phind-cli

phind: An AI-Enhanced, modern, intuitive, cross-platform command-line tool for finding files and directories recursively, designed with developers in mind.

---

Why phind and not find? Because phind offers a consistent, intuitive, cross-platform experience with sensible defaults tailored for developers, plus the power of AI-driven natural language search, unlike the fragmented and often complex native find commands.

---

Why Another find Tool?

The standard find command, while powerful, presents several challenges, especially for developers working across different environments:
1. AI-Powered Search: Native find requires precise, often complex, patterns. It cannot interpret natural language queries like "find all configuration files" or "show me the test setup".

2. Cross-Platform Inconsistency:
* The syntax and available options for find can differ significantly between operating systems (e.g., GNU find on Linux vs. BSD find on macOS). This forces users to remember different flags or consult documentation frequently.
* Windows lacks a direct, built-in equivalent. Developers on Windows often resort to using dir /s /b, installing Git Bash (which includes GNU find), using WSL, or employing PowerShell commands, none of which provide the same seamless experience or syntax as Unix-like find.

3. Verbose and Often Unintuitive Syntax:
* Standard find syntax can be complex, requiring explicit actions like -print and involving less intuitive operators for combining conditions (e.g., -o for OR, escaped parentheses for grouping).
* Excluding directories effectively (pruning) often involves cumbersome patterns like -path './node_modules' -prune -o -print.

4. Lack of Sensible Defaults for Developers:
This is a major pain point. When searching in a typical software project, you almost never want to see results from node_modules, .git, .gradle, build directories, etc. Standard find requires you to explicitly exclude these every single time*. This adds significant boilerplate to common search commands.
5. No User-Level Configuration:
Beyond project-specific ignores (like .gitignore, which find doesn't natively understand anyway), there's no standard way to tell find to always ignore certain patterns across all projects. Think about system files (.DS_Store, Thumbs.db), editor configuration (.vscode/, .idea/), temporary files (.bak, .swp), or compiled artifacts (.pyc, .o). With standard find, you must manually add exclusion flags for these user-specific or system-specific nuisances every single time you run a command*, or resort to complex shell aliases or wrapper scripts.

The Solution: phind

phind aims to solve these problems by providing a unified and intelligent file finding experience:

* True Cross-Platform Compatibility: Runs consistently on Linux, macOS, and Windows with the same command and options.
* Sensible Developer Defaults: Crucially, phind automatically excludes common directories like node_modules, .git, and .gradle by default. This drastically simplifies finding relevant project files without manual exclusion flags in most cases.
AI-Powered Natural Language Search: Leverages Google Gemini to interpret your natural language queries (--ai "your query") and identify the most relevant files from the search results. Find files based on intent*, not just exact patterns. Requires a GEMINI_API_KEY.
* Intuitive Glob Patterns (Standard Mode): Uses familiar glob patterns (like those in .gitignore) for both including (--name) and excluding (--exclude) files and directories when not using AI mode.
* Simplified Options: Offers clear, easy-to-understand flags for common tasks like filtering by type (-t f/-t d), limiting depth (-d), case-insensitive search (-i), and relative path output (-r).
* Global Ignore File: Supports a global ignore file (~/.config/phind/ignore or platform equivalent) for persistent, user-defined excludes across all projects.
* Performance: Leverages modern Node.js features like async I/O and fs.Dirent for efficient traversal.

In essence, phind is designed to be the find command you wish you had – consistent, simple for common tasks, intelligently configured for software development workflows out of the box, and now with the power of AI understanding.

Key Features

* ✅ Cross-Platform: Works identically on Windows, macOS, and Linux.
* 🧠 Smart Defaults: Automatically ignores node_modules, .git, and .gradle.
* ✨ Intuitive Syntax: Uses simple flags and familiar glob patterns.
* 🤖 AI-Powered Search: Use natural language queries to find relevant files (via --ai and Google Gemini).
* 🔍 Flexible Filtering: Filter by name/path (--name), type (--type), and depth (--maxdepth).
* 🚫 Powerful Exclusions: Exclude patterns via CLI (--exclude) or a global ignore file.
* ⚙️ Configurable: Skip global ignores (--skip-global-ignore), control case sensitivity (--ignore-case).
* 📄 Output Control: Print relative (default) or absolute paths (--relative=false).
* 🚀 Modern & Performant: Built with TypeScript and modern Node.js APIs. (Defaults to relative paths!)

