Context
Tree line B1 (#14564 map) — Option E, the graduated v1 OPENER: the timed fresh-install run that measures reality BEFORE any self-configuration machinery claims wins. The operator's sellable number (TTFP) starts existing here.
The Problem
Every adoption claim without a baseline is marketing. The harness executes a scripted fresh install against a containerized reference environment and emits the three Ring-2 metrics' raw timings — producing the manual-baseline BEFORE-number on day one (the same-morning #14454 gate-first precedent, applied to install).
The Architectural Reality
Standalone runner (buildScripts/ or ai/scripts class — structural-pre-flight at claim; sibling: the benchmark runners); containerized reference env per the Option-E falsifier (if env-scripting costs rival leaf D, E collapses into an AC — the falsifier stays live and honest); consumes only public install paths (the docs' own steps — the harness IS the docs' lie detector).
The Fix
One PR: the runner (env up → scripted install steps → timestamped phase log → teardown) + the phase schema (provision-start, deps-done, config-done, first-persistence) + one recorded baseline run committed as the before-number.
Acceptance Criteria
Out of Scope
Metric NODE emission (G1) · the other two metrics' probes (B3/B4 lines) · any install-path improvement (that's the point — measure first).
Related
Parent #14564 (line B1) · Option-E authority from #14456 · #14230 contract anchor · feeds G1-G3. Sweeps: 03:47Z clean.
Origin Session ID: fa2a6fd5-7488-4af6-a0d2-3855c86003e4
Retrieval Hint: "fresh install acceptance runner TTFP baseline phases"
Context
Tree line B1 (#14564 map) — Option E, the graduated v1 OPENER: the timed fresh-install run that measures reality BEFORE any self-configuration machinery claims wins. The operator's sellable number (TTFP) starts existing here.
The Problem
Every adoption claim without a baseline is marketing. The harness executes a scripted fresh install against a containerized reference environment and emits the three Ring-2 metrics' raw timings — producing the manual-baseline BEFORE-number on day one (the same-morning #14454 gate-first precedent, applied to install).
The Architectural Reality
Standalone runner (buildScripts/ or ai/scripts class — structural-pre-flight at claim; sibling: the benchmark runners); containerized reference env per the Option-E falsifier (if env-scripting costs rival leaf D, E collapses into an AC — the falsifier stays live and honest); consumes only public install paths (the docs' own steps — the harness IS the docs' lie detector).
The Fix
One PR: the runner (env up → scripted install steps → timestamped phase log → teardown) + the phase schema (provision-start, deps-done, config-done, first-persistence) + one recorded baseline run committed as the before-number.
Acceptance Criteria
Out of Scope
Metric NODE emission (G1) · the other two metrics' probes (B3/B4 lines) · any install-path improvement (that's the point — measure first).
Related
Parent #14564 (line B1) · Option-E authority from #14456 · #14230 contract anchor · feeds G1-G3. Sweeps: 03:47Z clean. Origin Session ID: fa2a6fd5-7488-4af6-a0d2-3855c86003e4 Retrieval Hint: "fresh install acceptance runner TTFP baseline phases"