Scalex

What to know before you start

Scalex is an open-source developer tool for exploring Git-tracked Scala 2, Scala 3, and Java source code through coding agents. It analyzes source directly with abstract syntax tree parsing, allowing an agent to locate definitions, inspect types, find reference candidates, follow inheritance, examine imports, and read method bodies without compiling the repository or starting a build server. The project is designed for repository understanding rather than as a replacement for a compiler, language server, or refactoring engine.

The tool is distributed as a native executable with an agent skill and bootstrap script. The documented installation paths cover Claude Code and Codex CLI or desktop through plugin marketplaces. For other coding agents, the repository provides instructions for copying the shared skill, command reference, and binary bootstrap into an agent's skill directory. The bootstrap downloads a pinned release on first use, verifies its SHA-256 checksum, and caches the executable. The binary does not require a JVM at runtime, although the documented build process uses GraalVM and the repository's Mill launcher.

Scalex exposes commands intended to reduce the number of exploratory tool calls an agent needs. The `explain` operation combines a definition, documentation, members, implementations, and import information in one response. Other documented capabilities include symbol search, definitions, categorized reference candidates, inheritance hierarchies, wildcard-import resolution, package and annotation navigation, method-body extraction, batch queries, JSON output, result limits, and composable filters such as kind, path, test exclusion, exact matching, category, package scope, and maximum output. The plugin invokes a bundled CLI script rather than adding a `scalex` command to the user's PATH.

Its indexing model is based on Git-tracked files, Git object IDs, source parsing, binary indexes, and bloom filters. On a repository, Scalex identifies tracked files, compares cached object IDs, parses changed or new records, and stores repository data in a `.scalex/index.bin` file. Subsequent queries can reuse unchanged records, while batch queries share one index load. The README documents Scala parsing through Scalameta and Java parsing through JavaParser. It also describes a per-repository cache and a native executable, so there is no daemon or continuously running process to manage.

The project website presents performance measurements from a production monorepo and the Scala 3 compiler, including cold and warm indexing times, symbol counts, and query examples. These figures are project-published measurements rather than independently verified benchmarks. The site also describes import analysis, transitive hierarchy traversal, and query batching with example comparisons against grep. Those examples help explain the intended workflow, but they should not be treated as universal results for every repository, language feature, machine, or workload.

Scalex has explicit operating boundaries. It performs source-level analysis and does not provide compiler type resolution, verified test coverage, or rename refactoring. Reference results are candidates that can include comments or unrelated symbols with the same name, even though import information and confidence categories help narrow them. Test-related commands find mentions in test files rather than executed lines or verified assertions. Untracked files are excluded because indexing follows Git-tracked content, and unstaged edits can leave cached results stale; the documentation recommends checking fresh edits directly and using the `diff` command where appropriate. The documented supported platforms are macOS Apple Silicon, macOS Intel, and Linux x64. Pricing beyond the repository's open-source MIT license is not separately documented.