Skip to content 
  ██████╗ ██████╗ ██╗ ██████╗██╗  ██╗  Prompt Rich Templating CLI Engine
  ██╔══██╗██╔══██╗██║██╔════╝██║  ██║  For LLM Prompts and Shell Commands
  ██████╔╝██████╔╝██║██║     ███████║  with Multi-Step Processing Pipelines
  ██╔═══╝ ██╔══██╗██║██║     ██╔══██║
  ██║     ██║  ██║██║╚██████╗██║  ██║
  ╚═╝     ╚═╝  ╚═╝╚═╝ ╚═════╝╚═╝  ╚═╝

prich

CLI tool for prompt-rich, shareable LLM templates with pre/post-processing and multi-step pipeline workflows — use it for your code review, git-diff summaries, CSV insights, logs and results analysis, and more!

  • Why prich? Reusable Jinja2 templates, flexible steps (shell/Python/LLM), and isolated per-template venvs for dependable runs. Teams can share a .prich/ folder or zip packages and install with one command.
  • Supported providers: Ollama API, OpenAI-compatible HTTP, MLX (local, mac), and a safe echo/STDIN mode for dry-runs or custom bridges.

Status: early, under active development.

What you can do

  • Chain steps: run git diff, parse files, then ask the LLM to summarize or review.
  • Share and install templates from a remote repo or local zips/folders.

90-second Quickstart

# Install (recommended)
pipx install git+https://github.com/oleks-dev/prich

# Initialize config (global)
prich init --global

# Create & run your first template (this will create a very simple one step template)
prich create my-template --global
prich run my-template

By default, prich uses the echo provider (renders your prompt only). Switch providers by editing ~/.prich/config.yaml (see Reference → Config → Providers).

Learn more