workledger

Observability tells you what ran. workledger tells you what work happened.

This report turns raw traces into normalized observations, rolled work units, and explicit review-needed work. Policy and economics remain downstream views.

work_unit_count
2
total_direct_cost
0.084
total_blended_cost
0.0966
avg_compression_ratio
1.12
avg_confidence
0.0
review_burden
1
pending_review_count
1
total_source_spans
8

Dataset Context

DatasetSplitAdapterRowsTracesObservations
kshitijthakkar/smoltrace-traces-20260130_053009trainsmoltrace228

Raw Trace Excerpt

TitleTraceStepsExcerptSource ref
Work through multi-step prompttrace-smol-25Work through multi-step prompt -> Retrieve docs -> Run calculator -> Reason over docs -> Return answerhf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-1
Solve benchmark prompttrace-smol-13Solve benchmark prompt -> Search background facts -> Draft final answerhf://kshitijthakkar/smoltrace-traces-20260130_053009/train/0#span-1

Normalized Observations

TraceSpanKindNameSource ref
trace-smol-2root-2agentWork through multi-step prompthf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-1
trace-smol-2retrieve-2retrieverRetrieve docshf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-2
trace-smol-2tool-2toolRun calculatorhf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-3
trace-smol-2llm-2allmReason over docshf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-4
trace-smol-2llm-2bllmReturn answerhf://kshitijthakkar/smoltrace-traces-20260130_053009/train/1#span-5
trace-smol-1root-1agentSolve benchmark prompthf://kshitijthakkar/smoltrace-traces-20260130_053009/train/0#span-1
trace-smol-1tool-1toolSearch background factshf://kshitijthakkar/smoltrace-traces-20260130_053009/train/0#span-2
trace-smol-1llm-1llmDraft final answerhf://kshitijthakkar/smoltrace-traces-20260130_053009/train/0#span-3

Rolled Work Units

TitleKindReviewTrustEvidenceSpansTotal cost
Work through multi-step promptaugmented_ai_workqueuedunreviewed65$0.0644
Solve benchmark promptaugmented_ai_workunreviewedunreviewed43$0.0322

Review-Needed Work

TitleKindReviewReasonSpansEvidence
Work through multi-step promptaugmented_ai_workqueuedqueued during trace-to-work rollup56

Cost by Policy Outcome

Policy outcomeBlended costWork units
No policy outcome totals available

Pending Review Queue

TitleCategoryOutcomeCostConfidenceSpansCompressionCompeting
No pending review items

Compression Proof Point

Work through multi-step prompt condensed 5 source spans into a 1.25x business work signal.

Top Ambiguous Items

TitleCategoryConfidenceGap
No ambiguous items

Low-Trust High-Cost Outputs

TitleCategoryCostTrust
No low-trust high-cost outputs