perfgate-sensor

Cockpit mode integration via sensor.report.v1 envelopes.

CI dashboards and fleet-monitoring tools need a uniform envelope around every tool's output. This crate wraps perfgate's typed receipts into a generic SensorReport that any cockpit system can ingest without knowing anything about performance budgets.

What It Does

A SensorReport envelope contains:

Tool metadata -- name, version
Run metadata -- timestamps, duration, capability flags (baseline available, engine features)
Verdict -- pass/warn/fail with counts
Findings -- individual check results, each with a stable SHA-256 fingerprint
Artifacts -- links to detailed run/compare/report receipts
Data -- opaque summary payload for downstream consumers

Key API

use perfgate_sensor::SensorReportBuilder;

let report = SensorReportBuilder::new(tool_info, started_at)
    .ended_at(ended_at, duration_ms)
    .baseline(true, None)
    .artifact("extras/perfgate.run.v1.json".into(), "run_receipt".into())
    .build(&perfgate_report);

Builder Methods

Method	Purpose
`new(tool, start)`	Create builder with tool info and start time
`ended_at(t, ms)`	Set end time and duration
`baseline(ok, reason)`	Declare baseline availability
`max_findings(n)`	Cap findings (default 100, adds truncation notice)
`artifact(path, type)`	Attach an artifact link
`build(report)`	Single-bench envelope from a `PerfgateReport`
`build_error(msg, stage, code)`	Error envelope when the tool itself fails
`build_aggregated(outcomes)`	Multi-bench envelope from `Vec<BenchOutcome>`

Other Exports

Item	Purpose
`sensor_fingerprint(findings)`	SHA-256 fingerprint over a set of findings
`default_engine_capability()`	Platform-aware capability detection
`BenchOutcome`	Success or Error outcome for aggregation

License