Token Burn

Burn is not proof. Output is not victory. Accepted work that repeats is leverage.

today · in flight

1.8M active

14 sessions · spend 30.5M · Claude Code 1.8M
by project: mainstay-av 1.8M · <home> 0
top session 5af22ad7 (mainstay-av, 1.8M, —)

2026-08-22
as of 2026-08-22T12:55:31Z

Stupid button · 24m ago · gemma4:e4b · 4053t in / 853t out · 29.6s

Trend vs last week

Waste is UP 20.6M (467.9% up) vs last week.
Unsettled is DOWN 12.4M (-73.9% down) vs last week.
Accepted has no prior-window comparison.

What worked

pal: 4.6M
pal/dashboards: 1.7M
las: 1.4M
memory: 606K
pal/templari: 427K

What wasted tokens

mainstay-av: 0.0 (measurable)
<home>: 0.0 (UNMEASURABLE — ran outside any git repo, so 0.0 is the meter failing to see the work, not a judgement on it). The waste volume here is 24.8M.
The largest waste volume is in <home> at 24.8M.
The second largest waste volume is in the overall waste total of 25.0M.

Why the big number is big

A subagent spawn forks its parent, so the child starts holding the parent's whole transcript and the same conversation is charged once per fork, on every turn. The root 019fee2e-1781 involved 13 threads and 12 forks, with a max depth of 2. The total charged is 579.6M, and the cost per output is 315.8. The agents involved are Averroes, Copernicus, Faraday, Galileo, Halley, and Hooke.

What needs a verdict

mainstay-av (5af22ad7, 2026-08-22, mainstay-av/docs/entity): suggested_verdict is review. (open_loop, 1.8M burn — worth looking)
pal (5af22ad7, 2026-08-21, pal): suggested_verdict is review. (open_loop, 1.4M burn — worth looking)
pal (5af22ad7, 2026-08-20, pal): suggested_verdict is not_useful. (open_loop, 476K burn — small miss)
pal (83ba2858, 2026-08-17, pal): suggested_verdict is not_useful. (open_loop, 388K burn — small miss)
mainstay-av (5af22ad7, 2026-08-21, mainstay-av/docs/entity): suggested_verdict is not_useful. (open_loop, 233K burn — small miss)
panacea-ai (83ba2858, 2026-08-16, panacea-ai): suggested_verdict is not_useful. (open_loop, 53K burn — small miss)
dreamweaver (83ba2858, 2026-08-16, dreamweaver): suggested_verdict is not_useful. (open_loop, 41K burn — small miss)
dreamweaver (5af22ad7, 2026-08-19, dreamweaver): suggested_verdict is not_useful. (open_loop, 25K burn — small miss)

One contrarian observation

The <home> project shows a 0.0 acceptance_ratio, but its waste_active_h of "24.8M" is the primary driver of the total waste_active_h of "25.0M".

---

Measurability caveat (generated, not model-authored): <home> (24.8M) ran outside any git repo. There is no diff to credit, so a 0.0 acceptance ratio there is the meter failing to see the work — not a judgement on it. Only ratios on measurable: true projects are verdicts.

Rollups — acceptance lane

Headline is accepted tokens (active lane, manual label wins over git auto-classification). Unsettled = committed-not-merged + open-loop. "Active" excludes cache replay; "spend" includes it.

Last 7 days

8.7M accepted

unsettled 4.4M · acceptance 23%

spend 972.3M · 174 sessions · Codex 23.6M / Claude Code 14.4M

Last 30 days

8.7M accepted

unsettled 182.0M · acceptance 4%

spend 9.41B · 686 sessions · Claude Code 154.8M / Codex 65.2M

Last 90 days

8.7M accepted

unsettled 298.8M · acceptance 3%

spend 12.04B · 805 sessions · Claude Code 222.5M / Codex 114.4M

Daily heatmap — last 60 days

Each cell = one UTC day, color intensity = log10(active tokens). Hover for date + counts.

LessMore· log-scale, exact lane

Top 10 days

Date	Active	Spend	Sessions	Top project	Outcomes
2026-07-28	34.5M	1.78B	47	<home> (29.6M)	?=50
2026-08-01	24.5M	1.57B	10	payday (6.1M)	?=20
2026-07-16	23.3M	797.0M	24	<home> (16.3M)	?=27
2026-07-31	23.1M	2.10B	9	payday (8.1M)	?=34
2026-08-18	22.3M	554.5M	38	<home> (21.6M)	no_repo=38 · accepted=5 · PRs #27
2026-07-21	20.6M	435.9M	3	<home> (9.8M)	?=8
2026-07-30	19.4M	645.3M	26	payday (16.5M)	?=31
2026-08-06	16.5M	592.1M	51	<home> (12.0M)	?=58
2026-07-29	11.2M	118.2M	25	payday (10.8M)	?=27
2026-07-24	10.8M	403.2M	26	las (5.9M)	?=28

By project — last 7 days

Project	Active
<home>	24.8M
pal	6.9M
mainstay-av	2.0M
pal/dashboards	1.7M
las	1.4M
memory	606K
pal/templari	427K
dreamweaver	66K
.claude	60K
panacea-ai	53K
hooks	25K
18	18K

By outcome — last 7 days

accepted 14 #32 (MERGED), #16 (MERGED), #27 (MERGED), #26 (MERGED)

open_loop 8

no_repo 182

Effectiveness — last 30 days

Where delegated work shipped (top) vs. where high burn never crossed an acceptance line (bottom). Bars: accepted · unsettled · waste. Normalized to the widest project in window.

