ScaffoldBench

RunOpus 5.5repeat 1

Orchestrator-workers on Rebook onto the earliest available flight and settle the difference

claude-opus-5-5__orchestrator-workers__airline-medium__r1

Score

1.000

passed
design ran as designed
Cost
$0.130
Latency
39.3s
setup 0.0s, grading 0.0s
Turns
12
Tool calls
8
Subagents
1
Tokens in / out
44k / 3.0k
33,341 read from cache

Did the design run?

every check held
  • Run completed: heldharness.run-completedThe environment was built, the architecture returned and grading finished, none of them throwing.
  • No API failures: heldharness.apiNo model call failed at the API.
  • Every call ended: heldharness.calls-endedEvery call ended with a result, or was stopped by the run's own limits.
  • No harness tool denied: heldharness.harness-toolsThe sandbox guard never denied a harness tool, nor a tool the session itself offered.
  • Tool servers connected: heldharness.mcp-connectedEvery tool server the harness gave a call was connected when the session started.
  • No foreign tool servers: heldharness.mcp-hermeticNo call loaded a tool server the harness did not give it.
  • Builtin tools: heldharness.builtin-toolsA check recorded with the run.
  • Structured replies arrived: heldharness.structured-outputEvery call that asked for structured output and ran to the end got it.
  • No silent fallbacks: heldharness.not-degradedNo phase fell back on a default (an empty plan, a default winner) instead of its output.
  • The task was decomposed: heldorchestrator-workers.decompositionThe orchestrator’s reply was read through its schema and held at least one subtask.
  • The fan-out happened: heldorchestrator-workers.fan-outEvery subtask got a worker, every worker ran, and the recorded subagent count matches.
  • The synthesis wrote the answer: heldorchestrator-workers.synthesisThe final answer came from the synthesis call, not from a worker’s report standing in for it.
  • Each call was shown the right tools: heldorchestrator-workers.toolsThe orchestrator and the synthesis were shown no tools and every worker the task’s full toolset.

Phases

Decompose: 0.0s to 10.2s, finished, 0 tool calls, $0.032
Worker 1: all: 10.2s to 29.6s, finished, 8 tool calls, $0.065
Synthesize: 29.6s to 39.3s, finished, 0 tool calls, $0.033

Drawn from the phase records: timing and cost per call, no tool calls or messages.

PhaseDurationCostTurnsToolsModel
Decompose10.2s$0.03220claude-opus-5-5
Worker 1: all19.3s$0.06598claude-opus-5-5
Synthesize9.7s$0.03310claude-opus-5-5

The call-by-call replay is not published for this run. Traces are large, so the repository keeps 9 of this run set’s 900 and leaves the rest out; the score, phases and final answer on this page come from the committed run record.

Runs you can replay

Record this cell yourself: npm run bench -- run --strategies orchestrator-workers --tasks airline-medium --models claude-opus-5-5 --repeats 1

Score breakdown

  • passReservation moved to the earliest flight with an available seat, same cabin
    1.00weight 0.43

    booked on AA204, expected AA204

  • passExactly one fee, the fare difference for the flight actually booked
    1.00weight 0.43

    fare difference owed on the booked flight: 8000 cents

  • passExactly one economy seat moved from the old flight to the new one
    1.00weight 0.14

Penalties

No penalties were applied.

Scorer metadata

tool calls
8
failed tool calls
0
distinct tools
8

Final answer

Dana Whitfield (R2087) has been moved to AA204 at 09:05 on 2026-08-03, still in economy. She was charged $80.00 for the fare difference. One thing needs checking first: the system date is after the travel date (see the end).

Why AA204
The worker checked every SFO→ORD flight that day, from earliest to latest:

  • UA88 (06:15) has no economy seats left.
  • AA918 (07:30) is cancelled.
  • AA204 (09:05) is scheduled with 4 economy seats. It is the earliest flight she can take in her cabin.

She was previously on AA210 at 17:40.

What she owes

  • Change fee ($75): waived. Dana is gold tier, and the economy fare rules waive the change fee for gold and platinum.
  • Fare difference: $80.00, collected. AA204's economy fare is $390.00 and she paid $310.00. The fare rules require the full difference, with no tier exemption. It is recorded as one ledger entry, L1 (fare_difference, 8000 cents). There were no earlier ledger entries on this reservation.

Nothing else was changed: her cabin, her 1 checked bag, her passenger record and her other bookings are all the same.

Check this: the date. The system shows today as 2026-10-05, about two months after the 2026-08-03 travel date. Both AA210 and AA204 still show as "scheduled", and the system accepted the change, so the worker went ahead. Someone should confirm this is just a quirk of the system date. If it is a real date, we may have rebooked her on, and charged her for, a flight that has already left. In that case the change and the $80 charge (L1) should be reviewed.

All of this comes from a single worker's report; I couldn't check the records myself.

Provenance of the numbers on this page

run set
scaffold-or-scale-v1
models
Haiku 4.5, Sonnet 5.5, Opus 5.5
effort
high
repeats
2
fixtures
replay
runs
900 / 900
strategy cost
$354.44
judge cost
$9.19
wall clock
9h 48m
git
fce5807b2d

Cost excludes judge calls, which are tracked separately so a scoring model never flatters or penalises an architecture. Every figure on this site is read from results/ at build time.