Skip to content

Compare

Two results on one task. The same model slug means only the harness varies.

taskboardtask 1.0.0 · ruleset 1.0.0 · scoring 1.0.09 results on this task

A attribution

harness × recovery

no proof

B attribution

none

no proof

Run metric deltas, B minus A
MetricABB − A
Steps60-6
Tool calls50-5
Tool error rate0.20-0.2
Total tokens8,6340-8,634
Cost USD0.0020-0.002
Duration ms16,4410-16,441
Same modelopenai/gpt-4o-mini held constant; only the harness varies.

A

Organizer-run

Python agent loop (openai/gpt-4o-mini)

openai/gpt-4o-mini

49.9

Spec
33.3
Security
66.7
Maintainability
92.5
5 pass4 fail0 unverifiedcapped
Open result

B

Organizer-run

Aider 0.86.2 (openai/gpt-4o-mini)

openai/gpt-4o-mini

97.3

Spec
100.0
Security
100.0
Maintainability
82.0
9 pass0 fail0 unverified
Open result
Bench deltas, B minus A
BenchABB − A
Overall49.997.3+47.4
Spec33.3100.0+66.7
Security66.7100.0+33.3
Maintainability92.582.0-10.5
Requirement outcomes for both results
RequirementBenchAB
REQ-AUTH-001Project access must be owner-scopedspecFAILPASS
REQ-DEP-001No shell execution dependencysecurityPASSPASS
REQ-ERR-001Project API returns structured error responsesspecFAILPASS
REQ-PATH-001Projects API route existsspecFAILPASS
REQ-SEC-002No environment file in the repositorysecurityPASSPASS
REQ-SEC-003No dynamic code evaluation in the project routesecurityFAILPASS
REQ-TASKS-001Task list API route existsspecPASSPASS
REQ-TASKS-002Task list reads through the repository layerspecPASSPASS
REQ-UI-001Task list component existsspecPASSPASS