← charleslikesdata.com afk cockpit
6 repos · 2,909 attempts · 135 merged · data through Jun 1 – Jul 5, 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.
5%
landed rate
n = 2,703 gate-passed
5%
landed rate (executor repos)
n = 6 exec repos
$6.33
API-equiv $ / merged PR
n = 135 merged
7%
quarantine (safety) rate
n = 2,909 attempts
5%
escalation rate
n = 2,909 attempts
0%
rejected rate
n = 836 proposals
Pipeline

Stage funnel

proposed
836
promoted
568
attempted
2,909
passed gate
2,703
published
143
merged
135
By repository

Outcomes per repo

landed quarantined other
Orbit-Wars
973 · 0%
afk-agent-system
960 · 7%
afk-cockpit
334 · 7%
claude-workflow
413 · 3%
household-bms
203 · 21%
my-brain
26 · 0%
Producers

Proposal quality

dropped
125
harvest
5
mis-sized
31
pending
105
rejected
1
shipped
568
superseded
1
Safety axis

Quarantine by category

206 total
other
77
no-op
36
build-crash
16
capability
16
integration-conflict
14
review
10
setup-crash
10
test-conflict
6
deny-surface
5
idle-timeout
4
timeout
4
coupling
2
scope
2
unrecoverable-signature
2
flaky-test
1
security
1
Cost · API-equivalent
fleet telemetry · per attempt · Jun 1 – Jul 5, 2026

Spend by model tier

$854.34 total across the window
sonnet$753.41 · avg $1.21/attempt · n=621
opus$100.93 · avg $2.06/attempt · n=49
Claude Code transcripts, afk worktrees only · per API call · Jun 6 – Jul 5, 2026
$0.03
median / call
$0.08
p90 / call
46,331
calls logged
whole machine, all projects: $6,131.61 across 67,467 calls
Cost dial

Per-API-call cost distribution

36768
7125
1953
342
115
27
1
<$.05.05–.10.10–.15.15–.20.20–.30.30–.50$.50+
Model mix

Opus share by day

Jun 6 range: 0%–100% Jul 5
Throughput

Slices per day

2,909 total
06-0106-0706-1106-1506-1906-2406-2807-02
Context cost

Tokens by type, per day

2.79B tokens · 95.0% cache reads — the context is overwhelmingly cached
cache read cache write input output