ctxpack 0.1.0

Automatically gather project context and generate Markdown prompt packages for LLMs

ctxpack

Automatically gather project context and generate Markdown prompt packages for LLMs.

Stop manually copy-pasting files into ChatGPT. ctxpack intelligently selects the most relevant files from your project and packages them into a single Markdown document ready for any LLM.

Installation

pip install ctxpack

Quick Start

# Initialize ctxpack in your project
cd my-project
ctxpack init

# Generate a context package
ctxpack pack

# Generate with smart mode (import/reference analysis)
ctxpack pack --mode smart

# Watch for changes and auto-regenerate
ctxpack watch --out context.md

Commands

ctxpack init

Creates a .ctxpack/ configuration folder and .ctxpackignore file with sensible defaults.

ctxpack init          # Initialize in current directory
ctxpack init --force  # Overwrite existing configuration

ctxpack pack [PATH]

Scans your project, ranks files by relevance, and outputs a Markdown context package.

ctxpack pack                           # Pack current directory
ctxpack pack ./src                     # Pack specific path
ctxpack pack --max-files 10            # Limit to 10 files
ctxpack pack --max-bytes 50000         # Limit to 50KB
ctxpack pack --extensions .py,.js      # Only Python and JavaScript
ctxpack pack --include-tests           # Include test files
ctxpack pack --mode smart              # Use import graph analysis
ctxpack pack --out context.md          # Output to file
ctxpack pack --verbose                 # Show scoring details

ctxpack watch [PATH]

Watches for file changes and auto-regenerates the context package.

ctxpack watch --out context.md         # Watch and output to file
ctxpack watch --debounce 2.0           # Wait 2 seconds after changes

Configuration

After running ctxpack init, you can customize .ctxpack/config.toml:

[pack]
max_files = 20
max_bytes = 100000
extensions = [".py", ".ts", ".js", ".md"]
exclude_tests = true
mode = "simple"

[scoring]
recency_weight = 0.7
size_weight = 0.3

Selection Modes

• Simple mode (default): Ranks files by recency and size
• Smart mode: Adds import/reference graph analysis to boost frequently-used files

License

MIT