Full System Audit

Direct answer to "is everything functioning independently and called at appropriate times." Two different questions, answered separately, because the honest answer to each is different.

PurposeSeparate "does this module work on its own" from "is this module actually wired into anything" — a real distinction the rest of this site's audit pages don't split out explicitly.

StatusVerified — every module below confirmed to import cleanly and run standalone before being listed

Built fromDirect audit, not assumption

Depends on—

Superseded by—

EvidenceReal, specific drift risk found below — not hypothetical.

Everything below imports cleanly and runs standalone — confirmed, not assumed

multi_stage_pipeline.py, linear_and_four_channel_pipeline_example.py, pipeline_front_gate.py, calibration_forward_walk.py, story_logic_checker.py, red_team_panel.py, epistemic_protocols.py, verification_registry.py, all COMPONENT_A_* files, the ablation harnesses, channel_correlation_diagnostic.py, severity_spectrum_test.py.

The real, honest gap: independent function does not mean connected function

Genuinely orphaned — real, tested, working, but called by nothing else:

calibration_forward_walk.py
story_logic_checker.py
red_team_panel.py
epistemic_protocols.py
verification_registry.py
COMPONENT_A_FIXED_PIPELINE.py — superseded by later fixes, but the later fixes don't call back to it either
COMPONENT_A_NUANCED_ESTIMATORS.py — contains the real Fix 5 formulas, imported by nothing
COMPONENT_A_v2_RECALIBRATED.py
ablation_harness_v2.py
channel_correlation_diagnostic.py
severity_spectrum_test.py

Genuinely wired in:

multi_stage_pipeline.py — used by the alternating pipeline example and the front gate
COMPONENT_A_MIND_DIVERSITY_FIXED.py, COMPONENT_A_TILT_SPECTRUM_FIXED.py — used by ablation_harness_v3, severity tests
ablation_harness_v3.py — used by severity_spectrum_test.py

A real, specific drift risk found during this audit, not hypothetical

pipeline_front_gate.py re-implements the schema_health structural-gap formula as an inline lambda, duplicating — not calling — the real fixed version in COMPONENT_A_NUANCED_ESTIMATORS.py. Two copies of the same fix now exist. If one gets updated and the other doesn't, they silently diverge, with no error raised anywhere. This is exactly the failure mode a real kernel or part registry is supposed to prevent — found by auditing the actual call graph, not by assuming the architecture diagram matched the code.

What this means, directly

Nothing here is broken. Everything listed works on its own terms, verified before being listed. The actual gap is the one this audit was built to find: there is no single decision layer that knows these modules exist and calls the right one at the right time. Right now, using any of the orphaned modules requires a person remembering it exists and writing a new import by hand — which is exactly how the pipeline_front_gate duplication happened in the first place.