# LLM Context
[](https://opensource.org/licenses/Apache-2.0)
[](https://pypi.org/project/llm-context/)
LLM Context is a command-line tool designed to help developers efficiently copy and paste relevant context from code / text repositories into the web-chat interface of Large Language Models (LLMs). It leverages `.gitignore` patterns for smart file selection and uses the clipboard for seamless integration with LLM interfaces.
> **Note**: This project was developed in collaboration with Claude-3.5-Sonnet, using LLM Context itself to share code context during development. All code in the repository is human-curated (by me 😇, @restlessronin).
## Current Usage Patterns
- **LLM Integration**: Primarily used with Claude (Project Knowledge) and GPT (Knowledge), but supports all LLM chat interfaces.
- **Project Types**: Suitable for code repositories and collections of text/markdown/html documents.
- **Project Size**: Optimized for projects that fit within an LLM's context window. Large project support is in development.
## Installation
Use [pipx](https://pypa.github.io/pipx/) to install LLM Context:
```
pipx install llm-context
```
## Usage
LLM Context enables rapid project context updates for each AI chat.
### Quick Start and Typical Workflow
1. [Install LLM Context](#installation) if you haven't already.
2. Navigate to your project's root directory.
3. Run `lc-init` to set up LLM Context for your project (only needed once per repository).
4. For chat interfaces with built-in context storage (e.g., Claude Pro Projects, ChatGPT Plus GPTs):
- Set up your custom prompt manually in the chat interface.
- A default prompt is available in `.llm-context/templates/lc-prompt.md`.
5. (Optional) Edit `.llm-context/config.toml` to [add custom ignore patterns](#customizing-ignore-patterns).
6. Run `lc-sel-files` to select files for full content inclusion.
7. (Optional) [Review the selected file](#reviewing-selected-files) list in `.llm-context/curr_ctx.toml`.
8. Generate and copy the context:
- For chat interfaces with built-in storage: Run `lc-context`
- For other interfaces (including free plans): Run `lc-context --with-prompt` to include the default prompt
9. Paste the generated context:
- For interfaces with built-in storage: Into the Project Knowledge (Claude Pro) or GPT Knowledge (ChatGPT Plus) section
- For other interfaces: Into the system message or the first chat message, as appropriate
10. Start your conversation with the LLM about your project.
To maintain up-to-date AI assistance:
- Repeat steps 6-9 at the start of each new chat session. This process takes only seconds.
- For interfaces with built-in storage, update your custom prompt separately if needed.
### Handling LLM File Requests
When the LLM asks for a file that isn't in the current context:
1. Copy the LLM's file request (typically in a markdown block) to your clipboard.
2. Run `lc-read-cliplist` to generate the content of the requested files.
3. Paste the generated file contents back into your chat with the LLM.
### Configuration
#### Customizing Ignore Patterns
Add custom ignore patterns in `.llm-context/config.toml` to exclude specific files or directories not covered by your project's `.gitignore`. This is useful for versioned files that don't contribute to code context, such as media files, large generated files, detailed changelogs, or environment-specific configurations.
Example:
```toml
# /.llm-context/config.toml
[gitignores]
full_files = [
"*.svg",
"*.png",
"CHANGELOG.md",
".env",
# Add more patterns here
]
```
#### Reviewing Selected Files
Review the list of selected files in `.llm-context/curr_ctx.toml` to check what's included in the context. This is particularly useful when trying to minimize context size.
```toml
# /.llm-context/curr_ctx.toml
[context]
full = [
"/llm-context.py/pyproject.toml",
# more files ...
]
```
## Command Reference
- `lc-init`: Initialize LLM Context for your project (only needed once per repository)
- `lc-sel-files`: Select files for full content inclusion
- `lc-sel-outlines`: Select files for outline inclusion (experimental)
- `lc-context`: Generate and copy context to clipboard
- Use `--with-prompt` flag to include the prompt for LLMs without built-in storage
- `lc-read-cliplist`: Read contents for LLM-requested files, and copy to clipboard
## Experimental: Handling Larger Repositories
For larger projects, we're exploring a combined approach of full file content and file outlines. Use `lc-sel-outlines` after `lc-sel-files` to experiment with this feature.
**Note:** The outlining feature currently supports the following programming languages:
C, C++, C#, Elisp, Elixir, Elm, Go, Java, JavaScript, OCaml, PHP, Python, QL, Ruby, Rust, and TypeScript. Files in unsupported languages will not be outlined and will be excluded from the outline selection.
### Feedback and Contributions
We welcome feedback, issue reports, and pull requests on our [GitHub repository](https://github.com/cyberchitta/llm-context.py).
## Acknowledgments
LLM Context evolves from a lineage of AI-assisted development tools:
- This project succeeds [LLM Code Highlighter](https://github.com/restlessronin/llm-code-highlighter), a TypeScript library I developed for IDE integration.
- The concept originated from my work on [RubberDuck](https://github.com/rubberduck-ai/rubberduck-vscode) and continued with later contributions to [Continue](https://github.com/continuedev/continuedev).
