code2prompt
is a command-line tool (CLI) that converts your codebase into a single LLM prompt with a source tree, prompt templating, and token counting.
You can run this tool on the entire directory and it would generate a well-formatted Markdown prompt detailing the source tree structure, and all the code. You can then upload this document to either GPT or Claude models with higher context windows and ask it to:
.gitignore
.You can customize the prompt template to achieve any of the desired use cases. It essentially traverses a codebase and creates a prompt with all source files combined. In short, it automates copy-pasting multiple source files into your prompt and formatting them along with letting you know how many tokens your code consumes.
Download the latest binary for your OS from Releases OR install with cargo
:
cargo install code2prompt
For unpublished builds:
cargo install --git https://github.com/mufeedvh/code2prompt
For building code2prompt
from source, you need to have these tools installed:
git clone https://github.com/mufeedvh/code2prompt.git
cd code2prompt/
cargo build --release
The first command clones the code2prompt
repository to your local machine. The next two commands change into the code2prompt
directory and build it in release mode.
Generate a prompt from a codebase directory:
code2prompt path/to/codebase
Use a custom Handlebars template file:
code2prompt path/to/codebase -t path/to/template.hbs
Filter files using glob patterns:
code2prompt path/to/codebase --include="*.rs,*.toml"
Exclude files using glob patterns:
code2prompt path/to/codebase --exclude="*.txt,*.md"
Exclude files/folders from the source tree based on exclude patterns:
code2prompt path/to/codebase --exclude="*.npy,*.wav" --exclude-from-tree
Display the token count of the generated prompt:
code2prompt path/to/codebase --tokens
Specify a tokenizer for token count:
code2prompt path/to/codebase --tokens --encoding=p50k
Supported tokenizers: cl100k
, p50k
, p50k_edit
, r50k_bas
.
[!NOTE]
See Tokenizers for more details.
Save the generated prompt to an output file:
code2prompt path/to/codebase --output=output.txt
Print output as JSON:
code2prompt path/to/codebase --json
The JSON output will have the following structure:
{
"prompt": "<Generated Prompt>",
"directory_name": "codebase",
"token_count": 1234,
"model_info": "ChatGPT models, text-embedding-ada-002",
"files": []
}
Generate a Git commit message (for staged files):
code2prompt path/to/codebase --diff -t templates/write-git-commit.hbs
Generate a Pull Request with branch comparing (for staged files):
code2prompt path/to/codebase --git-diff-branch 'main, development' --git-log-branch 'main, development' -t templates/write-github-pull-request.hbs
Add line numbers to source code blocks:
code2prompt path/to/codebase --line-number
Disable wrapping code inside markdown code blocks:
code2prompt path/to/codebase --no-codeblock
I initially wrote this for personal use to utilize Claude 3.0's 200K context window and it has proven to be pretty useful so I decided to open-source it!
code2prompt
comes with a set of built-in templates for common use cases. You can find them in the templates
directory.
document-the-code.hbs
Use this template to generate prompts for documenting the code. It will add documentation comments to all public functions, methods, classes and modules in the codebase.
find-security-vulnerabilities.hbs
Use this template to generate prompts for finding potential security vulnerabilities in the codebase. It will look for common security issues and provide recommendations on how to fix or mitigate them.
clean-up-code.hbs
Use this template to generate prompts for cleaning up and improving the code quality. It will look for opportunities to improve readability, adherence to best practices, efficiency, error handling, and more.
fix-bugs.hbs
Use this template to generate prompts for fixing bugs in the codebase. It will help diagnose issues, provide fix suggestions, and update the code with proposed fixes.
write-github-pull-request.hbs
Use this template to create GitHub pull request description in markdown by comparing the git diff and git log of two branches.
write-github-readme.hbs
Use this template to generate a high-quality README file for the project, suitable for hosting on GitHub. It will analyze the codebase to understand its purpose and functionality, and generate the README content in Markdown format.
write-git-commit.hbs
Use this template to generate git commits from the staged files in your git directory. It will analyze the codebase to understand its purpose and functionality, and generate the git commit message content in Markdown format.
improve-performance.hbs
Use this template to generate prompts for improving the performance of the codebase. It will look for optimization opportunities, provide specific suggestions, and update the code with the changes.
You can use these templates by passing the -t
flag followed by the path to the template file. For example:
code2prompt path/to/codebase -t templates/document-the-code.hbs
code2prompt
supports the use of user defined variables in the Handlebars templates. Any variables in the template that are not part of the default context (absolute_code_path
, source_tree
, files
) will be treated as user defined variables.
During prompt generation, code2prompt
will prompt the user to enter values for these user defined variables. This allows for further customization of the generated prompts based on user input.
For example, if your template includes {{challenge_name}}
and {{challenge_description}}
, you will be prompted to enter values for these variables when running code2prompt
.
This feature enables creating reusable templates that can be adapted to different scenarios based on user provided information.
Tokenization is implemented using tiktoken-rs
. tiktoken
supports these encodings used by OpenAI models:
Encoding name | OpenAI models |
---|---|
cl100k_base |
ChatGPT models, text-embedding-ada-002 |
p50k_base |
Code models, text-davinci-002 , text-davinci-003 |
p50k_edit |
Use for edit models like text-davinci-edit-001 , code-davinci-edit-001 |
r50k_base (or gpt2 ) |
GPT-3 models like davinci |
For more context on the different tokenizers, see the OpenAI Cookbook
code2prompt
makes it easy to generate prompts for LLMs from your codebase. It traverses the directory, builds a tree structure, and collects information about each file. You can customize the prompt generation using Handlebars templates. The generated prompt is automatically copied to your clipboard and can also be saved to an output file. code2prompt
helps streamline the process of creating LLM prompts for code analysis, generation, and other tasks.
Ways to contribute:
Licensed under the MIT License, see LICENSE for more information.
If you liked the project and found it useful, please give it a :star: and consider supporting the authors!