Test Coverage Mcp

Reads LCOV files and gives your AI agent structured coverage data without burning tokens on massive raw files. Exposes four tools: coverage_summary for project-wide percentages, coverage_file_summary for individual file metrics, start_recording to snapshot a baseline, and get_diff_since_start to measure coverage impact during a session. Useful when you want agents to maintain or improve coverage as they write code, or when testing agents need to identify uncovered areas. Works with any language that generates LCOV output. The baseline tracking is clever because agents don't have to hold initial coverage numbers in their context window throughout the entire conversation.