Token Burn

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

today · in flight

4.1M active

15 sessions · spend 38.6M · Claude Code 4.0M / Codex 93K
by project: las 4.0M · pal 93K · <home> 0
top session 4b15a997 (las, 4.0M, —)

2026-09-08
as of 2026-09-08T12:55:09Z

Stupid button · 23m ago · gemma4:e4b · 3655t in / 810t out · 143.6s

Trend vs last week

Accepted tokens are DOWN 9.4M (-73.5%) vs last week.
Unsettled tokens are UP 4.9M (204.5%) vs last week.
Waste tokens are DOWN 855K (-45.8%) vs last week.

What worked

las accounted for 2.3M accepted tokens.
memory accounted for 1.1M accepted tokens.

What wasted tokens

pal has an acceptance ratio of 0.0.
The <home> directory has a waste volume of 1.0M. Its acceptance_note is "UNMEASURABLE — ran outside any git repo, so 0.0 is the meter failing to see the work, not a judgement on it."

Why the big number is big

The root 01a04b01-5edc involved 6 threads, 3 forks, and a max_depth of 1. It charged 33.8M, with a replay_pct of 94.3% and a cost_per_output of 273.5. The agents involved were Chandrasekhar, Faraday, Harvey, James, and Pascal. 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.

What needs a verdict

las + 4b15a997 + 2026-09-08 + las/.worktrees/mvp: suggested_verdict is "review" (PR open — work in flight).
las + 4b15a997 + 2026-09-06 + las/.worktrees/mvp: suggested_verdict is "review" (PR open — work in flight).
pal + 01a07347 + 2026-09-08 + pal: suggested_verdict is "review" (open_loop, 872K burn — worth looking).
las + 4b15a997 + 2026-09-05 + las/.worktrees/mvp: suggested_verdict is "review" (open_loop, 645K burn — worth looking).
pal + 01a07a8f + 2026-09-07 + pal: suggested_verdict is "not_useful" (open_loop, 192K burn — small miss).
pal + 01a07347 + 2026-09-07 + pal: suggested_verdict is "not_useful" (open_loop, 191K burn — small miss).
las + 4b15a997 + 2026-09-03 + las/.worktrees/mvp: suggested_verdict is "not_useful" (open_loop, 165K burn — small miss).
pal + 01a07ec0 + 2026-09-08 + pal: suggested_verdict is "not_useful" (open_loop, 106K burn — small miss).

One contrarian observation

The las project has an acceptance_ratio of 0.286, which is significantly lower than the 1.0 ratio achieved by memory, despite las having a much larger accepted_active_h at 2.3M.

---

Measurability caveat (generated, not model-authored): <home> (1.0M) 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

3.4M accepted

unsettled 7.3M · acceptance 29%

spend 180.2M · 169 sessions · Claude Code 9.2M / Codex 2.5M

Last 30 days

16.2M accepted

unsettled 67.8M · acceptance 19%

spend 2.05B · 746 sessions · Claude Code 58.0M / Codex 28.8M

Last 90 days

16.2M accepted

unsettled 164.9M · acceptance 9%

spend 5.33B · 949 sessions · Codex 98.5M / Claude Code 85.5M

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	29.1M	1.09B	22	<home> (29.1M)	?=22
2026-08-18	22.3M	554.5M	38	<home> (21.6M)	?=43
2026-08-06	16.5M	592.1M	46	<home> (12.0M)	?=53
2026-07-16	15.7M	522.1M	23	<home> (15.7M)	?=23
2026-08-12	7.7M	116.0M	25	dreamweaver (6.5M)	?=27
2026-08-08	6.1M	358.6M	27	dreamweaver (3.0M)	?=37
2026-08-07	6.0M	352.8M	26	<home> (4.1M)	?=33
2026-08-31	5.9M	144.8M	26	las (5.8M)	no_repo=24 · open_loop=3 · accepted=2 · PRs #18
2026-06-15	5.9M	132.6M	9	<home> (5.9M)	?=9
2026-08-19	5.6M	205.7M	27	pal (2.3M)	?=36

By project — last 7 days

Project	Active
las	8.0M
pal	1.5M
memory	1.1M
<home>	1.0M
mainstay-av	30K

By outcome — last 7 days

accepted 3 #17 (MERGED)

committed 2 #21 (OPEN)

open_loop 11

no_repo 164

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
las	15.0M	66%
memory	1.1M	64%
pal	0	0%
mainstay-av	0	0%
pal/templari	0	0%
Downloads	0	0%

