kopipasta 0.54.0

A CLI tool to generate prompts with project structure and file contents

kopipasta

kopipasta bridges the gap between your local file system and LLM context windows.

A CLI tool for taking full, transparent control of your prompt. No black boxes.

➜  ~ kopipasta

  📁 Project Files
  |-- 📂 src/
  |   |-- ● 📄 main.py (4.2 KB)
  |   |-- ○ 📄 utils.py (1.5 KB)
  |-- ○ 📄 README.md (2.1 KB)

  Current: src/main.py | Selected: 1 full | ~4,200 chars

You Control the Context

Many AI coding assistants automatically find what they think is relevant context. This is a black box. When the LLM gives a bad answer, you can't debug it because you don't know what context it was actually given.

kopipasta is the opposite. I built it on the principle of explicit context control. You are in the driver's seat. You decide exactly what files, functions, and snippets go into the prompt.

It's a "smart copy" command for your project, not a magic wand.

How It Works

The workflow is a fast, iterative cycle:

1. Context: Run kopipasta to select files and define your task.
2. Generate: Paste the prompt into your LLM (ChatGPT, Claude, etc.).
3. Patch: Press p in kopipasta and paste the LLM's response to apply changes locally.
4. Iterate: Review with git diff, then repeat for the next step.

Session Management

kopipasta introduces a Quad-Memory Architecture to prevent context drift in long tasks.

🧠 Sessions (AI_SESSION.md)

Instead of re-explaining the task in every prompt, use a Session to maintain state.

1. Start (n): Creates AI_SESSION.md (a scratchpad) and snapshots your git commit.
2. Work: The session file is pinned to your prompt. The LLM updates it (via patches) to track progress.
3. Update (u): Generates a handover prompt to compress the session state for the next LLM window.
4. Finish (f): Harvests architectural learnings into AI_CONTEXT.md (Constitution), deletes the session file, and offers to squash your work into a clean commit.

📜 Context (AI_CONTEXT.md)

The "Laws of Physics" for your project. This file is always pinned to the prompt if it exists. Use it for:

• Tech stack constraints (e.g., "Use Poetry", "No React Class Components").
• Architecture decisions.
• Common pitfalls to avoid.

👤 Profile (~/.config/kopipasta/ai_profile.md)

Your global preferences (e.g., "I use VS Code", "I prefer TypeScript"). Injected into every prompt automatically.

Use Cases

• Targeted Refactoring: Select just the module you are cleaning up and its immediate dependencies.
• Test Generation: Pipe your implementation file and a similar existing test file to the LLM to generate consistent new tests.
• Docs to Code: Select an API documentation file (or web URL) and your source file to implement a feature against a spec.
• Bug Fixing: Grab the relevant traceback files and the config to diagnose issues without distracting the LLM with the whole repo.

Installation

Installation

# Using uv (recommended)
uv tool install kopipasta

# For development
uv sync

Usage

kopipasta has two main modes: creating prompts and applying patches.

Creating a Prompt

By default kopipasta opens tree selector on the current dir.

You may also use the command line arguments:

kopipasta [options] [files_or_directories_or_urls...]

Arguments:

• [files_or_directories_or_urls...]: One or more paths to files, directories, or web URLs to use as the starting point for your context.

Options:

• -t TASK, --task TASK: Provide the task description directly on the command line, skipping the editor.

Applying Patches

kopipasta automatically injects strict instructions into your prompt, teaching the LLM how to format code for this tool. kopipasta can apply changes suggested by an LLM directly to your codebase, assuming you are in a Git repository.

1. Press p in the file selector.
2. Paste the entire markdown response from your LLM.
3. The tool robustly detects code blocks, handles indentation quirks, and applies changes (full files or diffs).
4. If a patch fails, the tool provides diagnostic feedback telling you exactly why (e.g., missing headers).
5. Always review changes with git diff before committing.

Example of supported LLM output formats:

# FILE: src/utils.py
def new_feature():
    print("kopipasta handles full file creation")

# FILE: src/main.py
@@ -10,2 +10,3 @@
 def main():
-    pass
+    new_feature()

Key Features

• Total Context Control: Interactively select files, directories, or snippets. You see everything that goes into the prompt.
• Smart Dependency Analysis: Press d on a Python or TypeScript/JavaScript file, and kopipasta will scan imports to find and add related local files to your context automatically.
• Robust Code Patcher: Applies LLM suggestions directly. Handles indentation, various comment styles (#, //, <!--), and multiple files per block.
• Built-in Search: Press g to grep for text patterns inside directories to find relevant files.
• Transparent & Explicit: No hidden RAG. You know exactly what's in the prompt because you built it. This makes debugging LLM failures possible.
• Web-Aware: Pulls in content directly from URLs—perfect for API documentation.
• Safety First:
  • Automatically respects your .gitignore rules.
  • Detects if you're about to include secrets from a .env file and asks what to do.
• Context-Aware: Keeps a running total of the prompt size (in characters and estimated tokens) so you don't overload the LLM's context window.
• Developer-Friendly:
  • Provides a rich, interactive prompt for writing task descriptions in terminal.
  • Copies the final prompt directly to your clipboard.
  • Provides syntax highlighting during chunk selection.

Interactive Controls

Key	Action
Space	Toggle file/directory selection
s	Toggle Snippet Mode (include only the first 50 lines)
d	Analyze Dependencies (find and add imported files)
g	Grep (search text in directory)
a	Add all files in directory
p	Apply Patch (paste LLM response)
r	Reuse selection from previous run
n	Start Session (Init AI_SESSION.md)
u	Update Session (Handover/Checkpoint)
f	Finish Task (Harvest & Squash)
Enter	Expand/Collapse directory
q	Quit and finalize selection