Flyto Indexer

Flyto Indexer provides a Python-based MCP server that gives AI-assisted development tooling through code intelligence. You can index your codebase locally, expose 23 MCP tools to your AI assistants, and work entirely offline with pure Python.

How to use

You use this MCP server by connecting an AI client to either a local Python-based MCP server or by running the server from source. Start by indexing your project to create a local index, then run the MCP server so your editor or AI assistant can call the 23 available tools for impact analysis, references, search, and more. The server runs entirely on your machine and keeps your data private.

How to install

Prerequisites: you need Python installed on your machine. You will use the Python package manager to install Flyto Indexer or you can run it directly from source.

# Option A: Install from PyPI
pip install flyto-indexer

# Index your project (creates .flyto-index/)
flyto-index scan /path/to/your/project

# Start MCP server
python -m flyto_indexer.mcp_server

# Option B: Run from source

git clone https://github.com/flytohub/flyto-indexer.git
cd flyto-indexer
pip install -e .

# Index your project
flyto-index scan /path/to/your/project

# Start MCP server (from repo)
python -m src.mcp_server

Configuration and start tips

You can connect your MCP client to the local server. If you are using Claude Code or another client, configure the MCP server as follows.

{
  "mcpServers": {
    "flyto-indexer": {
      "command": "python3",
      "args": ["-m", "flyto_indexer.mcp_server"]
    }
  }
}

Connecting to Claude Code from source

If you are running from source, you can connect Claude Code by configuring the MCP server with the appropriate command.

{
  "mcpServers": {
    "flyto-indexer": {
      "command": "python3",
      "args": ["-m", "src.mcp_server"],
      "cwd": "/path/to/flyto-indexer"
    }
  }
}

Available tools

impact_analysis

Analyze the blast radius of modifying a symbol to identify what could break.

find_references

Find all callers and importers of a symbol across the indexed code.

search_code

Symbol-aware search across all indexed projects to locate relevant code.

dependency_graph

Show import chains and dependent relationships for a symbol or file.

get_symbol_content

Retrieve the full source content of a specific function or class.

find_dead_code

Detect unreferenced or unused code that may be safely removed.

get_file_symbols

List all symbols defined in a given file.

get_file_info

Get metadata about a file, including purpose, category, keywords, and dependencies.

fulltext_search

Search inside comments, strings, and TODO markers across indexed code.

check_index_status

Check whether the index is up-to-date with the current workspace.

list_projects

List all indexed projects with summary statistics.

list_categories

Show code categories such as authentication, payment, etc.

list_apis

List all API endpoints found within the codebase.

update_description

Write or update a file description to reflect its purpose.

get_description

Get the semantic one-liner description for a file.

brief

Generate a high-level project outline (L0) for quick context.

describe

Annotate files with their purposes to improve discoverability.

check_and_reindex

Detect file changes and trigger re-indexing as needed.

session_track

Track workspace events to boost search quality.

session_get

Inspect the current session state used by the MCP client.

edit_impact_preview

Preview impact changes before performing renames or signature changes.

get_file_context

Provide a complete context package for a file, including symbols and dependencies.