Context Engine MCP Server

Index the current workspace for semantic search.

This tool scans all source files in the workspace and builds a semantic index that enables fast, meaning-based code search.

When to use this tool:

• First time using the context engine with a new project

• After making significant changes to the codebase

• When semantic_search or enhance_prompt returns no results

What gets indexed (50+ file types):

• TypeScript/JavaScript (.ts, .tsx, .js, .jsx, .mjs, .cjs)

• Python (.py, .pyi)

• Flutter/Dart (.dart, .arb)

• Go (.go)

• Rust (.rs)

• Java/Kotlin/Scala (.java, .kt, .kts, .scala)

• C/C++ (.c, .cpp, .h, .hpp)

• .NET (.cs, .fs)

• Swift/Objective-C (.swift, .m)

• Web (.vue, .svelte, .astro, .html, .css, .scss)

• Config (.json, .yaml, .yml, .toml, .xml, .plist, .gradle)

• API schemas (.graphql, .proto)

• Shell scripts (.sh, .bash, .ps1)

• DevOps (Dockerfile, .tf, Makefile, Jenkinsfile)

• Documentation (.md, .txt)

What is excluded (optimized for AI context):

• Generated code (*.g.dart, *.freezed.dart, .pb.)

• Dependencies (node_modules, vendor, Pods, .pub-cache)

• Build outputs (dist, build, .dart_tool, .next)

• Lock files (package-lock.json, pubspec.lock, yarn.lock)

• Binary files (images, fonts, media, archives)

• Files over 1MB (typically generated or data files)

• Secrets (.env, *.key, *.pem)

The index is saved to .augment-context-state.json in the workspace root and will be automatically restored on future server starts.