Skip to main content

Director.y

A natural-language TUI for safely managing your Windows filesystem with AI-generated PowerShell scripts.

Director.y is a terminal-based tool that lets you navigate, inspect, and manage your Windows filesystem using natural language. Describe a task or ask a question - Director.y will generate a safe PowerShell script, validate it, and execute it only with your approval.


Features

  • Natural Language Interface: Manage files without memorizing commands.
  • Modern TUI: Clean, responsive interface built with Textual.
  • Safety-First Execution: Sandboxing, path validation, script classification, and automatic retries.
  • Two Modes
    • Query Mode: Read-only questions about your filesystem.
    • Task Mode: File operations that require explicit approval.
  • Live Directory Tree: Auto-refreshing, interactive view of your working directory.
  • Flexible LLM Integration: Provider and model agnostic

Installation

pip install directory-agent

After installation, verify the dy command is available:

dy --version

Note: If the dy command isn't recognized, ensure Python's Scripts directory is in your PATH.

From Source (For Development)

git clone https://github.com/raytonc/Director.y.git
cd Director.y

# Install in development mode
pip install -e .

Configuration

Run the interactive configuration wizard to set up your API key:

dy --configure

The wizard will guide you through:

  1. Selecting your AI provider
  2. Entering and validating your API key
  3. Choosing a model
  4. Configuring timeouts and output limits

Your configuration will be saved to %APPDATA%\directory\config.toml (Windows).


Usage

Director.y enforces a sandbox for safety and must be run from a folder inside C:\Users\<YourName>.

# Navigate to a folder under your user directory
cd C:\Users\YourName\Downloads

# Run Director.y
dy

How It Works

Query Mode (Read-Only)

Ask questions like:

  • "What's taking up the most space?"
  • "How many PDFs are here?"
  • "Which files were modified today?"

Flow:

  1. AI generates a read-only PowerShell script
  2. The script is validated
  3. It runs safely in the sandbox
  4. Results are summarized in plain English

Task Mode (With Approval)

Examples:

  • "Organize files by type"
  • "Move all PDFs to Documents"
  • "Delete empty folders"
  • "Rename photos with dates"

Flow:

  1. AI analyzes your request
  2. A read-only planning script is generated
  3. An execution script is produced
  4. You review and approve all changes
  5. The script executes in the sandbox
  6. Results are summarized

Safety Features

Multi-Layer Protection

  1. Sandboxing -- Never operates outside the directory you run it from
  2. Path Validation -- All paths must stay within the sandbox
  3. Script Classification -- Read-only, write, or unsafe
  4. Syntax Validation -- PowerShell syntax checked before execution
  5. Manual Approval -- Required for all write operations
  6. Recycle Bin Safety -- Deletions are soft (moved to a temp location)
  7. Output Size Limits -- Default: 100 KB
  8. Timeouts -- 60s for read scripts, 300s for write scripts

Blocked Operations

  • Registry access
  • Process creation/execution
  • Network requests
  • Operations outside the sandbox
  • Dangerous cmdlets (e.g., Invoke-Expression)

Architecture

Core Components

src/directory/
├── agents/           # AI agents for different tasks
│   ├── query.py      # Generates read-only scripts
│   ├── planner.py    # Plans task execution
│   ├── executor.py   # Generates write scripts
│   └── summary.py    # Summarizes results
├── text/             # System prompts and text content
├── tui/              # Terminal user interface
├── execution.py      # Script validation and execution
├── workflows.py      # Orchestrates multi-step workflows
└── config.py         # Configuration management

Workflow Overview

Query Flow

User Question → Query Agent → Validation → Execution → Summary → User

Task Flow

User Task → Planner Agent → Planning Script → Execution Plan
         → Executor Agent → Validation → Syntax Check → User Approval
         → Execute → Summary → User

Commands

In-App Commands

  • help or /help -- Show help
  • quit or exit -- Exit Director.y
  • /quit or /exit -- Alternative exit aliases

Keyboard Shortcuts

  • Tab -- Switch between Query and Task modes
  • Ctrl+C -- Cancel current operation
  • Ctrl+Q -- Quit Director.y

CLI Options

dy --help          # Show help message
dy --version       # Show version information
dy --configure     # Configure API keys and settings

Development

Running Tests

pytest tests/ -v

Project Structure Highlights

  • Agents -- Specialized AI agents for planning and execution
  • Execution Layer -- Validates and runs PowerShell scripts safely
  • TUI Layer -- Interactive interface built with Textual
  • Workflow Engine -- Manages multi-step AI interactions with retry logic

Adding New Features

  1. Add system prompt to src/directory/text/
  2. Create agent in src/directory/agents/
  3. Add workflow in src/directory/workflows.py
  4. Add tests in tests/

Requirements

  • Windows (PowerShell required)
  • Python 3.11+
  • Must be run from directories under C:\Users\*

Important: Director.y executes PowerShell scripts. Always review proposed changes before approving. Multiple safety layers are in place, but you make the final decision.

Metadata

Release files for directory-agent 1.0.3

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for directory-agent 1.0.3
File Size Uploaded Provenance
directory_agent-1.0.3.tar.gz 35.3 kB Without provenance Details

Built distribution (wheel)

Table of built distributions (wheels) for directory-agent 1.0.3
File Interpreter ABI Platform
directory_agent-1.0.3-py3-none-any.whl Python 3 none any Details

Total release size: 72.9 kB

Release files / directory_agent-1.0.3.tar.gz

Download URL directory_agent-1.0.3.tar.gz
Size 35.3 kB
Tags Source
SHA-256 checksum
How to use checksums
976a8248029c231e7c18a77cc53039b3f546f80eeb0f9f97536c97134b3180cc
BLAKE2b-256 checksum
How to use checksums
7c68ec29e48faf759ddd0caddf4568c59526773619c5d2fc7e4da86d4442944b
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.3

Release files / directory_agent-1.0.3-py3-none-any.whl

Download URL directory_agent-1.0.3-py3-none-any.whl
Size 37.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
3612054116a8229a9c5c6c526841f8bbde8de51252ce1428f7c9e2a6dd140d5b
BLAKE2b-256 checksum
How to use checksums
ef02ce12b57d0ea8254d4990cd456885a02d7198bc55b2cec7ded4b81b6a1539
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.13.3

Security for directory-agent 1.0.3

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page