SDD Review-Dispatch — Full Eval Matrix

runs

pass

fail

indeterminate

$633

total est. cost

681M

coding tokens

34.4h

coding wall time

Baseline (dev branch) · 5 runs

When (UTC)	Scenario	Config	Agent	Verdict	Coding min	Tokens	Coding $	Model split	Total $	Run
06-10 00:26	sdd-go-fractals	baseline	claude	pass	65	21.2M	16.07	O$7.68 S$8.39	21.81	2883
06-10 00:26	sdd-rejects-extra-features	baseline	claude	pass	6	1.9M	1.88	O$1.15 S$0.72	2.33	607b
06-10 00:26	sdd-svelte-todo	baseline	claude	indeterminate	6	2.0M	2.14	O$1.40 S$0.74	2.75	9297
06-10 00:33	spec-reviewer-catches-planted-flaws	baseline	claude	pass	1	0.2M	0.50	O$0.50	0.63	4b11
06-10 00:34	sdd-svelte-todo	baseline	claude	pass	80	27.3M	20.98	O$11.14 S$9.83	29.35	e69c
subtotal	157	53M	41.56		56.87

When (UTC)

Scenario

Config

Agent

Verdict

Coding min

Tokens

Coding $

Model split

Total $

Run

06-10 00:26

sdd-go-fractals

baseline

claude

pass

21.2M

16.07

O$7.68 S$8.39

21.81

2883

06-10 00:26

sdd-rejects-extra-features

baseline

claude

pass

1.9M

1.88

O$1.15 S$0.72

2.33

607b

06-10 00:26

sdd-svelte-todo

baseline

claude

indeterminate

2.0M

2.14

O$1.40 S$0.74

2.75

9297

06-10 00:33

spec-reviewer-catches-planted-flaws

baseline

claude

pass

0.2M

0.50

O$0.50

0.63

4b11

06-10 00:34

sdd-svelte-todo

baseline

claude

pass

27.3M

20.98

O$11.14 S$9.83

29.35

e69c

subtotal

157

53M

41.56

56.87

Design iterations 1-3 (Jun 9) · 20 runs

When (UTC)	Scenario	Config	Agent	Verdict	Coding min	Tokens	Coding $	Model split	Total $	Run
06-10 00:33	sdd-rejects-extra-features	branch (iter. state at run time)	claude	pass	6	3.4M	2.04	O$1.70 H$0.33	2.46	52b8
06-10 00:34	sdd-go-fractals	branch (iter. state at run time)	claude	indeterminate	30	7.6M	7.90	O$4.93 S$2.97	9.99	84d1
06-10 00:40	sdd-svelte-todo	branch (iter. state at run time)	claude	indeterminate	29	9.1M	7.40	O$3.99 S$3.25 H$0.17	9.88	2b6f
06-10 01:04	spec-reviewer-catches-planted-flaws	branch (iter. state at run time)	claude	pass	1	0.2M	0.44	O$0.44	0.60	0aa2
06-10 01:05	sdd-go-fractals	branch (iter. state at run time)	claude	pass	43	14.5M	15.61	O$13.44 S$2.17	18.50	c7b3
06-10 01:06	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	indeterminate	9	2.5M	2.51	O$1.52 S$0.99	3.45	47df
06-10 01:10	sdd-svelte-todo	branch (iter. state at run time)	claude	indeterminate	5	1.8M	2.73	O$2.73	3.37	1629
06-10 01:48	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	pass	8	3.6M	2.62	O$2.24 H$0.37	3.31	0c68
06-10 01:57	sdd-svelte-todo	branch (iter. state at run time)	claude	indeterminate	47	15.6M	12.60	O$6.49 S$6.12	20.86	4f68
06-10 02:46	sdd-svelte-todo	branch (iter. state at run time)	claude	indeterminate	8	2.6M	2.61	O$1.62 S$0.99	3.95	1c6f
06-10 02:55	sdd-svelte-todo	branch (iter. state at run time)	claude	pass	55	17.2M	22.18	O$22.18	23.97	9af1
06-10 04:19	sdd-go-fractals	branch (iter. state at run time)	claude	pass	70	32.2M	17.14	O$9.29 S$5.89 H$1.96	18.95	ad31
06-10 04:19	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	fail	12	3.1M	3.31	O$2.07 S$1.24	3.77	9a13
06-10 04:19	sdd-svelte-todo	branch (iter. state at run time)	claude	pass	83	29.2M	22.00	O$13.82 S$7.53 H$0.66	23.52	4033
06-10 04:49	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	indeterminate	12	3.8M	2.85	O$2.04 S$0.47 H$0.34	3.51	bc1c
06-10 05:43	sdd-go-fractals	branch (iter. state at run time)	claude	pass	68	22.9M	17.59	O$9.58 S$8.01	18.74	e161
06-10 06:58	sdd-go-fractals	branch (iter. state at run time)	claude	pass	48	15.7M	13.55	O$9.24 S$4.31	14.48	4d41
06-10 07:53	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	fail	8	2.1M	2.32	O$1.74 S$0.58	2.58	445a
06-10 07:53	sdd-rejects-extra-features	branch (iter. state at run time)	claude	pass	5	2.1M	1.37	O$1.12 H$0.25	1.64	3a9d
06-10 07:53	sdd-svelte-todo	branch (iter. state at run time)	claude	pass	73	25.4M	19.42	O$11.96 S$7.46	20.79	ecfc
subtotal	619	215M	178.20		208.30

