ScalaSemantic — Documentation

ScalaSemantic gives AI coding agents compiler-resolved Scala facts over MCP. It reads compiler-emitted SemanticDB, so answers reflect what the compiler resolved — not text matching.

Get started

• Quickstart — install in 5 minutes
• Integration — launcher, plain jar, logging
• FAQ — compile freshness, Metals/LSP, install choices

Reference

• Tool reference — all MCP tools and SemanticDB symbol grammar
• Tool Examples — real tool calls and responses, executed at docs build time

Understanding the trade-offs

• ScalaSemantic vs grep — where each wins; measured token savings
• The motivation — why SemanticDB matters

Project & Development

• Development — repository layout, build, test, cross-version
• Design decisions — implementation approach and extension points
• ADR 0001: Claude Code guard hook — why setup installs a hook that denies text tools on .scala sources
• Releasing — Sonatype Central release process
• Property-based testing audit — conversion candidates for property/golden tests

Research & Planning

• Claude interaction study — measured tool usage, tool recommendations
• Token metrics methodology — measurement definitions and generated results
• Token metrics live run — raw agent-session token counts, with/without MCP
• Token metrics findings — end-to-end agent token usage, coverage, and limitations
• ScalaSemantic vs grep response metrics — 337-call log analysis of tool vs grep token cost
• Compat fixture sources — reusable Scala/SemanticDB fixture corpora
• sbt build inventory — baseline for the sbt-to-Mill migration