Installation

``sh
npm install -g phind-cli
`

Alternatively, you can use it directly without installation via npx:

`sh
npx phind-cli [path] [options]
`

Usage

The basic command structure is:

`sh
phind [path] [options]
`

* [path]: (Optional) The directory to start searching from. Defaults to the current directory (.).
* [options]: (Optional) Flags to control filtering, exclusion, output format, etc.

Options

Here is a detailed breakdown of every available command-line option:

---

$3

* Description: The directory to search within. This is the starting point for the recursive traversal.
* Type: string
* Default: . (the current working directory)
Details: If provided, it must be the first argument that isn't part of an option flag. phind . --name ".js" and phind --name "*.js" are effectively the same if run from the same directory. It must be a valid directory.

---

$3

* Description: Activates AI Mode. Uses AI (Google Gemini) to filter the file list based on a natural language query provided as the argument. This allows you to find files based on their likely purpose or content description, rather than just their name or path.
* Type: string
* Requires: The GEMINI_API_KEY environment variable must be set with a valid API key for Google Gemini. You can obtain one from Google AI Studio.
* Behavior:
1. phind first performs an initial, broad file traversal starting from [path].
2. This initial traversal respects default excludes (node_modules, .git, .gradle) and global ignores (unless --skip-global-ignore is used). It does not use --name, --type, --maxdepth, --ignore-case or CLI --exclude patterns for this initial collection. It collects all non-excluded file paths (relative paths are used internally for the AI).
3. The collected list of file paths and your natural language are sent to the Google Gemini model (gemini-2.5-pro-preview-03-25).
4. The AI analyzes the list and query to identify the files it deems most relevant.
5. phind prints only the file paths returned by the AI.
Interaction with other flags: When --ai is used, it takes precedence for determining the final output. Other filtering flags like --name, --exclude (CLI), --type, --maxdepth, and --ignore-case are ignored during the AI processing stage. The initial file collection only* respects default and global excludes (unless --skip-global-ignore is used).
* Example: phind --ai "find all React components"

---

$3

* Description: Do not load patterns from the global ignore file. Applies to both standard mode and the initial file collection step in AI mode.
* Type: boolean
* Default: false
* Details: Use this flag if you want to temporarily ignore your global configuration (see Global Ignore File section below) and rely only on the hardcoded defaults (node_modules, .git, .gradle) and any patterns provided via --exclude (in standard mode).

---


--- Standard Filtering & Output Options (Ignored in AI Mode) ---

The following options are used for standard phind operation but are ignored when --ai is active, as the AI handles the final filtering based on your query.

---

$3

* Description: Glob pattern(s) for filenames or paths to include. You can specify this option multiple times.
* Type: string (array)
Default: [''] (matches all files and directories unless excluded)
Default Description: "" (all files/dirs)
* Details: Uses micromatch for globbing. Patterns are matched against the item's base name (e.g., file.txt), its full absolute path, and its relative path (if --relative is used). Hidden files/directories (starting with .) are matched by default because micromatch's dot: true option is enabled internally.
Example: --name ".ts" finds all files ending in .ts.
Example: --name "src//.ts" finds TypeScript files within the src directory and its subdirectories.
Example: --name package.json --name ".lock" finds package.json files and all lock files.

---

$3

* Description: Glob pattern(s) to exclude files or directories. Also reads patterns from the global ignore file unless --skip-global-ignore is used. You can specify this option multiple times.
* Type: string (array)
* Default: [] (CLI arguments default to none, but hardcoded defaults and global ignores are applied)
* Default Description: "node_modules", "".git", "".gradle" (These are the hardcoded defaults that are always added unless overridden by specific includes).
Details: Patterns are matched similarly to --name (base name, absolute path, relative path). If a directory matches an exclude pattern, it is pruned* – phind will not descend into it. This is highly efficient for skipping large directories like node_modules. Exclude patterns take priority over include patterns. The final list of excludes combines hardcoded defaults (node_modules, .git), patterns from the global ignore file, and patterns provided via the CLI --exclude flag.