When (UTC)

Scenario

Config

Agent

Verdict

Coding min

Tokens

Coding $

Model split

Total $

Run

06-10 00:33

sdd-rejects-extra-features

branch (iter. state at run time)

claude

pass

3.4M

2.04

O$1.70 H$0.33

2.46

52b8

06-10 00:34

sdd-go-fractals

branch (iter. state at run time)

claude

indeterminate

7.6M

7.90

O$4.93 S$2.97

9.99

84d1

06-10 00:40

sdd-svelte-todo

branch (iter. state at run time)

claude

indeterminate

9.1M

7.40

O$3.99 S$3.25 H$0.17

9.88

2b6f

06-10 01:04

spec-reviewer-catches-planted-flaws

branch (iter. state at run time)

claude

pass

0.2M

0.44

O$0.44

0.60

0aa2

06-10 01:05

sdd-go-fractals

branch (iter. state at run time)

claude

pass

14.5M

15.61

O$13.44 S$2.17

18.50

c7b3

06-10 01:06

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

indeterminate

2.5M

2.51

O$1.52 S$0.99

3.45

47df

06-10 01:10

sdd-svelte-todo

branch (iter. state at run time)

claude

indeterminate

1.8M

2.73

O$2.73

3.37

1629

06-10 01:48

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

pass

3.6M

2.62

O$2.24 H$0.37

3.31

0c68

06-10 01:57

sdd-svelte-todo

branch (iter. state at run time)

claude

indeterminate

15.6M

12.60

O$6.49 S$6.12

20.86

4f68

06-10 02:46

sdd-svelte-todo

branch (iter. state at run time)

claude

indeterminate

2.6M

2.61

O$1.62 S$0.99

3.95

1c6f

06-10 02:55

sdd-svelte-todo

branch (iter. state at run time)

claude

pass

17.2M

22.18

O$22.18

23.97

9af1

06-10 04:19

sdd-go-fractals

branch (iter. state at run time)

claude

pass

32.2M

17.14

O$9.29 S$5.89 H$1.96

18.95

ad31

06-10 04:19

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

fail

3.1M

3.31

O$2.07 S$1.24

3.77

9a13

06-10 04:19

sdd-svelte-todo

branch (iter. state at run time)

claude

pass

29.2M

22.00

O$13.82 S$7.53 H$0.66

23.52

4033

06-10 04:49

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

indeterminate

3.8M

2.85

O$2.04 S$0.47 H$0.34

3.51

bc1c

06-10 05:43

sdd-go-fractals

branch (iter. state at run time)

claude

pass

22.9M

17.59

O$9.58 S$8.01

18.74

e161

06-10 06:58

sdd-go-fractals

branch (iter. state at run time)

claude

pass

15.7M

13.55

O$9.24 S$4.31

14.48

4d41

06-10 07:53

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

fail

2.1M

2.32

O$1.74 S$0.58

2.58

445a

06-10 07:53

sdd-rejects-extra-features

branch (iter. state at run time)

claude

pass

2.1M

1.37

O$1.12 H$0.25

1.64

3a9d

06-10 07:53

sdd-svelte-todo

branch (iter. state at run time)

