loctree-mcp
Structural code intelligence for AI agents. Scan once, query everything ā dead exports, circular imports, dependency graphs, and more. CLI + MCP server.
README
<p align="center"> <img src="https://loct.io/assets/loctree-logo.png" width="128" alt="loctree logo"/> </p>
<h1 align="center">loctree</h1>
<p align="center"> <strong>Scan once, query everything.</strong><br/> AI-oriented static analysis for dead exports, circular imports, dependency graphs, and holographic context slices. </p>
<p align="center"> <a href="https://crates.io/crates/loctree"><img src="https://img.shields.io/crates/v/loctree.svg" alt="crates.io"/></a> <a href="https://crates.io/crates/loctree"><img src="https://img.shields.io/crates/d/loctree.svg" alt="downloads"/></a> <a href="https://docs.rs/loctree"><img src="https://docs.rs/loctree/badge.svg" alt="docs.rs"/></a> <a href="https://github.com/Loctree/loctree-ast/actions/workflows/ci.yml"><img src="https://github.com/Loctree/loctree-ast/actions/workflows/ci.yml/badge.svg" alt="CI"/></a> <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT%20OR%20Apache--2.0-blue.svg" alt="License"/></a> </p>
Install
curl -fsSL https://loct.io/install.sh | sh # CLI + loctree-mcp
cargo install --locked loctree loctree-mcp # Cargo, reproducible
npm install -g loctree # CLI only; published targets follow the latest npm release
brew install loctree/cli/loct # CLI via Homebrew tap
brew install loctree/mcp/loctree-mcp # MCP via Homebrew tap
Public install channels track the latest published release, which can lag behind
the workspace version on main. If you're validating a specific release, check
crates.io, npm, or GitHub Releases rather than assuming branch parity.
Quick Start
Artifacts are stored in your OS cache dir by default (override via LOCT_CACHE_DIR).
loct is the canonical CLI command. loctree remains available as a quiet compatibility alias.
loct # Scan project, write cached artifacts
loct --for-ai # AI-optimized overview (health, hubs, quick wins)
loct slice src/App.tsx --consumers # Context: file + deps + consumers
loct find useAuth # Find symbol definitions
loct find 'Snapshot FileAnalysis' # Cross-match: where terms meet
loct impact src/utils/api.ts # What breaks if you change this?
loct health # Quick summary: cycles + dead + twins
loct dead --confidence high # Unused exports
loct cycles # Circular imports
loct twins # Dead parrots + duplicates + barrel chaos
loct audit # Full codebase review
What It Does
loctree captures your project's real dependency graph in a single scan, then answers structural questions instantly from the snapshot. Designed for AI agents that need focused context without reading every file.
Core capabilities:
- Holographic Slice - extract file + dependencies + consumers in one call
- Cross-Match Search - find where multiple terms co-occur (not flat grep)
- Dead Export Detection - find unused exports across JS/TS, Python, Rust, Go, Dart
- Circular Import Detection - Tarjan's SCC algorithm catches runtime bombs
- Handler Tracing - follow Tauri commands through the entire FE/BE pipeline
- Impact Analysis - see what breaks before you delete or refactor
- jq Queries - query snapshot data with jq syntax (
loct '.files | length')
Why loctree
| grep/rg | LSP | loctree | |
|---|---|---|---|
| Knows imports vs definitions | No | Per-file | Whole graph |
| Dead export detection | No | No | Yes (multi-lang) |
| Cross-file impact analysis | No | Limited | Full transitive |
| AI agent integration | No | No | MCP server + --for-ai |
| Speed on 1M LOC repo | Fast (text) | Slow (indexing) | ~3s (structural) |
| Setup | None | Per-editor | One binary |
MCP Server
loctree ships as an MCP server for seamless AI agent integration:
loctree-mcp # Start via stdio (configure in your MCP client)
7 tools: repo-view, slice, find, impact, focus, tree, follow. Each tool accepts a project parameter ā auto-scans on first use, caches snapshots in RAM. Project-agnostic: analyze any repo without configuration.
{
"mcpServers": {
"loctree": {
"command": "loctree-mcp",
"args": []
}
}
}
Direct download users can also fetch signed release assets from the monorepo
GitHub release page, which mirrors both the CLI and loctree-mcp tarballs.
Language Support
| Language | Dead Export Accuracy | Notes |
|---|---|---|
| Rust | ~0% FP | Tested on rust-lang/rust (35K files) |
| Go | ~0% FP | Tested on golang/go (17K files) |
| TypeScript/JavaScript | ~10-20% FP | JSX/TSX, React patterns, Flow, WeakMap |
| Python | ~20% FP | Library mode, __all__, stdlib detection |
| Svelte | <15% FP | Template analysis, .d.ts re-exports |
| Vue | ~15% FP | SFC support, Composition & Options API |
| Dart/Flutter | Full | pubspec.yaml detection |
Auto-detects stack from Cargo.toml, tsconfig.json, pyproject.toml, pubspec.yaml, src-tauri/.
Holographic Slice
Extract 3-layer context for any file:
loct slice src/App.tsx --consumers
Slice for: src/App.tsx
Core (1 files, 150 LOC):
src/App.tsx (150 LOC, ts)
Deps (3 files, 420 LOC):
[d1] src/hooks/useAuth.ts (80 LOC)
[d2] src/contexts/AuthContext.tsx (200 LOC)
[d2] src/utils/api.ts (140 LOC)
Consumers (2 files, 180 LOC):
src/main.tsx (30 LOC)
src/routes/index.tsx (150 LOC)
Total: 6 files, 750 LOC
Cross-Match Search
Multi-term queries show where terms meet, not flat OR:
loct find 'Snapshot FileAnalysis'
=== Cross-Match Files (9) ===
src/snapshot.rs: Snapshot(6), FileAnalysis(4)
src/slicer.rs: Snapshot(2), FileAnalysis(3)
...
=== Symbol Matches (222 in cross-match files) ===
src/snapshot.rs:20 - Snapshot [struct]
src/types.rs:15 - FileAnalysis [struct]
...
=== Parameter Matches (4 cross-matched) ===
src/slicer.rs:45 - snapshot: &Snapshot in build_slice(analyses: &[FileAnalysis])
jq Queries
Query snapshot data directly:
loct '.dead_parrots' # Dead code findings
loct '.files | length' # Count files
loct '.edges[] | select(.from | contains("api"))' # Filter edges
loct '.summary.health_score' # Health score
CI Integration
loct lint --fail --sarif > results.sarif # SARIF for GitHub/GitLab
loct findings | jq '.dead_exports.total' # Check dead export count
loct findings --summary | jq '.health_score' # Health summary JSON
Crates
| Crate | Description |
|---|---|
loctree |
Core analyzer + CLI (loct, loctree) |
report-leptos |
HTML report renderer (Leptos SSR) |
loctree-mcp |
MCP server for AI agents |
Development
make precheck # fmt + clippy + check (run before push)
make install # Install loct, loctree, loctree-mcp
make test # Run all workspace tests
make publish # Cascade publish to crates.io
Badge
[](https://crates.io/crates/loctree)
License
MIT OR Apache-2.0. See LICENSE-MIT and LICENSE-APACHE.
š šššššššššš. with AI Agents ā 2025-2026 Loctree Team
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.