codegpt


Namecodegpt JSON
Version 0.3.1 PyPI version JSON
download
home_pagehttps://github.com/morganpartee/codegpt
SummaryA CLI tool for developers using GPT-3.
upload_time2022-12-23 19:12:29
maintainer
docs_urlNone
authorJohn Partee
requires_python>=3.8,<4.0
licenseGPL-3.0-or-later
keywords
VCS
bugtrack_url
requirements No requirements were recorded.
Travis-CI No Travis.
coveralls test coverage No coveralls.
            # Codegpt

## 0.3

A tool for using GPT just a little quicker. A nearly truly automated footgun. Learn how to revert with git before trying please.

Posting about progress here:

[![Twitter Follow](https://img.shields.io/twitter/follow/_JohnPartee?style=social)](https://twitter.com/_JohnPartee)

## Getting Started

`pip install codegpt --upgrade`

And set your openapi API key as an environment variable like they recommend:
[In their docs here](https://help.openai.com/en/articles/5112595-best-practices-for-api-key-safety)

Windows users can also use `setx` like:

`$ setx OPENAI_SECRET_KEY=<YOUR_API_KEY>`

from an admin console.

## Your first (safe) command

One cool thing is generating documentation. GPT-3 has a token limit of 4000 for completions, so larger files will be chunked up.

```bash
codegpt docs <paths>
```

And it'll generate docs, one per file. This is great when you're coming into a codebase you've never seen before.

## Unsafe Commands

Everything else can modify files. Have someone hold your beer and try some of these (after you check it into git):

Usage
To try Codegpt, you can run the following command:

```bash
codegpt do <instructions (quoted)> -f readme.md 
```

It can do basically anything. Try handing in some files for context and telling it to generate something new - SQL queries, new features, documentation, whatever.

Or use the quick command to do some neat stuff, like:

Generate docs

```bash
codegpt quick docs <filenames>
```

Change variable names to be more readable

```bash
codegpt quick varnames <filenames>
```

Add comments to your code automatically

```bash
codegpt quick comment <filenames>
```

Check for bugs (iffy, but worth a shot)

```bash
codegpt quick bugs <filenames>
```

Check for vulnerabilities (even more iffy, but worth a shot)

```bash
codegpt quick vulns <filenames>
```

Try to make code less miserable

```bash
codegpt quick ugh <filenames>
```

Keep in mind that using GPT-3 for code generation is paid, with a cost of 2 cents per 1,000 tokens.

Just like with a Jr Dev, it's best to break up your tasks into smaller pieces to improve the results.

            

