FA-99936 / Agent workflow termination / Open access

Workflow: workflow bypasses its finalization barrier · case 01

The run closes before all required wrap-up checks finish.

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

ROOT CAUSE

Completion is decided from a terminal response without checking barriers.

THE FAILURE

Completion is decided from a terminal response without checking barriers.

Unsuccessful approach: Waiting for at least one barrier rejects workflows whose barrier list is empty.

Case contract

Return true iff final_seen and all barrier flags are true; an empty barrier list is ready.

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(final_seen, barriers):
    return bool(final_seen)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('final all ready', solve(True,[True]*N),True)
check('one barrier pending', solve(True,[True,False]),False)
check('no final', solve(False,[True]),False)
check('no barriers', solve(True,[]),True)
check('empty and unfinished', solve(False,[]),False)
check('single ready', solve(True,[True]),True)
check('single pending', solve(True,[False]),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
final all ready	`True`	`True`	Passed
one barrier pending	`True`	`False`	Failed
no final	`False`	`False`	Passed
no barriers	`True`	`True`	Passed
empty and unfinished	`False`	`False`	Passed
single ready	`True`	`True`	Passed
single pending	`True`	`False`	Failed

SHA-256 / d98529f48deeed876dcb5fdaa1b54431c5eb2a3b09de53b5f3039017cdf032da

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(final_seen, barriers):
    return bool(final_seen) and bool(barriers) and all(barriers)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('final all ready', solve(True,[True]*N),True)
check('one barrier pending', solve(True,[True,False]),False)
check('no final', solve(False,[True]),False)
check('no barriers', solve(True,[]),True)
check('empty and unfinished', solve(False,[]),False)
check('single ready', solve(True,[True]),True)
check('single pending', solve(True,[False]),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
final all ready	`True`	`True`	Passed
one barrier pending	`False`	`False`	Passed
no final	`False`	`False`	Passed
no barriers	`False`	`True`	Failed
empty and unfinished	`False`	`False`	Passed
single ready	`True`	`True`	Passed
single pending	`False`	`False`	Passed

SHA-256 / 4c8872f42e3d5670e2db30c02700cac1f0e7c7ec6513a9e5c187cebf408a5629

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

Barrier booleans represent already evaluated completion prerequisites. 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 2026-09-30T16:48:29.887005+00:00.

Case digest / 725fb87c1da93fd21dd5682247c9a7f49113ec3580085d6c3ea867abfe7e56b0