sourceprompt

SourcePrompt

Преобразует вашу кодовую базу в подсказку, которую вы можете подавать в LLM.

Короче говоря: он сканирует все текстовые файлы в указанном каталоге / GIT Repostiory и помещает их контент (вместе с их пути) в структурированную отметки.

По умолчанию он предоставляет предопределенную подсказку, которая может быть опущена с помощью -r или заменена с помощью параметров -p .

Установка

Скачать двоичный файл из выпусков или установить, используя:

go install github.com/rostrovsky/sourceprompt@latest

Игнорируемые файлы / пути

• Файлы/каталоги начинаются с .
• Двоичные файлы

Флаги

• -i, --include -режима режима путей, которые должны быть включены.
• -e, --exclude -режима режима путей, которые должны быть исключены.
• -o, --output -Путь выходного файла. Когда он не указан, выход будет напечатан в stdout .
• -p, --prompt -Приглашение файла или URL -адрес. Позволяет указать пользовательскую подсказку, которая будет поставлена ​​в начале вывода. Если не указано, будет использоваться подсказка по умолчанию.
• -r, --raw -удаляет приглашение с вывода.
• -v, --verbose -Отпечатает информацию отладки при включении.

Примеры

Вывод в Stdout

 # output to stdout
sourceprompt /path/to/dir
sourceprompt /path/to/dir -v # debug
sourceprompt https://github.com/some/repo

# use remote git repo for codebase and remote prompt
sourceprompt https://github.com/some/repo -p https://raw.githubusercontent.com/another/repo/prompt.md -o out.md

# output to stdout without default prompt
sourceprompt /path/to/dir -r

# output to file
sourceprompt /path/to/dir -o out.md

# output to file with custom prompt
sourceprompt /path/to/dir -o out.md -p my_prompt.txt

# include only src/ files
sourceprompt /path/to/dir -o out.md -i ' ^src '

# include only C source files
sourceprompt /path/to/dir -o out.md -i ' .c$ ' -i ' .h$ '

# exclude markdown files
sourceprompt /path/to/dir -o out.md -e ' .md$ '

# exclude multiple types
sourceprompt /path/to/dir -o out.md -e ' .md$ ' -e ' .txt$ ' -e ' LICENSE ' 

Подсказка по умолчанию

 You will be provided with a markdown text (under the "---" separator) containing the contents of a codebase. Each code snippet will be enclosed in code fences, along with the corresponding file name. Your task is to analyze the codebase and gain a comprehensive understanding of its structure, functionality, and key features.

Please follow these steps:

1. Read through the entire codebase carefully, paying attention to the file names and the code within each code fence.
2. Identify the main components, modules, or classes of the codebase and their responsibilities. Summarize the purpose and functionality of each significant component.
3. Analyze the relationships and dependencies between different parts of the codebase. Identify any important interactions, data flow, or control flow between the components.
4. Extract the most important features and functionalities implemented in the codebase. Highlight any critical algorithms, data structures, or design patterns used.
5. Consider the overall architecture and design of the codebase. Identify any architectural patterns or principles followed, such as MVC, MVVM, or microservices.
6. Evaluate the code quality, readability, and maintainability. Note any areas that could be improved or any potential issues or vulnerabilities.
7. Provide a summary of your analysis, including the key insights, strengths, and weaknesses of the codebase. Offer suggestions for improvements or optimizations, if applicable.
8. Based on your understanding of the codebase, provide guidance on how AI agents can effectively operate across the entire codebase. Identify the entry points, important functions, or APIs that the agents should focus on for interaction and manipulation.
9. Discuss any specific considerations or challenges that AI agents may face when working with this codebase, such as dependencies, external libraries, or platform-specific requirements.
10. Conclude your analysis by providing a high-level overview of the codebase's functionality, architecture, and potential use cases. Highlight any notable features or aspects that make this codebase unique or valuable.

Your analysis should be thorough, insightful, and aimed at enabling AI agents to effectively understand and operate within the given codebase. Provide clear explanations and examples to support your findings and recommendations.

---