claude

pass

25.4M

19.42

O$11.96 S$7.46

20.79

ecfc

subtotal

619

215M

178.20

208.30

Optimization experiments + gates (Jun 10) · 21 runs

When (UTC)	Scenario	Config	Agent	Verdict	Coding min	Tokens	Coding $	Model split	Total $	Run
06-10 08:00	spec-reviewer-catches-planted-flaws	branch (iter. state at run time)	claude	pass	2	0.2M	0.50	O$0.50	0.61	75b9
06-10 09:09	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	fail	8	2.3M	2.34	O$1.80 S$0.55	2.69	c127
06-10 09:20	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	fail	8	2.5M	2.59	O$2.05 S$0.54	2.84	8817
06-10 09:35	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	pass	11	2.6M	2.14	O$1.67 S$0.26 H$0.21	2.58	719d
06-10 09:35	sdd-svelte-todo	branch (iter. state at run time)	claude	pass	66	22.4M	18.16	O$11.84 S$6.32	19.60	be5b
06-10 10:44	sdd-go-fractals	branch (iter. state at run time)	claude	pass	44	13.4M	11.67	O$7.13 S$4.54	12.57	a8d1
06-10 10:44	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	fail	7	1.6M	1.73	O$1.17 S$0.55	2.16	b6ab
06-10 10:44	sdd-svelte-todo	branch (iter. state at run time)	claude	pass	63	19.7M	15.76	O$8.61 S$7.09 H$0.06	17.06	9fb3
06-10 10:53	sdd-rejects-extra-features	branch (iter. state at run time)	claude	pass	5	2.0M	1.31	O$1.09 H$0.23	1.51	8bf5
06-10 11:00	spec-reviewer-catches-planted-flaws	branch (iter. state at run time)	claude	pass	1	0.2M	0.50	O$0.50	0.62	29f1
06-10 11:50	sdd-quality-reviewer-catches-planted-defect	branch (iter. state at run time)	claude	pass	13	3.2M	3.12	O$2.03 S$1.09	3.50	d8c1
06-10 16:17	sdd-go-fractals	branch (iter. state at run time)	claude	pass	57	20.0M	14.63	O$7.87 S$6.53 H$0.23	15.81	309b
06-10 16:39	sdd-go-fractals	lean-controller (E)	claude	pass	50	13.7M	11.73	O$5.78 S$5.96	12.94	9a73
06-10 16:39	sdd-go-fractals	file-handoffs (E)	claude	pass	47	14.9M	17.43	O$16.18 S$1.25	18.47	a65c
06-10 16:52	sdd-go-fractals	durable-progress (E)	claude	pass	48	15.8M	13.37	O$8.85 S$4.43 H$0.09	14.45	b260
06-10 17:17	sdd-go-fractals	pipelined-review (E)	claude	pass	49	14.1M	13.43	O$8.27 S$5.16	14.68	9d88
06-10 17:44	sdd-go-fractals	combo	claude	pass	55	14.4M	12.81	O$7.41 S$5.40	13.96	98b7
06-10 18:09	sdd-go-fractals	pipelined-review (E)	claude	pass	54	17.8M	14.84	O$8.75 S$6.08	16.08	310e
06-10 19:10	sdd-go-fractals	combo	claude	pass	54	16.6M	14.31	O$8.67 S$5.64	15.51	941a
06-10 19:10	sdd-svelte-todo	combo	claude	pass	55	19.3M	14.99	O$7.76 S$7.23	16.50	2824
06-10 19:10	sdd-quality-reviewer-catches-planted-defect	combo	claude	pass	10	2.9M	2.77	O$1.77 S$0.99	3.14	2ad0
subtotal	708	220M	190.12		207.28

When (UTC)

Scenario

Config

Agent

Verdict

Coding min

Tokens

Coding $

Model split

Total $

Run

06-10 08:00

spec-reviewer-catches-planted-flaws

branch (iter. state at run time)

claude

pass

0.2M

0.50

O$0.50

0.61

75b9

06-10 09:09

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

fail

2.3M

2.34

O$1.80 S$0.55

2.69

c127

06-10 09:20

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

fail

2.5M

2.59

O$2.05 S$0.54

2.84

8817

06-10 09:35

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

pass

2.6M

2.14

O$1.67 S$0.26 H$0.21

2.58

719d

06-10 09:35

sdd-svelte-todo

branch (iter. state at run time)

claude

pass

22.4M

18.16

O$11.84 S$6.32

19.60

be5b

06-10 10:44

sdd-go-fractals

branch (iter. state at run time)

claude

pass

13.4M

11.67

O$7.13 S$4.54

12.57

a8d1

06-10 10:44

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

fail

1.6M

1.73

O$1.17 S$0.55

2.16

b6ab

06-10 10:44

sdd-svelte-todo

branch (iter. state at run time)

