🧠 gpdevmate CLI — A Simple AI Code Review from Terminal

The gpdevmate CLI allows developers to run AI-powered code reviews directly from the terminal, using local source files and LLMs like GPT-4 or CodeLlama.

It’s part of the gpdevmate project, a project by @sgprakas, aiming to build AI tooling for developers.

💡 Great for devs or learners — get AI-powered feedback on your code
to help you learn, understand, and improve — not just generate.

⚠️ Note: This is an early preview (v0.1.*).
While gpdevmate provides useful suggestions, they may not always be accurate or context-aware.
Use it as a guide — not a source of truth.

🚧 In future versions, gpdevmate aims to learn from your codebase
to provide more personalized and context-aware reviews.

🚀 Installation

npm install -g @sgprakas/gpdevmate

You can now run gpdevmate from anywhere.

⚡ Quick Start

1. Configure the LLM provider (interactive):

gpdevmate config --update

This will walk you through selecting and setting up your model provider (e.g., OpenAI or Ollama). Passing --provider option only will update the default provider. Passing both --provider and --update will update the default provider and also prompt for config to update.

2. Run your first review:

gpdevmate review ./src --limit 5 --output-format markdown

This will review up to 5 files from the ./src directory and print/save the output in Markdown format.

⚙️ CLI Usage

gpdevmate [options] [command]

Available Commands:

| Command | Description | | ----------------- | -------------------------------------------------------------- | | review <target> | Run code review on a directory or file | | config | Set model provider and its configuration (OpenAI, Ollama, etc) | | check | Check whether the current model provider is reachable | | help [command] | Show help information for a specific command |

Global Options:

| Option | Description | | --------------- | ---------------------- | | -V, --version | Output the CLI version | | -h, --help | Display help menu |

🔍 review Command Usage

gpdevmate review [options] <target>

Arguments:

| Argument | Description | | ---------- | ------------------------------- | | <target> | Path to the directory to review |

Options:

| Option | Description | | --------------------------------------- | ----------------------------------------------------------- | | --provider <provider> | LLM Provider: openai | ollama (default: openai) | | --limit <number> | Max number of files to review (default: 5) | | --skip-dirs <dirs> | Comma-separated list of directories to skip | | --dry-run | Show which files are going to be reviewed | | --of, --output-format <output_format> | Output format: console | markdown (default: console) | | -o, --output <output> | Output directory to save the review as a .md file | | -h, --help | Display help for the review command |

⚙️ config Command Usage

gpdevmate config [options]

Description:

Configure and update the default LLM provider and its settings.

Options:

| Option | Description | | ----------------------- | ----------------------------------------------------- | | --provider <provider> | Set default LLM Provider: openai | ollama | | --update | Update configuration values for the selected provider | | -h, --help | Display help for the config command |

🧠 Example Output

File: sample.ts

export async function handleRequest(req, res) {
    const userData = await getUserData(req.params.id)

    if (userData === null) {
        res.send('User not found')
    }
    else {
        res.send(userData)
    }
}

async function getUserData(id) {
    let db = connectToDb()
    let result

    try {
        result = await db.query("SELECT * FROM users WHERE id = " + id)
    } catch (err) {
        console.log(err)
    }

    return result.rows[0]
}

function connectToDb() {
    return {
        async query(sql) {
            return { rows: [{ id: 1, name: "John" }] }
        }
    }
}

Suggestions:
• SQL Injection Risk: The current implementation of `getUserData` is vulnerable to SQL injection. Use parameterized queries to prevent this.
• Error Handling: The error handling in `getUserData` is insufficient. Consider returning a meaningful error response to the client if the database query fails.
• Return Type: Specify the return type for `getUserData` and `handleRequest` functions for better type safety.
• Response Status Codes: Use appropriate HTTP status codes when sending responses (e.g., 404 for "User not found").

Improved Code Example:

export async function handleRequest(req, res) {
    try {
        const userData = await getUserData(req.params.id);
        if (!userData) {
            res.status(404).send('User not found');
        } else {
            res.status(200).send(userData);
        }
    } catch (error) {
        console.error('Error handling request:', error);
        res.status(500).send('Internal Server Error');
    }
}

async function getUserData(id: string): Promise<any | null> {
    const db = connectToDb();
    try {
        const result = await db.query("SELECT * FROM users WHERE id = $1", [id]); // Use parameterized query
        return result.rows[0] || null;
    } catch (err) {
        console.error('Database query error:', err);
        throw new Error('Database query failed');
    }
}

function connectToDb() {
    return {
        async query(sql: string, params: any[]) {
            // Simulate a database query with parameters
            return { rows: [{ id: 1, name: "John" }] };
        }
    };
}

🙋‍♂️ Author

Created by SG Prakash Fullstack dev building AI tools to learn & grow 🚀

🐛 Found a Bug or Issue?

If you’ve found a bug or something isn’t working as expected:

👉 Create an issue here
Help us improve by sharing details — even small issues matter!

🗣 Got Ideas or Fixes?

We’d love your contributions!

⚠️ Before opening a pull request, please create an issue first — even for small changes.

This gives us a chance to:

• Understand the context
• Discuss implementation
• Avoid duplicate work

➡️ Create an issue