- LLM Code Highlighter was heavily inspired by [Aider Chat](https://github.com/paul-gauthier/aider). I worked with GPT-4 to translate several Aider Chat Python modules into TypeScript, maintaining functionality while restructuring the code.
- This project uses tree-sitter [tag query files](src/llm_context/highlighter/tag-qry/) from Aider Chat.
- LLM Context exemplifies the power of AI-assisted development, transitioning from Python to TypeScript and back to Python with the help of GPT-4 and Claude-3.5-Sonnet.
I am grateful for the open-source community's innovations and the AI assistance that have shaped this project's evolution.
I am grateful for the help of Claude-3.5-Sonnet in the development of this project.
## License
This project is licensed under the Apache License, Version 2.0. See the [LICENSE](LICENSE) file for details.
Raw data
{
"_id": null,
"home_page": "https://github.com/cyberchitta/llm-context.py",
"name": "llm-context",
"maintainer": null,
"docs_url": null,
"requires_python": "<4.0,>=3.9",
"maintainer_email": null,
"keywords": "llm, ai, context, code, clipboard, chat",
"author": "restlessronin",
"author_email": "88921269+restlessronin@users.noreply.github.com",
"download_url": "https://files.pythonhosted.org/packages/a0/b6/237375a325e4f22cb255c374b4d11e3e07a42c76c43cfea82a11512899ff/llm_context-0.1.0.tar.gz",
"platform": null,
"description": "# LLM Context\n\n[](https://opensource.org/licenses/Apache-2.0)\n[](https://pypi.org/project/llm-context/)\n\nLLM Context is a command-line tool designed to help developers efficiently copy and paste relevant context from code / text repositories into the web-chat interface of Large Language Models (LLMs). It leverages `.gitignore` patterns for smart file selection and uses the clipboard for seamless integration with LLM interfaces.\n\n> **Note**: This project was developed in collaboration with Claude-3.5-Sonnet, using LLM Context itself to share code context during development. All code in the repository is human-curated (by me \ud83d\ude07, @restlessronin).\n\n## Current Usage Patterns\n\n- **LLM Integration**: Primarily used with Claude (Project Knowledge) and GPT (Knowledge), but supports all LLM chat interfaces.\n- **Project Types**: Suitable for code repositories and collections of text/markdown/html documents.\n- **Project Size**: Optimized for projects that fit within an LLM's context window. Large project support is in development.\n\n## Installation\n\nUse [pipx](https://pypa.github.io/pipx/) to install LLM Context:\n\n```\npipx install llm-context\n```\n\n## Usage\n\nLLM Context enables rapid project context updates for each AI chat.\n\n### Quick Start and Typical Workflow\n\n1. [Install LLM Context](#installation) if you haven't already.\n2. Navigate to your project's root directory.\n3. Run `lc-init` to set up LLM Context for your project (only needed once per repository).\n4. For chat interfaces with built-in context storage (e.g., Claude Pro Projects, ChatGPT Plus GPTs):\n - Set up your custom prompt manually in the chat interface.\n - A default prompt is available in `.llm-context/templates/lc-prompt.md`.\n5. (Optional) Edit `.llm-context/config.toml` to [add custom ignore patterns](#customizing-ignore-patterns).\n6. Run `lc-sel-files` to select files for full content inclusion.\n7. (Optional) [Review the selected file](#reviewing-selected-files) list in `.llm-context/curr_ctx.toml`.\n8. Generate and copy the context:\n - For chat interfaces with built-in storage: Run `lc-context`\n - For other interfaces (including free plans): Run `lc-context --with-prompt` to include the default prompt\n9. Paste the generated context:\n - For interfaces with built-in storage: Into the Project Knowledge (Claude Pro) or GPT Knowledge (ChatGPT Plus) section\n - For other interfaces: Into the system message or the first chat message, as appropriate\n10. Start your conversation with the LLM about your project.\n\nTo maintain up-to-date AI assistance:\n- Repeat steps 6-9 at the start of each new chat session. This process takes only seconds.\n- For interfaces with built-in storage, update your custom prompt separately if needed.\n\n### Handling LLM File Requests\n\nWhen the LLM asks for a file that isn't in the current context:\n\n1. Copy the LLM's file request (typically in a markdown block) to your clipboard.\n2. Run `lc-read-cliplist` to generate the content of the requested files.\n3. Paste the generated file contents back into your chat with the LLM.\n\n### Configuration\n\n#### Customizing Ignore Patterns\n\nAdd custom ignore patterns in `.llm-context/config.toml` to exclude specific files or directories not covered by your project's `.gitignore`. This is useful for versioned files that don't contribute to code context, such as media files, large generated files, detailed changelogs, or environment-specific configurations.