---

$3

* Description: Match only items of a specific type.
* Type: string
* Choices: f (file), d (directory)
* Default: null (matches both files and directories)
* Details:
* -t f: Only outputs files.
* -t d: Only outputs directories.

---

$3

* Description: Maximum directory levels to descend. 0 means only the starting path itself.
* Type: number
* Default: Infinity (represented internally as Number.MAX_SAFE_INTEGER)
* Details: Controls the recursion depth.
* --maxdepth 0: Only considers the [path] argument itself.
* --maxdepth 1: Considers the [path] argument and items directly within it.
* --maxdepth 2: Considers items up to two levels deep from the [path].

---

$3

* Description: Perform case-insensitive matching for --name and --exclude patterns.
* Type: boolean
* Default: false
Details: Affects how glob patterns are matched against file and directory names/paths. For example, with -i, --name ".jpg" would match image.jpg, image.JPG, and image.JpG.

---

$3

* Description: Print paths relative to the starting directory (default). Use --relative=false for absolute paths.
* Type: boolean
* Default: true
* Default Description: true (relative paths)
* Details:
* Default (true): Prints paths relative to the [path] argument (e.g., ./src/file.ts). The starting directory itself is represented as .. Paths are prefixed with ./ unless they start with ../.
* --relative=false: Use this explicitly to print absolute paths (e.g., /home/user/project/src/file.ts).

---

$3

* Description: Show the help message listing all options and exit.
* Type: boolean
* Details: Displays usage information, descriptions of all options, and their defaults.

---

Global Ignore File

phind supports a global ignore file to specify patterns that should always be excluded, regardless of the project you're currently in. This is useful for editor configuration files, OS-specific files, temporary files, etc.

Location:

The location of the global ignore file follows platform conventions:

1. Environment Variable: If PHIND_TEST_GLOBAL_IGNORE_PATH is set (primarily for testing), that path is used.
2. Windows: %APPDATA%\phind\ignore (e.g., C:\Users\YourUser\AppData\Roaming\phind\ignore)
3. Linux/macOS (XDG): $XDG_CONFIG_HOME/phind\ignore (if $XDG_CONFIG_HOME is set)
4. Linux/macOS (Fallback): ~/.config/phind/ignore (e.g., /home/youruser/.config/phind/ignore)

Format:

* Plain text file.
* One glob pattern per line.
* Lines starting with # are treated as comments and ignored.
* Blank lines are ignored.
Leading/trailing whitespace on a line is trimmed after* removing comments.

Example ignore file:

`gitignore


OS-specific files

.DS_Store
Thumbs.db

Editor/IDE config

.vscode/
.idea/
*.sublime-project
*.sublime-workspace

Temporary files

*.tmp
*.bak
*.swp

Build output common across projects

dist/
build/
out/
`

Disabling:

Use the --skip-global-ignore flag to prevent loading this file for a specific command execution.

Examples

Let's assume the following directory structure for the examples:

`
my_project/
├── src/
│ ├── main.ts
│ ├── utils/
│ │ └── helper.ts
│ └── components/
│ └── Button.tsx
├── dist/ # Typically excluded by global ignore
│ └── bundle.js
├── node_modules/ # Excluded by default
│ └── some_lib/
│ └── index.js
├── test/
│ ├── main.test.ts
│ └── utils.test.js
├── public/
│ ├── index.html
│ └── styles.css
├── .git/ # Excluded by default
│ └── config
├── package.json
├── README.md
├── .env
└── config.yaml
`

---

$3

a. Find all items in the current directory (defaults applied)
`sh
cd my_project
phind
`

Output (Relative Paths - Default, node_modules & .git excluded):


