Module telemetry 

Purpose: Track ProjectAtlas token savings telemetry.

Structs

AgentEfficiencyArtifactIdentity

Identity retained from one validated benchmark artifact.

AgentEfficiencyBaselineRow

One matched baseline comparison projected from the benchmark artifact.

AgentEfficiencyBreakEven

Workload-specific setup/runtime break-even truth.

AgentEfficiencyCapabilityContribution

Trace-completed v0.4 MCP calls grouped by navigation responsibility.

AgentEfficiencyComparison

Optional controlled benchmark comparison attached to live token telemetry.

AgentEfficiencyMetricComparison

Candidate and baseline distribution summary for one navigation metric.

AgentEfficiencyProviderMetric

Descriptive-only candidate and baseline provider counter.

TokenAccountingTotals

Wide separated accounting totals before narrowing to the public report representation.

TokenBucketOverview

Aggregated token savings for one bucket and counting mode.

TokenCalibrationOverview

Optional local tokenizer calibration for indexed UTF-8 files.

TokenOverview

Token savings overview.

TokenTrendPeriod

Token trend aggregate for one period.

TokenTrendReport

Token savings trend report.

UsageEvent

Token savings event for a funnel command.

UsageInstanceId

One bounded CLI invocation or MCP process inside an authoritative project database.

Enums

AgentEfficiencyBaseline

Baseline arm compared with the v0.4 candidate.

AgentEfficiencyCapability

Durable ProjectAtlas navigation capability represented in the benchmark.

AgentEfficiencyEvidenceState

Validation state for optional agent-efficiency benchmark evidence.

AgentEfficiencyMetricKind

Closed navigation metric projected from matched benchmark trials.

AgentEfficiencyProviderMetricKind

Provider counter represented only as descriptive benchmark context.

TelemetryContractError

Typed telemetry-domain validation failure.

TokenTrendWindow

Token trend grouping window.

UsageDetailAvailability

Truth state for caller-label and raw telemetry detail.

UsageInstanceOwner

Runtime owner of one internal telemetry instance.

Constants

READ_AVOIDANCE_CONFIDENCE_MODELED

Read-avoidance confidence for modeled navigation events.

READ_AVOIDANCE_CONFIDENCE_NOT_RECORDED

Read-avoidance confidence when raw command evidence is unavailable.

READ_AVOIDANCE_CONFIDENCE_OBSERVED

Read-avoidance confidence for directly observed full-file compression events.

READ_AVOIDANCE_SCOPE

Human-facing explanation for likely read-avoidance counters.

TOKENIZER_BACKEND_HEURISTIC

Default token-count backend for offline estimates.

TOKEN_ACCOUNTING_MODELED_AVOIDANCE

Modeled counterfactual accounting layer.

TOKEN_ACCOUNTING_OBSERVED_DELTA

Observed before/after accounting layer.

TOKEN_ACCURACY_HEURISTIC

Accuracy label for the default offline estimator.

TOKEN_BASELINE_DIRECTORY_WALK

Baseline kind for broad directory-walk navigation savings.

TOKEN_BASELINE_FULL_FILE

Baseline kind for a concrete full-file comparison.

TOKEN_BASELINE_SELECTED_CANDIDATES

Baseline kind for inferred candidate-set navigation savings.

TOKEN_BUCKET_FULL_FILE_COMPRESSION

Bucket for source compression through summaries, outlines, search, or slices.

TOKEN_BUCKET_NAVIGATION_AVOIDANCE

Bucket for navigation that avoids broad folder/file exploration.

TOKEN_COMMAND_MCP_FILE_SUMMARY

MCP event label for file summaries.

TOKEN_COMMAND_MCP_OUTLINE

MCP event label for file outlines.

TOKEN_COMMAND_MCP_SEARCH

MCP event label for indexed search.

TOKEN_COMMAND_MCP_SLICE

MCP event label for source slices.

TOKEN_COMMAND_OUTLINE

CLI command label for file outlines.

TOKEN_COMMAND_SEARCH

CLI command label for indexed search.

TOKEN_COMMAND_SLICE

CLI command label for source slices.

TOKEN_COMMAND_SUMMARY

CLI command label for file summaries.

TOKEN_COMMAND_SYMBOL_SLICE

CLI command label for symbol slices.

TOKEN_CONFIDENCE_INFERRED

Confidence label for inferred navigation comparisons.

TOKEN_CONFIDENCE_OBSERVED

Confidence label for observed source-compression comparisons.

TOKEN_CONFIDENCE_POLICY_ESTIMATE

Confidence label for policy-modeled navigation comparisons.

TOKEN_DEDUPE_SCOPE_EVENT

Dedupe scope for measured one-off events.

TOKEN_DEDUPE_SCOPE_SESSION

Dedupe scope for repeated modeled workflow baselines in one session.

TOKEN_ESTIMATE_KIND

Token overview counting mode.

TOKEN_ESTIMATE_METHOD_HEURISTIC

Default method label for heuristic token estimates.

TOKEN_ESTIMATE_SCOPE

Token overview scope label.

TOKEN_ESTIMATOR

Token overview estimator identifier.

TOKEN_MODEL_UNKNOWN

Default model label when no model-specific counter is used.

TOKEN_PROVIDER_HEURISTIC

Default token-count provider label for offline estimates.

TOKEN_TRACE_HEURISTIC

Trace label for the default heuristic calculation.

Functions

default_accounting_layer

Default accounting layer for legacy usage events.

default_baseline_identity

Build a stable baseline identity from existing event context.

default_dedupe_scope

Default dedupe scope for legacy usage events.

default_denominator_kind

Default denominator kind for legacy usage events.

default_estimate_method

Default estimate method for legacy usage events.

default_read_avoidance_confidence

Default read-avoidance confidence for legacy serialized overviews.

default_read_avoidance_scope

Default read-avoidance scope for legacy serialized overviews.

default_token_accuracy

Default accuracy label for legacy usage events.

default_token_baseline_kind

Default baseline kind for legacy usage events.

default_token_confidence

Default confidence label for legacy usage events.

default_token_model

Default token model for legacy usage events.

default_token_provider

Default token provider for legacy usage events.

default_token_savings_bucket

Default token savings bucket for legacy usage events.

default_token_trace

Default calculation trace for legacy usage events.

default_tokenizer_backend

Default tokenizer backend for legacy usage events.

usage_from_estimates

Create a usage event from already-computed token estimates.

usage_from_estimates_with_accounting

Create a usage event from token estimates and explicit accounting semantics.

usage_from_estimates_with_context

Create a usage event from token estimates and explicit baseline semantics.

usage_from_text

Create a usage event from response text and baseline text.