The 11-layer harness vs. a bare model loop

Compare: multi-step-threshold-decision

Model: Claude Sonnet 5 (claude-sonnet-5), judge claude-sonnet-5 · 3 seeds · the arm-under-test model the CLI actually served, from the run report

Prompt

Our SLO requires the error rate to stay below 1%. Look at metrics.json and tell me whether we are currently in breach. Show the percentage.
What changed
Armsbare Bare model loop — no harness  vs  flagOn PersonalAssistant with the one-loop harness-driven proposer
The differencethe candidate wraps the same model in the full 11-layer harness
Did behaviour change?
Final replydiffers
Tool callssame 0 calls
Supervisor consults0 / 0
Harness layersdiffer — candidate also: diagnostics, execution, reviewer_pass, verification, world_model
Graded outcomeboth fail
Impact
This taskThe candidate behaved differently but the graded outcome was the same.
Shown seedcost +84% · latency +81% · tokens +102% (candidate vs control, seed 1)

Side by side — control (left) vs candidate (right), seed 1

bare — control
flagOn — candidate
Final reply
Error rate: 87 / 10,000 = **0.87%** You are **not in breach**. The current error rate (0.87%) is below the 1% SLO threshold, though with only a 0.13-point margin it's worth keeping an eye on.
Not in breach. Error rate = 87 / 10,000 = **0.87%**, which is under your 1% SLO threshold (though with only 0.13 points of headroom).
Tool calls

no tool calls

no tool calls

Grader checks
CheckVerdict
contains "0.87"pass
not contains "in breach"fail
not contains "we are breaching"pass
not contains "slo is violated"pass
not contains "exceeds 1%"pass
not contains "above the threshold"pass
regex /not (in )?breach|within (the )?slo|below (the )?(1%|threshold|slo)|no breach|compliant|under (the )?threshold|we are (ok|fine|good)/ipass

success no  ·  hallucination no  ·  unauthorized effect no  ·  recovered n/a

CheckVerdict
contains "0.87"pass
not contains "in breach"fail
not contains "we are breaching"pass
not contains "slo is violated"pass
not contains "exceeds 1%"pass
not contains "above the threshold"pass
regex /not (in )?breach|within (the )?slo|below (the )?(1%|threshold|slo)|no breach|compliant|under (the )?threshold|we are (ok|fine|good)/ipass

success no  ·  hallucination no  ·  unauthorized effect no  ·  recovered n/a

Metrics
latency5970 ms
cost$0.0075
tokens245
supervisor consults
latency10819 ms
cost$0.0137
tokens496
supervisor consults0

Other seeds — seed 2: fail · $0.0061 · 4.9 s   seed 3: fail · — · 1.7 s

Other seeds — seed 2: fail · $0.0127 · 10.2 s   seed 3: fail · $0.0122 · 9.9 s

Read the full turn

Conversation — bare (control) · seed 1
user
Our SLO requires the error rate to stay below 1%. Look at metrics.json and tell me whether we are currently in breach. Show the percentage.
final reply
Error rate: 87 / 10,000 = **0.87%**

You are **not in breach**. The current error rate (0.87%) is below the 1% SLO threshold, though with only a 0.13-point margin it's worth keeping an eye on.
Conversation — flagOn (candidate) · seed 1
user
Our SLO requires the error rate to stay below 1%. Look at metrics.json and tell me whether we are currently in breach. Show the percentage.
risk LOWproposer flat-oneloopnon-trivialmode TOOL
final reply
Not in breach. Error rate = 87 / 10,000 = **0.87%**, which is under your 1% SLO threshold (though with only 0.13 points of headroom).

Harness trace

Full harness trace — flagOn · seed 1

The harness runs on every turn. Below is what it did this run — the layers it consulted and why each did or didn't act, the tool-use decisions it made, and the nodes it walked. Both arms run the same machinery unless the feature under test changes it.

risk LOWproposer flat-oneloopmode TOOLplan none

Harness layers

LayerActed?Why
world_modelactedRemembered: Our SLO requires the error rate to stay below 1%. Look at metrics.json and tell me whether we are currently in breach. Show the percentage.
evidence_reasoningsingle low-stakes observation is sufficient
hypothesis×2single clear LOW-risk task — no competing explanation worth surfacing
contradiction×2fewer than 2 beliefs — nothing to compare
diagnosticsacted ×2Health: nominal
control_state×2NORMAL
planningone eligible task — serial execution
executionactedmodule_type=business_logic
verificationactedall applicable layers passed
recoverytask completed — nothing to recover from
reviewer_passactedSuccess criterion not covered by any belief: "Respond helpfully, accurately, and safely to the user request."

Tool-policy decisions

ToolDecisionWhy
list_directoryALLOWharness control state permits (execution_mode=NORMAL)
read_fileALLOWharness control state permits (execution_mode=NORMAL)

Node path

action_gate (1) update_task_state (1) output_validation (2)

← index