LLM Wiki CLI

The LLM Wiki CLI is the primary companion tool for authoring, validating, and querying this semantic knowledge vault. It parses markdown files, converts YAML/JSON frontmatter into RDF graphs, resolves blank nodes, and performs deductive reasoning under OWL-RL semantics.

Wiki workflows and authoring guide

This section covers common editing, auditing, and publishing procedures for contributing to the semantic knowledge vault.

Create a new page

Create a new markdown file in the wiki/ directory with standard frontmatter. Use lowercase kebab-case for the filename (e.g., my-new-article.md).

Fill in the semantic metadata

Always specify what fields are required for your document type. For example, if your page describes a person:

id: wiki:alice-smith
type: schema:Person
name: Alice Smith
givenName: Alice
familyName: Smith
context: Software Engineer working on the LLM Wiki CLI
status: permanent
dateCreated: 2026-05-08

Run the validation suite

Before committing any files, run wiki check to verify that all filenames are valid, all internal WikiLinks resolve correctly, and all SHACL constraints are satisfied.

wiki check -v

Render dynamic tables

If your page contains inline SPARQL query comments (<!-- sparql:start -->), run wiki render to update the markdown tables automatically:

wiki render -v

Wiki schema and active types

Active shapes are loaded from the input_dirs and executed against all matching pages during wiki check.

Current pages in the vault by type

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX schema: <https://schema.org/>

SELECT ?document ?type WHERE {
  ?document rdf:type ?type .
  FILTER(STRSTARTS(STR(?document), "wiki:"))
  FILTER(?type != schema:WebPage)
}
ORDER BY ?type

Person shape

• givenName: Required string (exactly 1)
• familyName: Required string (exactly 1)
• context: Required string (one-liner describing relationship)
• status: Required choice of either permanent or one-off
• dateCreated: Required date in YYYY-MM-DD format

Design philosophies

The architecture of the LLM Wiki CLI is governed by core Unix system design philosophies to ensure platform portability, scriptability, and robust performance in CI/CD pipelines.

Silence is golden

By default, action-oriented subcommands (like render and check --fix) exit silently with code 0 upon success. This adheres strictly to the Unix philosophy, preventing unnecessary output log clutter and ensuring that commands can be seamlessly integrated into automated scripting. Users can opt-in to interactive summaries and details at any time by appending the -v or --verbose flag.

Pipes and filters over custom hardware drivers

The CLI avoids platform lock-in and core scope creep by delegating hardware-level operations (like printing or typesetting) entirely to userland. Instead of maintaining complex margin-wrapping, font-measurement, or network spooling libraries, the CLI outputs clean, standard formats (markdown, json, csv, table) directly to standard output (stdout). This allows users to compose powerful piped chains using standard Unix utilities:

# Query names, format as a table, and print using system spooler
wiki query "SELECT ?name WHERE { ?s schema:name ?name }" | pr -h "Wiki Names" | lp

Flat and streamlined command surface

The CLI features a flat, intuitive set of top-level commands with zero nested command groups, making the tool extremely easy to learn and script.

Command reference

wiki check

Unifies all vault health inspections under a single entry point, executing both strict semantic validations and softer stylistic audits.

# Run audits and output details/warnings verbosely
wiki check -v

# Run in strict mode (warnings are promoted to errors and fail CI)
wiki check --strict

# Automatically normalize and standardize frontmatter casing inline
wiki check --fix

wiki query

Executes raw SPARQL SELECT or CONSTRUCT queries against your vault's unified graph.

wiki query "SELECT ?name WHERE { ?s schema:name ?name }"

wiki render

Finds any dynamic SPARQL query comments (<!-- sparql:start -->) inside your markdown files and updates their tables inline.

# Render all tables silently (silence is golden)
wiki render

# Render with detailed verbose output
wiki render -v

wiki export

Compiles all document frontmatter into a canonical JSON-LD array, acting as a clean extension point for bulk semantic integrations.

wiki export -o exported-wiki.json