KnowledgeAsCode.Tool 0.7.0

kac: knowledge as code

kac reads a folder of Markdown documents kept in git and holds each one to a schema. The folder a document sits in names its type. The corpus's own .schema/ holds one file per type, saying what fields and rules that type carries. A repository of those documents, with the schema it runs, is a corpus.

What the structure buys you: indexes built from the documents themselves, and a broken cross-reference that fails CI rather than rotting quietly. An agent contributing to the corpus can be told where a document goes.

kac carries no rules about any particular type. It reads what to enforce from the schema the corpus holds, so you add a knowledge type by writing a YAML file.

Install

dotnet tool install --global KnowledgeAsCode.Tool

Start a corpus

kac new turns the folder you are standing in into one. It takes the framework from the repository below, writes what a corpus is made of, and stops short of committing so you can read what it staged.

mkdir my-corpus && cd my-corpus
kac new          # asks four things, and has a default for each

Use

kac finds a corpus by walking up from the working directory for a .corpus.yaml, the descriptor naming the corpus, so run it from anywhere inside one.

kac validate     # frontmatter, links, structure, clauses and the graph
kac generate     # regenerate the indexes and the generated blocks in each type page
kac update       # take a newer framework from the template this corpus was made from
kac export       # write the corpus to .dist/export/ as data a consumer reads instead of cloning
kac bundle       # assemble that export and .plugin/ into an installable plugin
kac checks       # list every check the validator implements

kac update writes into the working tree and commits nothing, so git diff is where you read what arrived. kac update --check reports the same and writes nothing, which is the form a pipeline runs.

kac --help lists them, and every command carries its own --help.

Exit codes

Where to go next

The repository this tool is built from carries the framework's documentation, a page for each record type, and a reference for every command:

https://github.com/paul80nd/knowledge-as-code

It also holds a worked corpus you can read before you write your own.

Released under the MIT licence.