Raw data

            {
    "_id": null,
    "home_page": "https://github.com/morganpartee/codegpt",
    "name": "codegpt",
    "maintainer": "",
    "docs_url": null,
    "requires_python": ">=3.8,<4.0",
    "maintainer_email": "",
    "keywords": "",
    "author": "John Partee",
    "author_email": "",
    "download_url": "https://files.pythonhosted.org/packages/fb/bb/7db2ddf545ba1bcb4587e290189b9501089dcba5f2e053dd6ef7ae0d29c0/codegpt-0.3.1.tar.gz",
    "platform": null,
    "description": "# Codegpt\n\n## 0.3\n\nA tool for using GPT just a little quicker. A nearly truly automated footgun. Learn how to revert with git before trying please.\n\nPosting about progress here:\n\n[![Twitter Follow](https://img.shields.io/twitter/follow/_JohnPartee?style=social)](https://twitter.com/_JohnPartee)\n\n## Getting Started\n\n`pip install codegpt --upgrade`\n\nAnd set your openapi API key as an environment variable like they recommend:\n[In their docs here](https://help.openai.com/en/articles/5112595-best-practices-for-api-key-safety)\n\nWindows users can also use `setx` like:\n\n`$ setx OPENAI_SECRET_KEY=<YOUR_API_KEY>`\n\nfrom an admin console.\n\n## Your first (safe) command\n\nOne cool thing is generating documentation. GPT-3 has a token limit of 4000 for completions, so larger files will be chunked up.\n\n```bash\ncodegpt docs <paths>\n```\n\nAnd it'll generate docs, one per file. This is great when you're coming into a codebase you've never seen before.\n\n## Unsafe Commands\n\nEverything else can modify files. Have someone hold your beer and try some of these (after you check it into git):\n\nUsage\nTo try Codegpt, you can run the following command:\n\n```bash\ncodegpt do <instructions (quoted)> -f readme.md \n```\n\nIt can do basically anything. Try handing in some files for context and telling it to generate something new - SQL queries, new features, documentation, whatever.\n\nOr use the quick command to do some neat stuff, like:\n\nGenerate docs\n\n```bash\ncodegpt quick docs <filenames>\n```\n\nChange variable names to be more readable\n\n```bash\ncodegpt quick varnames <filenames>\n```\n\nAdd comments to your code automatically\n\n```bash\ncodegpt quick comment <filenames>\n```\n\nCheck for bugs (iffy, but worth a shot)\n\n```bash\ncodegpt quick bugs <filenames>\n```\n\nCheck for vulnerabilities (even more iffy, but worth a shot)\n\n```bash\ncodegpt quick vulns <filenames>\n```\n\nTry to make code less miserable\n\n```bash\ncodegpt quick ugh <filenames>\n```\n\nKeep in mind that using GPT-3 for code generation is paid, with a cost of 2 cents per 1,000 tokens.\n\nJust like with a Jr Dev, it's best to break up your tasks into smaller pieces to improve the results.\n",
    "bugtrack_url": null,
    "license": "GPL-3.0-or-later",
    "summary": "A CLI tool for developers using GPT-3.",
    "version": "0.3.1",
    "split_keywords": [],
    "urls": [
        {
            "comment_text": "",
            "digests": {
                "md5": "aa7330810943eb73806545445bfa95c7",
                "sha256": "7f5a0bc10aa610fe8a840ac5d75aee37bd51543e3b8234a74fc05967562733d1"
            },
            "downloads": -1,
            "filename": "codegpt-0.3.1-py3-none-any.whl",
            "has_sig": false,
            "md5_digest": "aa7330810943eb73806545445bfa95c7",
            "packagetype": "bdist_wheel",
            "python_version": "py3",
            "requires_python": ">=3.8,<4.0",
            "size": 22384,
            "upload_time": "2022-12-23T19:12:27",
            "upload_time_iso_8601": "2022-12-23T19:12:27.604216Z",
            "url": "https://files.pythonhosted.org/packages/91/d3/940fcfe71bd0082827bf8aef734a21715651735a6cb2bf01f6b6b18d162e/codegpt-0.3.1-py3-none-any.whl",
            "yanked": false,
            "yanked_reason": null
        },
        {
            "comment_text": "",
            "digests": {
                "md5": "d101dc66f572b3ef734fad8cde7c3c02",
                "sha256": "5cf7a553e91f814283c9198e2a6edc6827e1997a52b9da1efadfd6e95c31f542"
            },
            "downloads": -1,
            "filename": "codegpt-0.3.1.tar.gz",
            "has_sig": false,
            "md5_digest": "d101dc66f572b3ef734fad8cde7c3c02",
            "packagetype": "sdist",
            "python_version": "source",
            "requires_python": ">=3.8,<4.0",
            "size": 21375,
            "upload_time": "2022-12-23T19:12:29",
            "upload_time_iso_8601": "2022-12-23T19:12:29.259394Z",
            "url": "https://files.pythonhosted.org/packages/fb/bb/7db2ddf545ba1bcb4587e290189b9501089dcba5f2e053dd6ef7ae0d29c0/codegpt-0.3.1.tar.gz",
            "yanked": false,
            "yanked_reason": null
        }
    ],
    "upload_time": "2022-12-23 19:12:29",
    "github": true,
    "gitlab": false,
    "bitbucket": false,
    "github_user": "morganpartee",
    "github_project": "codegpt",
    "travis_ci": false,
    "coveralls": false,
    "github_actions": false,
    "lcname": "codegpt"
}
        
Elapsed time: 0.30141s