evaluation_group must also remain together; these controlled models are not independent production incidents.

Observations recorded using Python 3.12.14 at 2026-09-30T16:48:28.969417+00:00.

Case digest / 32f355b81a8924c418512b9b9122783d8e3441a0ba3452ac86952acb32208bf7