Continually improving our agent harness

kept by eddie may 01

Cursor improves its coding agent by iterating on context management, evaluation metrics, and model-specific harness customization, treating the harness as a software product.

Training Composer for longer horizons
Cursor trains Composer to generate its own compact summaries mid-task, reducing context errors by 50% while using 80% fewer tokens than traditional prompting.
also on CursorBench, Cursor, #ai-agents, #ai-coding, #software-engineering, #developer-tools
claude code is not making your product better
Coding agents increase senior engineer output but not product velocity, as the bottleneck is product taste and complexity management, not code production speed.
also on Linear, Cursor, Anthropic, #ai-agents, #user-experience, #software-engineering, #developer-tools
Building An Elite AI Engineering Culture In 2026
High-performing engineering teams in 2026 are defined by taste, discipline, and leverage—not by AI tool choice—and a small set of exemplars like Linear, Cursor, Vercel, and Stripe share structural practices like writing-driven culture, design-engineer fusion, spec-driven development, tiny teams, and ruthless quality habits.
also on Linear, Cursor, #ai-coding, #software-engineering, #developer-tools