FA-99886 / Agent workflow termination / Open access

Workflow: a turn starts without reserving its maximum output budget · case 01

A turn begins even though its required reserve would exceed the run budget.

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

ROOT CAUSE

The guard checks that some budget remains instead of checking the requested reserve.

THE FAILURE

The guard checks that some budget remains instead of checking the requested reserve.

Unsuccessful approach: Requiring strictly more than the reserve rejects an exactly fitting turn.

Case contract

Return whether remaining_units >= reserve_units.

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(remaining_units, reserve_units):
    return remaining_units > 0
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('exact reserve', solve(N,N),True)
check('one short', solve(N-1,N),False)
check('surplus', solve(N+1,N),True)
check('zero reserve', solve(0,0),True)
check('empty budget', solve(0,1),False)
check('one unit exact', solve(1,1),True)
check('positive budget zero reserve', solve(N,0),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
exact reserve	`True`	`True`	Passed
one short	`False`	`False`	Passed
surplus	`True`	`True`	Passed
zero reserve	`False`	`True`	Failed
empty budget	`False`	`False`	Passed
one unit exact	`True`	`True`	Passed
positive budget zero reserve	`True`	`True`	Passed

SHA-256 / 3c7be3c477b21da0908f801566e66c843f0ccffe3daad9e972dfc7d795762023

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(remaining_units, reserve_units):
    return remaining_units > reserve_units
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('exact reserve', solve(N,N),True)
check('one short', solve(N-1,N),False)
check('surplus', solve(N+1,N),True)
check('zero reserve', solve(0,0),True)
check('empty budget', solve(0,1),False)
check('one unit exact', solve(1,1),True)
check('positive budget zero reserve', solve(N,0),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
exact reserve	`False`	`True`	Failed
one short	`False`	`False`	Passed
surplus	`True`	`True`	Passed
zero reserve	`False`	`True`	Failed
empty budget	`False`	`False`	Passed
one unit exact	`False`	`True`	Failed
positive budget zero reserve	`True`	`True`	Passed

SHA-256 / 7a0b33d6a410f2bbbd7eaf9e2b9e4fda285dea1ed6624d1a9a25640e9615d81f

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

Budgets are nonnegative integer units; reservation consumption occurs after admission. 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 2026-09-30T16:48:29.419576+00:00.

Case digest / 1a519117bb83d9e6ec6f3a6445c7ce5eeda956be172150e15d8fc151931fed7d