FA-99856 / Agent workflow termination / Open access

Workflow: progress is measured over the wrong window · case 01

A productive run is stopped even though a progress event is recent.

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

ROOT CAUSE

The guard scans the oldest slice instead of the trailing window.

THE FAILURE

The guard scans the oldest slice instead of the trailing window.

Unsuccessful approach: Scanning the entire history misses a genuinely stalled trailing window.

Case contract

Given boolean progress events and positive width, report stalled when fewer than threshold true events occur in the trailing width entries.

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(events, width, threshold):
    return sum(events[:width]) < threshold
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('recent progress', solve([False]*N+[True],N,1),False)
check('recent stall', solve([True]+[False]*N,N,1),True)
check('below threshold', solve([True,False,False],3,2),True)
check('at threshold', solve([True,True,False],3,2),False)
check('empty history', solve([],N,1),True)
check('short productive history', solve([True],N,1),False)
check('zero threshold', solve([False]*N,N,0),False)
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
recent progress	`True`	`False`	Failed
recent stall	`False`	`True`	Failed
below threshold	`True`	`True`	Passed
at threshold	`False`	`False`	Passed
empty history	`True`	`True`	Passed
short productive history	`False`	`False`	Passed
zero threshold	`False`	`False`	Passed

SHA-256 / 6aadec45fe90c0a9b419a6e9d67c8bbe1a7945d70fe4d44d3bd8ad3e042ae6ac

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(events, width, threshold):
    return sum(events) < threshold
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('recent progress', solve([False]*N+[True],N,1),False)
check('recent stall', solve([True]+[False]*N,N,1),True)
check('below threshold', solve([True,False,False],3,2),True)
check('at threshold', solve([True,True,False],3,2),False)
check('empty history', solve([],N,1),True)
check('short productive history', solve([True],N,1),False)
check('zero threshold', solve([False]*N,N,0),False)
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
recent progress	`False`	`False`	Passed
recent stall	`False`	`True`	Failed
below threshold	`True`	`True`	Passed
at threshold	`False`	`False`	Passed
empty history	`True`	`True`	Passed
short productive history	`False`	`False`	Passed
zero threshold	`False`	`False`	Passed

SHA-256 / 178e71bfa7d7d61599ded4d170da29266cbee3f45c2e8632750e233e8625b641

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

Events are booleans; width is positive and threshold is nonnegative. 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 2026-09-30T16:48:28.970591+00:00.

Case digest / 94863c4264a741d968dc456c846d85975bf15db6677b0dace9dae72dd6de0125