Leverage — most accepted

Project	Accepted	Ratio
pal	4.6M	51%
pal/dashboards	1.7M	100%
las	1.4M	17%
memory	606K	40%
pal/templari	427K	100%
mainstay-av	0	0%

Waste candidates — lowest acceptance ratio (≥100K active)

Project	Accepted	Ratio	Mix
<home>	0	0%
payday	0	0%
dreamweaver	0	0%
panacea-ai	0	0%
pal/placeholders	0	0%
mainstay-av	0	0%

Needs validation — last 14 days (15 unsettled)

Sessions where auto-classification couldn't tell shipped work from waste. Highest-burn first. Manually label these with `label: "used"|"partial"|"redo"|"review"|"not_useful"` in token_burn.jsonl to override the auto verdict.

Date	Project	Session	Active	Auto	cwd
2026-08-12	dreamweaver	d476d902	6.5M	open_loop	.../dreamweaver/.worktrees/deploy-main
2026-08-09	dreamweaver	d476d902	3.4M	open_loop	.../dreamweaver
2026-08-14	dreamweaver	d476d902	2.2M	open_loop	.../dreamweaver/.worktrees/deploy-main
2026-08-10	pal/placeholders	d476d902	1.8M	open_loop	.../pal/placeholders
2026-08-22	mainstay-av	5af22ad7	1.8M	open_loop	.../mainstay-av/docs/entity
2026-08-10	dreamweaver	d476d902	1.6M	open_loop	.../dreamweaver
2026-08-21	pal	5af22ad7	1.4M	open_loop	.../pal
2026-08-12	pal/placeholders	d476d902	1.2M	open_loop	.../pal/placeholders
2026-08-20	pal	5af22ad7	476K	open_loop	.../pal
2026-08-17	pal	83ba2858	388K	open_loop	.../pal
2026-08-21	mainstay-av	5af22ad7	233K	open_loop	.../mainstay-av/docs/entity
2026-08-16	panacea-ai	83ba2858	53K	open_loop	.../panacea-ai
2026-08-16	dreamweaver	83ba2858	41K	open_loop	.../dreamweaver
2026-08-19	dreamweaver	5af22ad7	25K	open_loop	.../dreamweaver
2026-08-15	memory	83ba2858	6K	open_loop	/home/khoward/.claude/projects/-home-khoward/memory

Recent events — last 30 rows

Latest enriched rows from the JSONL (sorted by date desc, active desc as tiebreaker). M = manually labeled.

Date	Tool	Project	Session	Active	Output	Outcome
2026-08-22	CC	mainstay-av	5af22ad7	1.8M	57K	open_loop
2026-08-22	CC	<home>	053bb15b	0	0	no_repo
2026-08-22	CC	<home>	1e0d45ff	0	0	no_repo
2026-08-22	CC	<home>	20a4587d	0	0	no_repo
2026-08-22	CC	<home>	33f690ae	0	0	no_repo
2026-08-22	CC	<home>	6271cf5c	0	0	no_repo
2026-08-22	CC	<home>	750f33b9	0	0	no_repo
2026-08-22	CC	<home>	769a41c8	0	0	no_repo
2026-08-22	CC	<home>	b1809569	0	0	no_repo
2026-08-22	CC	<home>	b9db0b3b	0	0	no_repo
2026-08-22	CC	<home>	c949ef31	0	0	no_repo
2026-08-22	CC	<home>	d7b1ed7b	0	0	no_repo
2026-08-22	CC	<home>	fbb5d5a3	0	0	no_repo
2026-08-21	CC	pal	5af22ad7	1.4M	3K	open_loop
2026-08-21	CC	mainstay-av	5af22ad7	233K	118K	open_loop
2026-08-21	Codex	<home>	01a0220a	0	0	no_repo
2026-08-21	CC	<home>	01439347	0	0	no_repo
2026-08-21	CC	<home>	0b29620d	0	0	no_repo
2026-08-21	CC	<home>	0b9399c2	0	0	no_repo
2026-08-21	CC	<home>	1616a2ae	0	0	no_repo
2026-08-21	CC	<home>	29505b44	0	0	no_repo
2026-08-21	CC	<home>	37561d26	0	0	no_repo
2026-08-21	CC	<home>	44854411	0	0	no_repo
2026-08-21	CC	<home>	4aff64c1	0	0	no_repo
2026-08-21	CC	<home>	512a42fb	0	0	no_repo
2026-08-21	CC	<home>	65745fcb	0	0	no_repo
2026-08-21	CC	<home>	6dee75eb	0	0	no_repo
2026-08-21	CC	<home>	7ee2d8fe	0	0	no_repo
2026-08-21	CC	<home>	86cc9e84	0	0	no_repo
2026-08-21	CC	<home>	8a1048b8	0	0	no_repo

Generated 2026-08-22 12:55 UTC from ~/ideas/pal/dashboards/token_burn.jsonl (989 rows). Sources: ~/.codex/sessions/ (exact), ~/.claude/projects/-home-khoward/ (exact). No estimate lane currently active.