My Digital Garden

Fabric AI Prompter

Fabric AI Prompter

Fabric is a toolset published by Daniel Miessler which:

  • contains some proven ChatGPT prompts for specific tasks (Patterns in the language of this tool)
  • provides a set of shell scripts for running the patterns against source text and returning the ChatGPT response.
  • an optional server component to allow complete independence from Miessler's tooling


The tool appears to be designed to be run on a Mac (examples contain use of pbpaste command) but should be runnable on any Linux (it's shell scripts and Python).

My access to Unix-like behaviour is via Ubuntu running in WSL2 on Windows, and I make use of conda for managing python enviroments and dependencies, so I had to tweak the setup slightly.

Installing on WSL2 with conda

  1. Do steps 1-5 in the Quickstart (i.e. up to and including the installation of Poetry)

  2. Configure Poetry to work with Conda:

    poetry config virtualenvs.path $CONDA_ENV_PATH
    poetry config virtualenvs.create false

    where CONDA_ENV_PATH is the path to the base envs folder (e.g., /home/myuser/miniconda3/envs).

    Do Not make the mistake I did and export CONDA_ENV_PATH in your bash profile...

  3. Create a new conda environment and activate it

    # Update the conda package manager to the latest version in your base environment
    conda update -n base conda
    # If you use Anaconda, use conda to update Anaconda to the latest version in your base environment
    conda update -n base anaconda
    # create a new environment for Fabric
    conda create --name fabric-ai python=3.11
    # activate the new environment
    conda activate fabric-ai

    From this point on, all package management in that environment should be done using Poetry (see [[How to Use Poetry with Conda for Package Management]])

  4. Resume the Quickstart at step 6 (run

  5. To follow most of the examples, install scripts to mimic PBcopy and PBpaste in WSL

Further exploration

Still to do:

  • play with a few of the prompt patterns
  • explore server mode

See also