`
.
./.env # Example output format
./README.md
./config.yaml
./dist
./dist/bundle.js
./package.json
./public
./public/index.html
./public/styles.css
./src
./src/components
./src/components/Button.tsx
./src/main.ts
./src/utils
./src/utils/helper.ts
./test
./test/main.test.ts
./test/utils.test.js
`

b. Find all items using absolute paths
`sh
cd my_project
phind --relative=false
`
Output (Absolute Paths, platform-specific):
`
/path/to/my_project
/path/to/my_project/.env
/path/to/my_project/README.md
/path/to/my_project/config.yaml
/path/to/my_project/dist
/path/to/my_project/dist/bundle.js
/path/to/my_project/package.json
/path/to/my_project/public
/path/to/my_project/public/index.html
/path/to/my_project/public/styles.css
/path/to/my_project/src
/path/to/my_project/src/components
/path/to/my_project/src/components/Button.tsx
/path/to/my_project/src/main.ts
/path/to/my_project/src/utils
/path/to/my_project/src/utils/helper.ts
/path/to/my_project/test
/path/to/my_project/test/main.test.ts
/path/to/my_project/test/utils.test.js
`

c. Find all items starting in a specific subdirectory (src)
`sh
cd my_project
phind src
`
Output (Relative Paths - Default):
`
./src
./src/components
./src/components/Button.tsx
./src/main.ts
./src/utils
./src/utils/helper.ts
`

d. Find items in src using absolute paths
`sh
cd my_project
phind src --relative=false
`
Output (Absolute Paths):
`
/path/to/my_project/src
/path/to/my_project/src/components
/path/to/my_project/src/components/Button.tsx
/path/to/my_project/src/main.ts
/path/to/my_project/src/utils
/path/to/my_project/src/utils/helper.ts
`

---

$3

a. Find all TypeScript files (*.ts) (Standard Mode)
`sh
cd my_project
phind --name "*.ts" # Relative is default
`
Output:
`
./src/main.ts
./src/utils/helper.ts
./test/main.test.ts
`

b. Find all TypeScript and TSX files (Standard Mode)
`sh
cd my_project
phind --name ".ts" --name ".tsx" # Relative is default
`
Output:
`
./src/components/Button.tsx
./src/main.ts
./src/utils/helper.ts
./test/main.test.ts
`

c. Find specific configuration files (Standard Mode)
`sh
cd my_project
phind --name package.json --name config.yaml # Relative is default
`
Output:
`
./config.yaml
./package.json
`

d. Find all items within the src/utils directory (Standard Mode)
`sh
cd my_project
phind --name "src/utils/**" # Relative is default
`
Output:
`
./src/utils/helper.ts
`
(Note: src/utils/ matches items inside utils, not the directory itself)

e. Find the src/utils directory itself and its contents (Standard Mode)
`sh
cd my_project
phind --name "src/utils" --name "src/utils/**" # Relative is default
`
Output:
`
./src/utils
./src/utils/helper.ts
`

---

$3

a. Find only files (Standard Mode)
`sh
cd my_project
phind -t f # Relative is default
`
Output (Relative paths, All files, excluding default ignores):
`
./.env
./README.md
./config.yaml
./dist/bundle.js
./package.json
./public/index.html
./public/styles.css
./src/components/Button.tsx
./src/main.ts
./src/utils/helper.ts
./test/main.test.ts
./test/utils.test.js
`

b. Find only directories (Standard Mode)
`sh
cd my_project
phind -t d # Relative is default
`
Output (Relative paths, All directories, excluding default ignores):
`
.
./dist
./public
./src
./src/components
./src/utils
./test
`

c. Find only JavaScript files (*.js) (Standard Mode)
`sh
cd my_project
phind -t f --name "*.js" # Relative is default
`
Output:
`
./dist/bundle.js
./test/utils.test.js
`

---

$3

a. Find only items in the immediate directory (depth 0) (Standard Mode)
`sh
cd my_project
phind --maxdepth 0 # Relative is default
`
Output:
`
.
`

b. Find items at depth 0 and 1 (Standard Mode)
`sh
cd my_project
phind --maxdepth 1 # Relative is default
`
Output (Top-level files/dirs, excluding default ignores):
`
.
./.env
./README.md
./config.yaml
./dist
./package.json
./public
./src
./test
`

c. Find only directories at depth 0 and 1 (Standard Mode)
`sh
cd my_project
phind -t d --maxdepth 1 # Relative is default
`
Output:
`
.
./dist
./public
./src
./test
`

