Lightsage Research
Stay ahead of the Devtool Arena leaderboard
Join the mailing list for leaderboard findings, benchmark updates, and practical guidance on improving how AI coding tools discover and use your API.
What you'll get
- New Devtool Arena leaderboard benchmark releases
- Breakdowns of scoring changes and methodology updates
- Product updates from Lightsage relevant to API teams
Join the mailing list
We'll send occasional updates about the Lightsage leaderboard and key product announcements.
Research areas we cover
Lightsage research focuses on how AI coding agents interact with developer tools, APIs, and documentation. Our benchmarks help API teams understand and improve their agent experience.
AI Coding Agent Benchmarks
How Claude Code, Cursor, Codex, and GitHub Copilot discover and use developer APIs. We measure discoverability, tool call success rates, and error handling across 65+ APIs.
Agent Experience Scoring
Our methodology for grading APIs on agent readiness. Factors include documentation quality, SDK design, MCP server availability, and llms.txt implementation.
Market Trends
Data on how AI coding agents are changing API consumption patterns. Which documentation formats perform best, and how agent traffic compares to traditional developer traffic.