UofA Adversarial Coverage Report

HIT HIT+ MISS WRONG CLEAN-CORRECT CLEAN-WRONG GEN-INVALID

View 1 — Catalog self-coverage (UofA pattern × subtlety)

COU-dependent rules: No rules show COU-dependent firing behavior at the 30% disparity threshold.

Pattern	low	medium	high	per-COU breakdown
COMPOUND-01	—	—	0% (0/20)	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a
COMPOUND-03	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-AL-01	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-AL-02	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-AR-01	—	—	95% (19/20)	show Morrison COU1: 95% Morrison COU2: n/a Nagaraja COU1: n/a
W-AR-02	—	—	100% (19/19)	show Morrison COU1: 95% Morrison COU2: n/a Nagaraja COU1: n/a
W-AR-03	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-AR-04	—	—	0% (0/20)	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a
W-CON-01	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-CON-02	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-CON-03	—	—	0% (0/20)	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a
W-CON-04	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-CON-05	—	—	100% (19/19)	show Morrison COU1: 95% Morrison COU2: n/a Nagaraja COU1: n/a
W-EP-01	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-EP-02	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-EP-03	—	—	0% (0/20)	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a
W-EP-04	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-ON-01	—	—	not measurable	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a
W-ON-02	—	—	100% (20/20)	show Morrison COU1: 100% Morrison COU2: n/a Nagaraja COU1: n/a
W-PROV-01	—	—	55% (11/20)	show Morrison COU1: 55% Morrison COU2: n/a Nagaraja COU1: n/a
W-SI-01	—	—	not measurable	show Morrison COU1: 0% Morrison COU2: n/a Nagaraja COU1: n/a

View 2 — Literature coverage (gap_probe outcomes by taxonomy)

Source taxonomy	WRONG	Verdict
clarissa-machinery/workflow/eliminative-argumentation	5	alt rule fires
clarissa-machinery/workflow/residual-risk-justification	5	alt rule fires
clarissa-machinery/workflow/theory-preconditions	5	alt rule fires
gohar/contextual/configuration	5	alt rule fires
gohar/contextual/environmental-factors	5	alt rule fires
gohar/contextual/faults-physical	5	alt rule fires
gohar/contextual/faults-software	5	alt rule fires

View 3 — Precision / recall summary

Metric	Value	Source battery
Catalog recall (HIT + HIT+)	76.2%	confirm_existing (n=378)
Catalog precision (1 − FPR)	87.5%	negative_controls (n=16)
Gap-probe MISS rate	0.0%	gap_probe (n=35)

Outcome class counts

Class	Count
COV-CLEAN-CORRECT	14
COV-CLEAN-WRONG	2
COV-HIT-PLUS	297
COV-WRONG	125
GEN-INVALID	45

View 4 — Per-rule precision (M5-B)

Decomposes the catalog's package-level precision (= 0% on M5) into per-rule contributors. NC denominator: 16 evaluable negative_control rows. Sorted by NC FPR descending.

Rule	NC FPR	NC firings	Total firings	Targeted firings	Precision when fires
`W-CON-01`	12.5%	2	42	20	47.6%
`COMPOUND-01`	0.0%	0	95	0	0.0%
`COMPOUND-03`	0.0%	0	86	20	23.3%
`W-AL-01`	0.0%	0	422	20	4.7%
`W-AL-02`	0.0%	0	21	20	95.2%
`W-AR-01`	0.0%	0	19	19	100.0%
`W-AR-02`	0.0%	0	44	19	43.2%
`W-AR-03`	0.0%	0	20	20	100.0%
`W-AR-05`	0.0%	0	413	0	0.0%
`W-CON-02`	0.0%	0	20	20	100.0%
`W-CON-04`	0.0%	0	41	20	48.8%
`W-CON-05`	0.0%	0	19	19	100.0%
`W-EP-01`	0.0%	0	21	20	95.2%
`W-EP-02`	0.0%	0	351	20	5.7%
`W-EP-04`	0.0%	0	20	20	100.0%
`W-ON-02`	0.0%	0	422	20	4.7%
`W-PROV-01`	0.0%	0	11	11	100.0%

Performance characterization (v1.8 D2)

Hardware: hardware spec not configured (set UOFA_HW_SPEC env var)

Metric	Value (ms)
Mean total_eval_ms	2613
Median total_eval_ms	2625
p95 total_eval_ms	2701
Sample size	438

Slowest 20 packages

spec_id	variant	total_eval_ms
adv-2026-p2-121-residual-risk-justification	4	2746
adv-2026-p2-121-residual-risk-justification	2	2744
adv-2026-p2-121-residual-risk-justification	3	2743
adv-2026-p2-021-compound-01	1	2739
adv-2026-p2-122-theory-preconditions	1	2733
adv-2026-p2-121-residual-risk-justification	5	2728
adv-2026-p2-021-compound-01	2	2723
adv-2026-p2-021-compound-01	3	2723
adv-2026-p2-201-nc-clean-full-morrison-cou1	1	2723
adv-2026-p2-021-compound-01	5	2713
adv-2026-p2-005-w-ep-01	8	2711
adv-2026-p2-020-w-prov-01	9	2711
adv-2026-p2-005-w-ep-01	11	2709
adv-2026-p2-209-nc-clean-low-confidence-but-documented	2	2709
adv-2026-p2-005-w-ep-01	10	2708
adv-2026-p2-005-w-ep-01	9	2705
adv-2026-p2-020-w-prov-01	8	2703
adv-2026-p2-209-nc-clean-low-confidence-but-documented	1	2702
adv-2026-p2-116-environmental-factors	5	2702
adv-2026-p2-112-faults-physical	4	2702

n = 483 package rows.