FA-99906 / Agent workflow termination / Open access

Workflow: batch is settled when any tool call returns · case 01

The workflow advances after the first result of a multi-call batch.

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

ROOT CAUSE

Batch readiness is computed with any instead of all.

THE FAILURE

Batch readiness is computed with any instead of all.

Unsuccessful approach: Requiring a result in a nonempty batch makes an empty batch appear unsettled.

Case contract

A batch is settled iff it is empty or every call status is result.

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(statuses):
    return not statuses or any(s == 'result' for s in statuses)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('all results', solve(['result']*N),True)
check('one outstanding', solve(['result','pending']),False)
check('all pending', solve(['pending']*N),False)
check('empty batch', solve([]),True)
check('singleton result', solve(['result']),True)
check('singleton pending', solve(['pending']),False)
check('failure is settled result contract false', solve(['failed']),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
all results	`True`	`True`	Passed
one outstanding	`True`	`False`	Failed
all pending	`False`	`False`	Passed
empty batch	`True`	`True`	Passed
singleton result	`True`	`True`	Passed
singleton pending	`False`	`False`	Passed
failure is settled result contract false	`False`	`False`	Passed

SHA-256 / 7debaaff680d4786264bdd06a3a53e0a43c36000d1000e8a1f5d4920883cedd2

2 / The unsuccessful fix

Exit 1

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

N = 1
observations = []
def solve(statuses):
    return bool(statuses) and all(s == 'result' for s in statuses)
def check(label, actual, expected):
    observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('all results', solve(['result']*N),True)
check('one outstanding', solve(['result','pending']),False)
check('all pending', solve(['pending']*N),False)
check('empty batch', solve([]),True)
check('singleton result', solve(['result']),True)
check('singleton pending', solve(['pending']),False)
check('failure is settled result contract false', solve(['failed']),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
all results	`True`	`True`	Passed
one outstanding	`False`	`False`	Passed
all pending	`False`	`False`	Passed
empty batch	`False`	`True`	Failed
singleton result	`True`	`True`	Passed
singleton pending	`False`	`False`	Passed
failure is settled result contract false	`False`	`False`	Passed

SHA-256 / d3a734373a11baf437f6f31c9bec048937991f664c8d2c2d6ea7876db86bacb1

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

Only result marks a successful returned call; failed calls need a separate policy. 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 2026-09-30T16:48:29.418888+00:00.

Case digest / 9faa2f80cabe72ae166935a8ccfdbc3c40708317ac88dc58aeff93e66c3c6dff