Overview

· 150 finalized PRs in past 30d

Output Quality
Avg Post-Open Commits
1.5
↓ 0.3 vs prior 30d
Lower is better
CI Success Rate
82%
↑ 4% vs prior 30d
First-run pass rate
Avg Line Revisit Rate
0.16
↑ 0.02 vs prior 30d
Cross-PR file overlap
Prompt Efficiency
Avg Iteration Depth
8
↓ 1 vs prior 30d
Human-agent turn pairs
Avg Token Cost
$1.80
↑ $0.11 vs prior 30d
116 of 150 PRs with session data
Avg Cache Hit Rate
65%
↑ 5% vs prior 30d
Prompt cache utilization
Agent Behavior
Avg Sidechain Rate
12%
↓ 3% vs prior 30d
Dead-end reasoning paths
Avg Re-Read Rate
1.65
↓ 0.15 vs prior 30d
File read redundancy
Avg Autonomy Score
7.6
↑ 0.8 vs prior 30d
Agent independence ratio
View all pull requests →Browse teams →