Llm Filesystem Tools

LLM Filesystem Tools

**Secure filesystem access for Large Language Models with governance-first design.**

[](https://badge.fury.io/py/llm-fs-tools) [](https://opensource.org/licenses/MIT) [](https://www.python.org/downloads/)

Stop reinventing filesystem tools for every LLM project. `llm-fs-tools` provides production-ready, secure file operations that work with any LLM supporting function calling - Ollama, OpenAI, Anthropic, and more.

The Problem

You want your AI assistant to analyze code, search files, or explore directories. You have three bad options:

1. Inject everything into the prompt - Wastes tokens, hits context limits, can't scale
2. Use heavy frameworks - LangChain/LlamaIndex lock you into their ecosystem
3. Roll your own - Reinvent security, path validation, and tool schemas every time

The Solution

pip install llm-fs-tools

from llm_fs_tools import FileSystemTools, SecurityPolicy

# Define security boundaries
policy = SecurityPolicy(
    allowed_roots=["./my-project"],
    max_file_size_mb=5,
    blocked_patterns=["*.env", ".git/*"]
)

# Initialize tools
fs_tools = FileSystemTools(policy)

# Use with any LLM (Ollama example)
import ollama

response = ollama.chat(
    model='qwen2.5-coder',
    messages=[{'role': 'user', 'content': 'Analyze the codebase structure'}],
    tools=fs_tools.get_tool_definitions()  # Auto-generates schemas
)

# Execute tool calls
for tool_call in response.message.tool_calls:
    result = fs_tools.execute(
        tool_call.function.name,
        tool_call.function.arguments
    )

**That's it.** Your model can now safely explore filesystems.

---

Features

🔒 Security First

• Path traversal protection - Validates all paths stay within allowed roots
• Configurable boundaries