claude

pass

19.7M

15.76

O$8.61 S$7.09 H$0.06

17.06

9fb3

06-10 10:53

sdd-rejects-extra-features

branch (iter. state at run time)

claude

pass

2.0M

1.31

O$1.09 H$0.23

1.51

8bf5

06-10 11:00

spec-reviewer-catches-planted-flaws

branch (iter. state at run time)

claude

pass

0.2M

0.50

O$0.50

0.62

29f1

06-10 11:50

sdd-quality-reviewer-catches-planted-defect

branch (iter. state at run time)

claude

pass

3.2M

3.12

O$2.03 S$1.09

3.50

d8c1

06-10 16:17

sdd-go-fractals

branch (iter. state at run time)

claude

pass

20.0M

14.63

O$7.87 S$6.53 H$0.23

15.81

309b

06-10 16:39

sdd-go-fractals

lean-controller (E)

claude

pass

13.7M

11.73

O$5.78 S$5.96

12.94

9a73

06-10 16:39

sdd-go-fractals

file-handoffs (E)

claude

pass

14.9M

17.43

O$16.18 S$1.25

18.47

a65c

06-10 16:52

sdd-go-fractals

durable-progress (E)

claude

pass

15.8M

13.37

O$8.85 S$4.43 H$0.09

14.45

b260

06-10 17:17

sdd-go-fractals

pipelined-review (E)

claude

pass

14.1M

13.43

O$8.27 S$5.16

14.68

9d88

06-10 17:44

sdd-go-fractals

combo

claude

pass

14.4M

12.81

O$7.41 S$5.40

13.96

98b7

06-10 18:09

sdd-go-fractals

pipelined-review (E)

claude

pass

17.8M

14.84

O$8.75 S$6.08

16.08

310e

06-10 19:10

sdd-go-fractals

combo

claude

pass

16.6M

14.31

O$8.67 S$5.64

15.51

941a

06-10 19:10

sdd-svelte-todo

combo

claude

pass

19.3M

14.99

O$7.76 S$7.23

16.50

2824

06-10 19:10

sdd-quality-reviewer-catches-planted-defect

combo

claude

pass

2.9M

2.77

O$1.77 S$0.99

3.14

2ad0

subtotal

708

220M

190.12

207.28

Batch A-E: strict-cost ladder probes (Jun 10/11 overnight) · 17 runs

A: lean-vs-combo cost-neutral → briefs stay for fidelity, not dollars. B: crisp 7-task plan −21% dispatches, $9.51–12.65. C: sonnet controller $6.68/$8.05 (~−40%), judgment clean, escalation unstressed. D: haiku reviewers DEAD — 0/10 planted defects cleanly flagged (they advocate for defects). E: svelte range $14.99–20.30. Up to 7-way parallel: trust $/tokens, not wall-clock.

When (UTC)	Scenario	Config	Agent	Verdict	Coding min	Tokens	Coding $	Model split	Total $	Run
06-10 22:06	sdd-go-fractals	lean (no briefs)	claude	pass	51	17.0M	13.25	O$7.30 S$5.95	14.72	e478
06-10 22:06	sdd-go-fractals	lean (no briefs)	claude	pass	43	13.8M	11.61	O$7.17 S$4.44	12.57	ee4d
06-10 22:06	sdd-go-fractals	lean (no briefs)	claude	pass	49	15.8M	12.27	O$6.46 S$5.81	13.41	751f
06-10 22:06	sdd-go-fractals	combo	claude	pass	60	18.4M	14.10	O$6.95 S$7.15	15.51	d170
06-10 22:07	sdd-go-fractals-crisp	combo	claude	pass	51	16.8M	12.65	O$6.33 S$6.22 H$0.11	13.67	f21c
06-10 22:07	sdd-go-fractals-crisp	combo	claude	pass	38	12.1M	9.51	O$4.93 S$4.50 H$0.08	10.30	45f0
06-10 22:07	sdd-svelte-todo	combo	claude	pass	70	24.1M	20.30	O$12.35 S$7.94	22.36	c47f
06-10 22:47	sdd-go-fractals	combo	claude-sonnet	indeterminate	0	—	0.00		0.13	48a0
06-10 22:54	sdd-go-fractals	combo	claude-sonnet	pass	31	12.6M	6.68	S$6.41 H$0.27	7.74	5038
06-10 22:59	sdd-go-fractals-crisp	combo	claude	pass	50	17.1M	12.65	O$6.70 S$5.88 H$0.06	13.68	15a0
06-10 23:27	sdd-go-fractals	combo	claude-sonnet	pass	41	16.1M	8.05	S$7.48 H$0.57	9.46	2175
06-10 23:28	sdd-quality-reviewer-catches-planted-defect	haiku reviewers	claude	indeterminate	12	3.4M	3.12	O$2.50 S$0.49 H$0.13	3.70	11cf
06-10 23:28	sdd-quality-reviewer-catches-planted-defect	haiku reviewers	claude	fail	13	3.8M	3.11	O$2.37 S$0.55 H$0.19	3.88	ed05
06-10 23:28	sdd-quality-reviewer-catches-planted-defect	haiku reviewers	claude	pass	11	2.9M	2.62	O$1.95 S$0.55 H$0.12	3.13	be87
06-10 23:44	sdd-quality-reviewer-catches-planted-defect	haiku reviewers	claude	fail	12	3.4M	2.76	O$2.25 S$0.34 H$0.18	3.34	6a71
06-10 23:44	sdd-quality-reviewer-catches-planted-defect	haiku reviewers	claude	pass	9	2.5M	2.31	O$1.72 S$0.47 H$0.12	2.59	4241
06-10 23:58	sdd-go-fractals	haiku reviewers	claude	pass	39	14.1M	9.61	O$5.09 S$4.10 H$0.43	10.55	0297
subtotal	578	194M	144.59		160.73

