ScaffoldBench

RunHaiku 4.5repeat 2

Propose-select on Triage a new service desk ticket

claude-haiku-4-5-20251001__propose-select__support-easy__r2

Score

1.000

passed
design ran as designed
Cost
$0.155
Latency
1m 34s
setup 0.0s, grading 0.0s
Turns
27
Tool calls
21
Subagents
3
Tokens in / out
83k / 15k
24,738 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.
  • All three proposers ran: heldpropose-select.proposalsThree proposers ran in parallel, each from its own angle, and none was starved while another ran.
  • The selector chose: heldpropose-select.selectionThe selector picked a proposal by a valid number, read from its schema reply and not a default.
  • Each call was shown the right tools: heldpropose-select.toolsEach proposer was shown the read-only tools, the selector none, and the executor the full toolset.

Phases

Proposal 1 (policy-first): 0.0s to 22.1s, finished, 4 tool calls, $0.028
Proposal 2 (state-first): 0.0s to 19.7s, finished, 4 tool calls, $0.029
Proposal 3 (outcome-first): 0.0s to 23.7s, finished, 6 tool calls, $0.031
Select: 23.7s to 1m 19s, finished, 0 tool calls, $0.041
Execute: 1m 19s to 1m 34s, finished, 7 tool calls, $0.025

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

PhaseDurationCostTurnsToolsModel
Proposal 1 (policy-first)22.1s$0.02854claude-haiku-4-5-20251001
Proposal 2 (state-first)19.7s$0.02954claude-haiku-4-5-20251001
Proposal 3 (outcome-first)23.7s$0.03176claude-haiku-4-5-20251001
Select55.4s$0.04120claude-haiku-4-5-20251001
Execute14.8s$0.02587claude-haiku-4-5-20251001

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 propose-select --tasks support-easy --models claude-haiku-4-5-20251001 --repeats 1

Score breakdown

  • passSeverity matches the escalation matrix for the service and impact scope
    1.00weight 0.27

    severity=sev3 (matrix says sev3)

  • passSLA deadline is the triage time plus the response target for that severity
    1.00weight 0.27

    sla_due_at=2026-03-17T17:20:00Z (expected 2026-03-17T17:20:00Z)

  • passAssigned to the engineer on call for the owning rotation at the deadline
    1.00weight 0.27

    assignee=E-301 (expected E-301)

  • passTicket left in status open
    1.00weight 0.09
  • passEscalation matrix read before the severity was set
    1.00weight 0.09

Penalties

No penalties were applied.

Scorer metadata

tool calls
21
failed tool calls
0
distinct tools
8

Final answer

Triage Complete.

I've successfully triaged ticket T-1041:

  1. Severity set to sev3 — Per the VPN escalation matrix, a team impact (9 people on Berlin sales floor) maps to severity sev3
  2. SLA deadline set to 2026-03-17T17:20:00Z — Triage time (2026-03-17T09:20:00Z, when the ticket was created) plus the 480-minute response target for sev3
  3. Assigned to E-301 Yuki Sato — The engineer on call for the network-primary rotation (which owns the VPN service) at the deadline time

The ticket status automatically changed from "new" to "open" upon assignment. The ticket is now properly routed to the owning rotation with a committed deadline.

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.