---

$3

a. Exclude all test files (.test.) (Standard Mode)
`sh
cd my_project
phind --exclude ".test." # Relative is default
`
(Output will include everything except ./test/main.test.ts and ./test/utils.test.js)

b. Exclude the public directory (pruning) (Standard Mode)
`sh
cd my_project
phind --exclude public # Relative is default
`
(Output will include everything except ./public directory and its contents: ./public/index.html, ./public/styles.css)

c. Exclude CSS files and the dist directory (Standard Mode)
`sh
cd my_project
phind --exclude "*.css" --exclude dist # Relative is default
`
(Output will not include ./public/styles.css, the ./dist directory, or ./dist/bundle.js)

d. Find all JS files, but exclude those in the test directory (Standard Mode)
`sh
cd my_project
phind --name ".js" --exclude "test/*" -t f # Relative is default
`
Output:
`
./dist/bundle.js
`
(Note: ./test/utils.test.js is excluded because it's inside the test directory)

---

$3

a. Find readme.md case-insensitively (Standard Mode)
`sh
cd my_project


Assuming file is README.md


phind --name readme.md -i # Relative is default
`
Output:
`
./README.md
`

b. Find all TS/TSX files, excluding button.tsx case-insensitively (Standard Mode)
`sh
cd my_project
phind --name "*.[tT][sS]x?" -i --exclude "button.tsx" # Relative is default
`
Output:
`
./src/main.ts
./src/utils/helper.ts
./test/main.test.ts
`
(Note: ./src/components/Button.tsx is excluded)

---

$3

a. Run normally (assuming global file excludes dist/ and .env) (Standard Mode)
`sh
cd my_project


Assumes global ignore contains 'dist/' and '.env'


phind # Relative is default
`
(Output will exclude node_modules, .git (default), AND dist/, .env (global), paths will be relative)

b. Skip the global ignore file (Standard Mode)
`sh
cd my_project


Assumes global ignore contains 'dist/' and '.env'


phind --skip-global-ignore # Relative is default
`
(Output will exclude node_modules, .git (default) but will include ./dist/ and ./.env because the global file was skipped, paths will be relative)

c. Combine global, default, and CLI excludes (Standard Mode)
`sh
cd my_project


Assumes global ignore contains 'dist/' and '.env'


phind --exclude "*.yaml" --skip-global-ignore=false # Relative is default
`
(Output excludes node_modules, .git (default), dist/, .env (global), AND ./config.yaml (CLI), paths will be relative)

---

$3

a. Find items inside node_modules (Standard Mode)
By default, node_modules is pruned. To find something inside in standard mode, you need an explicit --name pattern that targets the content, which overrides the pruning for that specific target.
`sh
cd my_project


Find the specific library index file


phind --name "node_modules/some_lib/index.js" # Relative is default
`
Output:
`
./node_modules/some_lib/index.js
`
(The default exclusion of node_modules directory printing/pruning is overridden because a specific include pattern targets content inside it)

b. List the node_modules directory itself and its contents (Standard Mode)
`sh
cd my_project
phind --name node_modules --name "node_modules/**" # Relative is default
`
Output:
`
./node_modules
./node_modules/some_lib
./node_modules/some_lib/index.js
`
(Here, --name node_modules explicitly includes the directory, overriding the default exclusion for listing it, and --name "node_modules/" includes the contents)

---

AI Mode Examples (--ai)

These examples demonstrate the AI-powered search. Remember to set the GEMINI_API_KEY environment variable first. AI Mode interprets your query and filters the initially collected file list (which respects default and global excludes).

Pre-requisite:
`sh
export GEMINI_API_KEY="YOUR_API_KEY_HERE"
cd my_project
`

a. Find Configuration Files
`sh
phind --ai "configuration files"
`
Console Output (before results):
`
AI Mode activated. Query: "configuration files"
AI Mode: Collecting all file paths...
AI Mode: Collected 21 paths to analyze. # (Number may vary based on global ignores)
AI Mode: Sending request to Gemini...
AI Mode: Received response.
AI Mode: Gemini identified 3 relevant files.