When (UTC)

Scenario

Config

Agent

Verdict

Coding min

Tokens

Coding $

Model split

Total $

Run

06-10 22:06

sdd-go-fractals

lean (no briefs)

claude

pass

17.0M

13.25

O$7.30 S$5.95

14.72

e478

06-10 22:06

sdd-go-fractals

lean (no briefs)

claude

pass

13.8M

11.61

O$7.17 S$4.44

12.57

ee4d

06-10 22:06

sdd-go-fractals

lean (no briefs)

claude

pass

15.8M

12.27

O$6.46 S$5.81

13.41

751f

06-10 22:06

sdd-go-fractals

combo

claude

pass

18.4M

14.10

O$6.95 S$7.15

15.51

d170

06-10 22:07

sdd-go-fractals-crisp

combo

claude

pass

16.8M

12.65

O$6.33 S$6.22 H$0.11

13.67

f21c

06-10 22:07

sdd-go-fractals-crisp

combo

claude

pass

12.1M

9.51

O$4.93 S$4.50 H$0.08

10.30

45f0

06-10 22:07

sdd-svelte-todo

combo

claude

pass

24.1M

20.30

O$12.35 S$7.94

22.36

c47f

06-10 22:47

sdd-go-fractals

combo

claude-sonnet

indeterminate

—

0.00

0.13

48a0

06-10 22:54

sdd-go-fractals

combo

claude-sonnet

pass

12.6M

6.68

S$6.41 H$0.27

7.74

5038

06-10 22:59

sdd-go-fractals-crisp

combo

claude

pass

17.1M

12.65

O$6.70 S$5.88 H$0.06

13.68

15a0

06-10 23:27

sdd-go-fractals

combo

claude-sonnet

pass

16.1M

8.05

S$7.48 H$0.57

9.46

2175

06-10 23:28

sdd-quality-reviewer-catches-planted-defect

haiku reviewers

claude

indeterminate

3.4M

3.12

O$2.50 S$0.49 H$0.13

3.70

11cf

06-10 23:28

sdd-quality-reviewer-catches-planted-defect

haiku reviewers

claude

fail

3.8M

3.11

O$2.37 S$0.55 H$0.19

3.88

ed05

06-10 23:28

sdd-quality-reviewer-catches-planted-defect

haiku reviewers

claude

pass

2.9M

2.62

O$1.95 S$0.55 H$0.12

3.13

be87

06-10 23:44

sdd-quality-reviewer-catches-planted-defect

haiku reviewers

claude

fail

3.4M

2.76

O$2.25 S$0.34 H$0.18

3.34

6a71

06-10 23:44

sdd-quality-reviewer-catches-planted-defect

haiku reviewers

claude

pass

2.5M

2.31

O$1.72 S$0.47 H$0.12

2.59

4241

06-10 23:58

sdd-go-fractals

haiku reviewers

claude

pass

14.1M

9.61

O$5.09 S$4.10 H$0.43

10.55

0297

subtotal

578

194M

144.59

160.73

Coding figures are the agent under test; Total adds the Gauntlet QA agent. Costs are run-time estimates from verdict.json economics. Wall-clock in Batch A-E ran under heavy parallel load.