Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Unified pipeline telemetry — collects per-stage performance metrics (CPU, memory, I/O) and diagnostic events into structured reports (JSON, text).
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | -73% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -73% | 0% |
This module provides unified pipeline telemetry for the infrastructure layer. It bridges StagePerformanceTracker resource metrics with DiagnosticReporter validation events into a single TelemetryCollector.
| Symbol | Type | Purpose | | --- | --- | --- | | TelemetryConfig | dataclass | Configuration with YAML-loadable from_dict() | | TelemetryCollector | class | Main lifecycle: start_stage() → end_stage() → finalize() | | StageTelemetry | dataclass | Per-stage record | | PipelineTelemetry | dataclass | Full report with warnings | | PerformanceWarning | dataclass | Individual anomaly record |
Other measured skills in the registry, with their headline benchmark lift.