\n\nExample:\n\n```toml\n# /.llm-context/config.toml\n[gitignores]\nfull_files = [\n \"*.svg\",\n \"*.png\",\n \"CHANGELOG.md\",\n \".env\",\n # Add more patterns here\n]\n```\n\n#### Reviewing Selected Files\n\nReview the list of selected files in `.llm-context/curr_ctx.toml` to check what's included in the context. This is particularly useful when trying to minimize context size.\n\n```toml\n# /.llm-context/curr_ctx.toml\n[context]\nfull = [\n \"/llm-context.py/pyproject.toml\",\n # more files ...\n]\n```\n\n## Command Reference\n\n- `lc-init`: Initialize LLM Context for your project (only needed once per repository)\n- `lc-sel-files`: Select files for full content inclusion\n- `lc-sel-outlines`: Select files for outline inclusion (experimental)\n- `lc-context`: Generate and copy context to clipboard\n - Use `--with-prompt` flag to include the prompt for LLMs without built-in storage\n- `lc-read-cliplist`: Read contents for LLM-requested files, and copy to clipboard\n\n## Experimental: Handling Larger Repositories\n\nFor larger projects, we're exploring a combined approach of full file content and file outlines. Use `lc-sel-outlines` after `lc-sel-files` to experiment with this feature.\n\n**Note:** The outlining feature currently supports the following programming languages:\nC, C++, C#, Elisp, Elixir, Elm, Go, Java, JavaScript, OCaml, PHP, Python, QL, Ruby, Rust, and TypeScript. Files in unsupported languages will not be outlined and will be excluded from the outline selection.\n\n### Feedback and Contributions\n\nWe welcome feedback, issue reports, and pull requests on our [GitHub repository](https://github.com/cyberchitta/llm-context.py).\n\n## Acknowledgments\n\nLLM Context evolves from a lineage of AI-assisted development tools:\n\n- This project succeeds [LLM Code Highlighter](https://github.com/restlessronin/llm-code-highlighter), a TypeScript library I developed for IDE integration.\n- The concept originated from my work on [RubberDuck](https://github.com/rubberduck-ai/rubberduck-vscode) and continued with later contributions to [Continue](https://github.com/continuedev/continuedev).\n- LLM Code Highlighter was heavily inspired by [Aider Chat](https://github.com/paul-gauthier/aider). I worked with GPT-4 to translate several Aider Chat Python modules into TypeScript, maintaining functionality while restructuring the code.\n- This project uses tree-sitter [tag query files](src/llm_context/highlighter/tag-qry/) from Aider Chat.\n- LLM Context exemplifies the power of AI-assisted development, transitioning from Python to TypeScript and back to Python with the help of GPT-4 and Claude-3.5-Sonnet.\n\nI am grateful for the open-source community's innovations and the AI assistance that have shaped this project's evolution.\n\nI am grateful for the help of Claude-3.5-Sonnet in the development of this project.\n\n## License\n\nThis project is licensed under the Apache License, Version 2.0. See the [LICENSE](LICENSE) file for details.\n",
"bugtrack_url": null,
"license": "Apache-2.0",
"summary": "A command-line tool for copying code context to clipboard for use in LLM chats",
"version": "0.1.0",
"project_urls": {
"Homepage": "https://github.com/cyberchitta/llm-context.py",
"Repository": "https://github.com/cyberchitta/llm-context.py"
},
"split_keywords": [
"llm",
" ai",
" context",
" code",
" clipboard",
" chat"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "24d134384f0886e1d79cacfb11883a145616a38a92cf321344c7eca7189f41cb",
"md5": "5f158afcffc4ef355f076d93d14bdc4d",
"sha256": "9bc0cee67f748b137b9d48e639fb84f38fe2c0c691e85caf05db42db6338a204"
},
"downloads": -1,
"filename": "llm_context-0.1.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "5f158afcffc4ef355f076d93d14bdc4d",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<4.0,>=3.9",
"size": 33913,
"upload_time": "2024-09-20T15:53:31",
"upload_time_iso_8601": "2024-09-20T15:53:31.361616Z",
"url": "https://files.pythonhosted.org/packages/24/d1/34384f0886e1d79cacfb11883a145616a38a92cf321344c7eca7189f41cb/llm_context-0.1.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "a0b6237375a325e4f22cb255c374b4d11e3e07a42c76c43cfea82a11512899ff",
"md5": "e0c808d0f1a5a24a4836420a9b0e4776",
"sha256": "8738f33322f441b7023d811475e7dc06b883928cddc18bf11159a610e8854e2e"
},
"downloads": -1,
"filename": "llm_context-0.1.0.tar.gz",
"has_sig": false,
"md5_digest": "e0c808d0f1a5a24a4836420a9b0e4776",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<4.0,>=3.9",
"size": 26785,
"upload_time": "2024-09-20T15:53:33",
"upload_time_iso_8601": "2024-09-20T15:53:33.718796Z",
"url": "https://files.pythonhosted.org/packages/a0/b6/237375a325e4f22cb255c374b4d11e3e07a42c76c43cfea82a11512899ff/llm_context-0.1.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-09-20 15:53:33",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "cyberchitta",
"github_project": "llm-context.py",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "llm-context"
}
JSON
