# Agent Usability

Run realistic integration tasks across coding agents and find the documentation, authentication, tool-call, and error-handling issues that block successful use.

## What We Measure

- API, CLI, and MCP discoverability
- Tool-call and task-completion success
- Authentication and error recovery
- Performance by endpoint and coding agent

## Outcome

Lightsage turns benchmark failures into specific engineering recommendations and lets teams compare results after fixes ship.

Canonical URL: https://lightsage.com/agent-usability