Morphiq Affinity Bench
What coding agents
pick
Look at what developer tools Codex, Claude Code, and Cursor discover and integrate.
We benchmarked 1,000 GitHub repositories across six skill use cases and 5 skill providers to measure which extra skills GPT-5.5-Codex and Claude Opus 4.7 reach for in real work.
Topskill providersacross all skill use cases
Skill Affinity
30.4%
GStack
Skill Affinity
25.1%
Anthropic Claude Code
Skill Affinity
21.1%
Superpowers
Skill Affinity forCode Review
Switch via the Skill filter above
Provider
Skill Affinity
The overall affinity podium and model spread above stay scoped to all six skill use cases. The provider table is scoped to the use case selected in the Skill filter inside the trend chart — currently Code Review. Each agent runs the same 1,000 GitHub repositories per use case. Latest snapshot: 2026-05-13.