Manual¶
So - how do we actually use this thing? That’s a very good question - I’m glad you asked! Outlined here is how to get started, put a complete project together, and some advanced concepts.
Sphinx includes a program-output directive; but for a myriad of reasons, this
can prove ineffective; perhaps the program cannot be run on the host that generates
your Sphinx documents, or credentials are required, or the colorisation of the output
has been removed. Our aidoc image turns console output into images - which can then
be placed into your .rst using a standard ..image:: directive..
$ aidoc image command -o ._static/images/ls.png 'ls -l'
We can also us AI to generate images. Check out our Prompts <target-prompt> documentation for creating and managing the prompts.
To create an image of the output of a prompt run, we can do something like this:
$ aidoc image prompt -o soap.png --chat shaving-soap soap-recipe
If you were writing a cook book, say, you could manage all of your recipes as prompts, then simply crafting prompts, and regnerating the images, you’d have a new updated revison of your book.
You can also write a prompt that will create an image. In this case, we’re generating a SVG, and we’re injecting a stylesheet into the prompt so that, say, if the image was for a website, it would have colorways that are aligned with the underlying desigh.
---
prompt:
- name: logo
system_message: |
You are an expert graphic designer and SVG rendering engine. Your sole task is to generate a professional, modern corporate logo.
CRITICAL CONSTRAINTS:
1. DO NOT simply output the name as plain text on a solid background.
2. You MUST design a distinct, abstract visual icon or geometric symbol that represents the core concept (e.g., automated documentation, source trees, or AI connectivity).
3. Combine the icon and the brand name cleanly within the viewBox. Scale them harmoniously so it looks like a production-ready software logo.
4. Use professional vector paths (<path>, <circle>, <g>) to construct the icon layout. Do not rely solely on basic shapes like primitive rectangles.
5. Ensure the visual aesthetic is minimalist, modern, and high-impact.
weight: HEAVYWEIGHT
# Standardised Input Schema Hardware
input_schema:
type: object
properties:
name:
type: string
description: "The name of the software suite to brand (e.g., 'aidoc')."
description:
type: string
description: "A description of what the tool does (e.g., 'A CLI utility built on Click/Rich to automate documentation, extract schemas, and manage system assets')."
stylesheet:
type: string
format: uri
description: "The structural file:/// URI locating your corporate branding CSV/CSS stylesheet rules."
required:
- name
- description
# Standardised Output Schema Hardware
output_schema:
type: object
properties:
svg_source:
type: string
description: "The complete, raw XML vector block code starting with '<svg>' and terminating with '</svg>'. Must include a valid viewBox configuration."
primary_color_used:
type: string
description: "The dominant hex color code pulled directly from the injected stylesheet configuration file."
secondary_color_used:
type: string
description: "The accent hex color code pulled directly from the injected stylesheet configuration file."
aspect_ratio:
type: string
description: "The generated geometric bounding box ratio expressed as a string segment (e.g., '1:1', '4:3')."
required:
- svg_source
- primary_color_used
- secondary_color_used
- aspect_ratio
additionalProperties: false
The prompt can then be run as follows:
$ aidoc image prompt -o logo.svg --chat my-website logo