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

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

Needs validation — last 14 days (21 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
2026-09-08	las	4b15a997	4.0M	committed · #21 (OPEN)	.../las/.worktrees/mvp
2026-08-29	mainstay-av	5af22ad7	1.6M	open_loop	.../mainstay-av/docs/entity
2026-09-06	las	4b15a997	913K	committed · #21 (OPEN)	.../las/.worktrees/mvp
2026-09-08	pal	01a07347	872K	open_loop	.../pal
2026-08-28	las	d33e7f08	689K	open_loop	.../las
2026-09-05	las	4b15a997	645K	open_loop	.../las/.worktrees/mvp
2026-09-07	pal	01a07a8f	192K	open_loop	.../pal
2026-09-07	pal	01a07347	191K	open_loop	.../pal
2026-09-03	las	4b15a997	165K	open_loop	.../las/.worktrees/mvp
2026-09-08	pal	01a07ec0	106K	open_loop	.../pal
2026-09-08	pal	01a07fb1	93K	open_loop	.../pal
2026-08-31	pal/templari	d33e7f08	61K	open_loop	.../pal/templari
2026-09-05	pal	01a07347	57K	open_loop	.../pal
2026-09-04	mainstay-av	d33e7f08	30K	open_loop	.../mainstay-av
2026-09-01	pal	5af22ad7	24K	open_loop	.../pal
2026-08-31	pal	5af22ad7	23K	open_loop	.../pal
2026-09-04	pal	d33e7f08	17K	open_loop	.../pal
2026-09-05	pal	01a0728d	13K	open_loop	.../pal
2026-08-31	pal	d33e7f08	10K	open_loop	.../pal
2026-08-28	pal/templari	d33e7f08	4K	open_loop	.../pal/templari
2026-08-28	pal	d33e7f08	2K	open_loop	.../pal

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
2026-09-08	CC	las	4b15a997	4.0M	113K	committed #21
2026-09-08	Codex	pal	01a07347	872K	43K	open_loop
2026-09-08	Codex	pal	01a07ec0	106K	4K	open_loop
2026-09-08	Codex	pal	01a07fb1	93K	865	open_loop
2026-09-08	CC	<home>	1b600c9e	0	0	no_repo
2026-09-08	CC	<home>	31fd65db	0	0	no_repo
2026-09-08	CC	<home>	63b12416	0	0	no_repo
2026-09-08	CC	<home>	66a9037c	0	0	no_repo
2026-09-08	CC	<home>	73db550e	0	0	no_repo
2026-09-08	CC	<home>	76a995e9	0	0	no_repo
2026-09-08	CC	<home>	a3c68085	0	0	no_repo
2026-09-08	CC	<home>	a5e0acd2	0	0	no_repo
2026-09-08	CC	<home>	b2986c6d	0	0	no_repo
2026-09-08	CC	<home>	c42d45bb	0	0	no_repo
2026-09-08	CC	<home>	d47b1461	0	0	no_repo
2026-09-08	CC	<home>	ecf194d5	0	0	no_repo
2026-09-07	Codex	pal	01a07a8f	192K	18K	open_loop
2026-09-07	Codex	pal	01a07347	191K	18K	open_loop
2026-09-07	CC	<home>	0a8c5706	0	0	no_repo
2026-09-07	CC	<home>	10ec3bf6	0	0	no_repo
2026-09-07	CC	<home>	1a47f770	0	0	no_repo
2026-09-07	CC	<home>	22fb4295	0	0	no_repo
2026-09-07	CC	<home>	242c0df7	0	0	no_repo
2026-09-07	CC	<home>	2b28801c	0	0	no_repo
2026-09-07	CC	<home>	32bd5352	0	0	no_repo
2026-09-07	CC	<home>	39210657	0	0	no_repo
2026-09-07	CC	<home>	393bec4c	0	0	no_repo
2026-09-07	CC	<home>	3f11a4cc	0	0	no_repo
2026-09-07	CC	<home>	446eb246	0	0	no_repo
2026-09-07	CC	<home>	52548297	0	0	no_repo

Generated 2026-09-08 12:55 UTC from ~/ideas/pal/dashboards/token_burn.jsonl (1081 rows). Sources: ~/.codex/sessions/ (exact), ~/.claude/projects/-home-khoward/ (exact). No estimate lane currently active.