Trunkate AI CLI

The trunkate CLI is the easiest way to integrate prompt optimization into CI/CD pipelines, shell scripts, or Makefile routines. It accepts direct flags and correctly handles STDOUT / STDERR routing to avoid breaking data pipes.

Installation

You can install the CLI globally via NPM:

npm install -g @trunkate-ai/cli

Authentication

Set your API token in your shell profile (~/.zshrc or ~/.bashrc), or pass it via the -k flag.

export TRUNKATE_API_KEY="tk_live_..."

Usage

# 1. Provide your long prompt as text
# 2. Add flags for task, budget, and target model
trunkate-ai \
  --text "Your very long context..." \
  --task "Summarize this document" \
  --budget 500 \
  --model "gpt-4o"

Global Flags

| Flag | Shorthand | Description | | :--- | :--- | :--- | | --text | -t | Required. The actual long prompt to optimize. | | --task | | Recommended context instructions. | | --budget | -b | Maximum target token budget. | | --model | -m | Target model architecture. | | --api-key | -k | Override environment key. | | --url | -u | Override target endpoint for testing. |