Token Budget Builder

Assemble, compress, and copy multi-file prompts with a live token budget for any model.

Instead of pasting files into a chat window and guessing whether you fit under the context limit, Token Budget Builder lets you pick the files that matter, compress them locally, and copy a clean formatted prompt to the clipboard in one click. No API keys. Everything runs locally.

How to Use

Open the Context Compressor panel in the Activity Bar.
In Context Files, add files via the toolbar or by right-clicking in the Explorer.
Check or uncheck files to include or exclude them. The token budget updates in real time.
In Build Prompt, click Model or Compression to adjust settings, then click Copy Prompt to assemble and copy.

To count tokens in a file or folder without adding it to context, right-click in the Explorer and choose Count Tokens in Selection.

Compression Modes

Mode	What it removes
None	nothing
Strip Comments	Line and block comments (JS/TS/Python/Go/Rust/C)
Collapse Whitespace	Blank lines and trailing spaces
Signatures Only	Function and class bodies; keeps signatures and docstrings

Supported Models

Model	Encoding	Context window
GPT-4o / GPT-4o mini	o200k_base	128 000
o1 / o3	o200k_base	200 000
GPT-4 / GPT-4 Turbo	cl100k_base	128 000
GPT-3.5 Turbo	cl100k_base	16 385
Claude (approx.)	cl100k_base	200 000

Token Budget Builder

Uchiha Labs

Token Budget Builder

How to Use

Compression Modes

Supported Models