← charleslikesdata.com afk cockpit
6 repos · 848 runs · 681 issues · 595 merged · data through Jun 7 – Jul 26, 2026

Autonomous-PR outcomes across the fleet. Cost is API-equivalent (notional on a Pro subscription). Quarantines are a safety signal — afk refused to ship — not failures.

Fleet state · not yet wired up
Live gate/promote/merge and dependency-frontier state isn't ingested yet — this panel is a placeholder for a future feature, not a fabricated number.
98%
landed rate
n = 605 gate-passed
98%
landed rate (executor repos)
n = 6 exec repos
$3.84
API-equiv $ / merged PR
n = 595 merged
fully loaded: includes failed runs, quarantines, and scout overhead
37%
quarantine (safety) rate
n = 681 issues
19%
escalation rate
n = 848 runs
0%
rejected rate
n = 1,144 proposals
Pipeline

Stage funnel

proposed
1,144
promoted
706
62% of proposed
attempted
681
96% of promoted
passed gate
605
89% of attempted
published
596
99% of passed gate
merged
595
100% of published
By repository

Outcomes per repo

landed quarantined other
Orbit-Wars
55 · 95%
afk-agent-system
353 · 89%
afk-cockpit
106 · 94%
household-bms
48 · 96%
the-vault
11 · 0%
the-workshop
108 · 78%
Producers

Proposal quality

dropped
194
harvest
5
mis-sized
40
pending
197
rejected
1
shipped
706
superseded
1
Safety axis

Quarantine by category

250 total
other
80
no-op
40
review
21
capability
19
integration-conflict
17
build-crash
16
setup-crash
10
scope
8
circuit-breaker
7
deny-surface
6
test-conflict
6
timeout
6
idle-timeout
4
coupling
3
unrecoverable-signature
3
flaky-test
2
security
2
Retries & recovery

Runs per issue

1.25 avg runs/issue
12345
Retries & recovery

Quarantine → retry rescue

retry rescues 59% of first-run quarantines · 97% of those actually retried
72%
first-try pass rate
16%
final-state quarantine (issues)
29%
quarantine rate (runs)
Cost · API-equivalent
fleet telemetry · per attempt · Jun 7 – Jul 26, 2026

Spend by model tier

$2,045.41 total across the window
sonnet$1198.08 · avg $1.51/attempt · n=794 attempts
unknown tier$674.64 · avg $1.69/attempt · n=399 attempts
opus$172.70 · avg $2.50/attempt · n=69 attempts
⚠ 33% of attempt spend ($674.64 · 399 attempts) has no recorded model_tier — telemetry gap, not a model
Overhead · orchestrator + scout ops
$240.94
208 cost events · additive to the tier spend above · not a model tier
Claude Code transcripts, afk worktrees only · per API call · Jun 29 – Jul 26, 2026
$0.03
median / call
$0.07
p90 / call
32,666
calls logged
whole machine, all projects: $7,877.12 across 65,776 calls
Cost dial

Per-API-call cost distribution

25871
5623
883
179
82
23
5
<$.05.05–.10.10–.15.15–.20.20–.30.30–.50$.50+
Model mix

Opus share by day

Jun 29 range: 0%–62% Jul 26
Throughput

Slices per day

848 runs
06-0706-1206-1706-2406-2907-0407-1407-22
Context cost

Tokens by type, per day

2.29B tokens · 95.8% cache reads — the context is overwhelmingly cached
cache read cache write input output