Testing Cursor vs Codex vs Claude Code on one SaaS: config pollution is the real story
SSShken · reddit · 2026-10-03
The author gave the same feature spec to Cursor, Codex and Claude Code on their own live SaaS — same branch, cheapest paid plan, default settings, and a 16-point acceptance list written upfront.
Key findings:
- Config follows you around: Cursor auto-imports Claude Code plugins (toggle on by default); a brand-new Claude account still loaded plugins, skills and an MCP server pointing at an unrelated repo, because they live in the home folder — only pointing CLAUDECONFIGDIR at an empty dir gave a clean session.
- Projects remember things outside the instructions file: generated doc lists hardcoded in four places led two agents to create a required doc without wiring it in; generation reported success and done-checks passed.
- Scope of responsibility, not model quality, separated the tools: with the local DB down, Cursor stopped with "couldn't verify anything"; Codex started Docker itself, applied migrations, wrote Playwright tests and found a serialization bug in its own code; Claude Code did the same plus wrote contrast tests and fixed two WCAG AA failures.
- Numbers: Claude Code 38 min / 6% of weekly limit; Cursor 52 min / 3% of monthly quota; Codex hit its 5h cap mid-feature.
Caveat: one run each, no noise floor measured.
Related event: Hands-on Comparison: Cursor, Codex and Claude Code Fail at Wiring, Not Code(2 posts)→
More from coding & agent
- Claude Code overtakes Cursor and Copilot in job-post mentions ahead of Anthropic IPO — antgoldbloom · 2026-11-19
- Dev submits PR adding periodic checkpoints to GKE sandboxes so agents don't lose progress — yoimnotkesku · 2026-11-19
- Orchestrator's Cloud Agents graduate from beta to general availability — julianweisser · 2026-11-19
- Zero users, 33 pages shipped: dev crowdsources audit for AI-agent trust schema — TheHamzaShahbaz · 2026-11-19
- She vibecoded her own photobooth with Claude Code — and shared the prompt — toolstelegraph · 2026-11-19
- Should AI Agents Write to Your Memory or Only Read? Dev Weighs In — jordimr · 2026-11-19