AI identified the following relevant files:
`
Potential AI Results (output order may vary):
`
./config.yaml
./package.json
./.env
`
Explanation: The AI understands that .yaml, .json (especially package.json), and .env often store configuration.

b. Find Test Files
`sh
phind --ai "all test files"
`
Console Output (before results):
`
AI Mode activated. Query: "all test files"


... (collection/API logs) ...

AI Mode: Gemini identified 2 relevant files.

AI identified the following relevant files:
`
Potential AI Results:
`
./test/main.test.ts
./test/utils.test.js
`
Explanation: The AI recognizes .test.ts and .test.js extensions and the test/ directory as common indicators of test files.

c. Find React Components
`sh
phind --ai "react components"
`
Console Output (before results):
`
AI Mode activated. Query: "react components"


... (collection/API logs) ...

AI Mode: Gemini identified 1 relevant files.

AI identified the following relevant files:
`
Potential AI Results:
`
./src/components/Button.tsx
`
Explanation: The AI associates the .tsx extension and the components/ directory structure with React components.

d. Irrelevant Query
`sh
phind --ai "image files of cats"
`
Output:
`
AI Mode activated. Query: "image files of cats"


... (collection/API logs) ...


AI did not identify any relevant files based on your query.
`
Explanation: The AI couldn't find any files in the list matching the description "image files of cats".

---

Why phind? (Comparison Summary)

| Feature | phind (phind-cli) | Standard find (Unix-like) | Standard dir (Windows) |
| :------------------ | :----------------------------------------------- | :--------------------------------------------- | :------------------------------ |
| Platform | ✅ Windows, macOS, Linux | ✅ Linux, macOS (Syntax varies) | ✅ Windows Only |
| Consistency | ✅ Identical behavior everywhere | ❌ Syntax/options differ (GNU vs BSD) | N/A |
| Default Excludes| ✅ node_modules, .git, .gradle | ❌ None (Requires manual -prune / -path) | ❌ None |
| AI Search | ✅ Yes (Google Gemini via --ai) | ❌ No | ❌ No |
| Exclusion Syntax| ✅ Simple globs (--exclude) + Global Ignore | ⚠️ Complex (-path ... -prune -o ...) | ⚠️ Limited (/A:-D) |
| Inclusion Syntax| ✅ Simple globs (--name) | ✅ Powerful but complex patterns (-name, -regex) | ✅ Simple wildcards (*, ?) |
| Type Filter | ✅ Simple (-t f, -t d) | ✅ -type f, -type d | ⚠️ Requires attribute check (/A:D, /A:-D) |
| Depth Limit | ✅ Simple (-d N) | ✅ -maxdepth N | ❌ No direct equivalent |
| Case Ignore | ✅ Simple (-i) | ✅ -iname, -iregex | ✅ /S searches, but no case flag |
| Path Style | ✅ Relative Default (./file), --relative=false for absolute | ⚠️ Relative Default (./file), complex absolute | ✅ Relative Default (file) |
| Ease of Use | ⭐⭐⭐⭐⭐ (High, Developer-focused) | ⭐⭐ (Medium-High, Steeper curve) | ⭐⭐⭐ (Medium, Simpler but limited) |

Development

1. Clone the repository:
`sh
git clone https://github.com/nodesman/phind-cli.git
cd phind-cli
`
2. Install dependencies:
`sh
npm install
`
3. Build: Compile TypeScript to JavaScript:
`sh
npm run build
`
(Outputs to the bin directory)
4. Watch: Compile TypeScript in watch mode:
`sh
npm run watch
`
5. Run Tests: Execute Jest tests:
`sh
npm test
`
Run tests in watch mode:
`sh
npm run test:watch
`
Generate coverage report:
`sh
npm run test:coverage
`
6. Run Development Version: Execute the CLI directly using ts-node:
``sh
npm run dev -- [path] [options]
#