Fast-Context

Fast-Context is a Rust codebase analysis library with Node.js and Python bindings.

It is built around Tree-sitter parsing, symbol extraction, dependency analysis, and graph operations. The project is meant for code search, indexing, editor tooling, and assistant-style codebase analysis.

Coming soon: the npm, PyPI, and crates.io packages have not been published yet. The repository and release flow are being finalized before the first public package release.

What It Does

• analyzes multi-language repositories
• extracts symbols and relationships
• finds symbols by kind or file
• resolves symbol dependencies
• exposes graph types for traversal and metrics
• supports file watching and incremental workflows

Current Surface

The actively maintained surfaces are:

• Rust core library
• Node.js package
• Python bindings
• Codex skills
• Claude plugin skeleton

Installation

The commands below describe the intended package install surface after the first public release.

Node.js

npm install fast-context

Python

pip install fast-context

Hosts and MCP

For host-specific installation and setup, see docs/HOST_SETUP.md.

That document covers:

• Codex skills, MCP setup, and optional hooks
• Claude plugin installation, bundled skills, plugin MCP, and optional hooks
• standalone MCP installation for non-Codex/non-Claude clients

Quick Start

Node.js

const { FastContextAnalyzer } = require('fast-context')

const analyzer = new FastContextAnalyzer({
  projectRoot: process.cwd(),
  ignorePatterns: ['node_modules/**', '.git/**', 'target/**'],
})

const result = analyzer.analyze()

console.log({
  files: result.fileCount,
  symbols: result.symbolCount,
  relationships: result.relationshipCount,
  skipped: result.skippedFileCount,
})

Python

import fast_context

config = fast_context.AnalyzerConfig(
    project_root=".",
    ignore_patterns=["node_modules/**", ".git/**", "target/**"],
)

analyzer = fast_context.FastContextAnalyzer(config)
result = analyzer.analyze()

print({
    "files": result.file_count,
    "symbols": result.symbol_count,
    "relationships": len(result.relationships),
    "skipped": len(result.skipped_files),
})

Node API Summary

FastContextAnalyzer currently exposes:

• analyze()
• getAnalysis()
• startWatching()
• stopWatching()
• findSymbolsByKind(kind)
• findSymbolsInFile(filePath)
• findDependencies(symbolName)
• findComplexSymbols(complexityThreshold)

Utility exports:

• getVersion()
• getSupportedLanguages()
• detectLanguage(filePath)
• checkConfiguration(config?)
• getSystemInfo()

For the current typed contract, see index.d.ts and docs/API_REFERENCE.md.

Supported Languages

Fast-Context includes parsers for:

• Rust
• JavaScript
• TypeScript
• Python
• Java
• Go
• C and C++
• C#
• Swift
• Objective-C
• PHP
• Ruby
• Scala
• Zig
• Lua
• Bash
• CSS
• HTML
• XML
• JSON
• YAML
• Markdown

Development

Rust

cargo check
cargo test
cargo clippy --all-targets --all-features -- -D warnings
cargo run --bin fast-context --features cli -- --help

Node.js

npm install
npm run build:debug
npm test

Python

pytest tests/python

If you are building native bindings locally, use the versions and host environments exercised by CI where possible.

Documentation

• API reference
• Deployment guide
• Host setup
• Host integration spec
• Host integration plan
• Release architecture
• Documentation index

Repository Notes

• The generated Node typings live in index.d.ts.
• The Node package metadata is rewritten during build/publish steps, so local package.json churn after native builds is expected.
• Cross-platform native artifact truth should come from the GitHub Actions workflows, not a single local machine.

License

Apache-2.0

Support

• GitHub repository
• Issue tracker