FA-99916 / Agent workflow termination / Open access
Workflow: budget exhaustion is reported as successful completion · case 01
A run with no remaining work is labeled completed despite having no final answer.
ROOT CAUSE
The finish reason is inferred from an empty queue alone.
THE FAILURE
The finish reason is inferred from an empty queue alone.
Unsuccessful approach: Prioritizing the budget label misclassifies a run that did produce its final answer.
Case contract
Given final_seen and remaining_steps, return complete when final_seen, budget_exhausted when steps are zero, else continue.
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, remaining_steps):
return 'complete' if remaining_steps == 0 else 'continue'
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('final at exhausted budget', solve(True,0),'complete')
check('budget without final', solve(False,0),'budget_exhausted')
check('active without final', solve(False,N),'continue')
check('final with room', solve(True,N),'complete')
check('positive step', solve(False,1),'continue')
check('extra budget without final', solve(False,N+1),'continue')
check('answer wins', solve(True,1),'complete')
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 at exhausted budget | complete | complete | Passed |
| budget without final | complete | budget_exhausted | Failed |
| active without final | continue | continue | Passed |
| final with room | continue | complete | Failed |
| positive step | continue | continue | Passed |
| extra budget without final | continue | continue | Passed |
| answer wins | continue | complete | Failed |
SHA-256 / 6be80ec96f0ce5eef881462f8ed672fc083c2b44d41cfda28a6d0594e1459290
2 / The unsuccessful fix
Exit 1"""Failure Map reference implementation. Python standard library only."""
import json
N = 1
observations = []
def solve(final_seen, remaining_steps):
return 'budget_exhausted' if remaining_steps == 0 else ('complete' if final_seen else 'continue')
def check(label, actual, expected):
observations.append({"check": label, "actual": actual, "expected": expected, "passed": actual == expected})
check('final at exhausted budget', solve(True,0),'complete')
check('budget without final', solve(False,0),'budget_exhausted')
check('active without final', solve(False,N),'continue')
check('final with room', solve(True,N),'complete')
check('positive step', solve(False,1),'continue')
check('extra budget without final', solve(False,N+1),'continue')
check('answer wins', solve(True,1),'complete')
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 at exhausted budget | budget_exhausted | complete | Failed |
| budget without final | budget_exhausted | budget_exhausted | Passed |
| active without final | continue | continue | Passed |
| final with room | complete | complete | Passed |
| positive step | continue | continue | Passed |
| extra budget without final | continue | continue | Passed |
| answer wins | complete | complete | Passed |
SHA-256 / d7fe2651f870976e495ad73777b0cf6dde1a2d101b297f91a0d7b120cac26b1d
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.
Sign in to the archive ↗Verification & scope
remaining_steps is nonnegative; failure and cancellation reasons are outside this precedence rule. 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 2026-09-30T16:48:29.449039+00:00.
Case digest / 5afa2c118f2ee54768e6025c3bace123cdaf4452805fc938b2bc72359b4c3048