Building local ZeroClaw wiki

build-local-zeroclaw-wiki.md

Below is an example how to build local ZeroClaw wiki (as it is on ZeroClaw official wiki):

➜  git clone https://github.com/zeroclaw-labs/zeroclaw.wiki.git && cd zeroclaw.wiki
➜  zeroclaw.wiki git:(master) ✗ cd src && ln -s Home.md SUMMARY.md && cd -
➜  zeroclaw.wiki git:(master) ✗ ./transform_links.sh -d src
Transformed: src/09.1-Memory-Backends.md
Transformed: src/09.2-Hybrid-Search.md
~...~
Transformed: src/07.1-Agent-Turn-Cycle.md
Transformed: src/12.3-Testing.md
Transformed: src/05.2-Custom-Providers.md
➜  zeroclaw.wiki git:(master) ✗ cat book.toml
[preprocessor.mermaid]
command = "mdbook-mermaid"
[output.html]
additional-js = ["mermaid.min.js", "mermaid-init.js"]
➜  zeroclaw.wiki git:(master) ✗ cargo install mdbook-mermaid                                                                                                                           
~...~
➜  zeroclaw.wiki git:(master) ✗ mdbook-mermaid install
[2026-04-07T08:18:27Z INFO  mdbook_mermaid] Reading configuration file ./book.toml
[2026-04-07T08:18:27Z INFO  mdbook_mermaid] Writing additional files to project directory at .
[2026-04-07T08:18:27Z INFO  mdbook_mermaid] Files & configuration for mdbook-mermaid are installed. You can start using it in your book.
[2026-04-07T08:18:27Z INFO  mdbook_mermaid] Add a code block like:
    ```mermaid
    graph TD;
        A-->B;
        A-->C;
        B-->D;
        C-->D;
    ```
➜  zeroclaw.wiki git:(master) ✗ mdbook build    # or: mdbook serve --open
 INFO Book building has started
 INFO Running the html backend
 INFO HTML book written to `/home/vit/prjs/my_ai/zeroclaw/zeroclaw.wiki/book`
➜  zeroclaw.wiki git:(master) ✗

Now open it by cd book && open index.html in browser.

#!/usr/bin/env bash

#
# transform_links.sh
# Transforms markdown links with GitHub wiki URLs to local .md file references
#
# Usage:
#   ./transform_links.sh <file.md> [file2.md ...]   # Transform specific files
#   ./transform_links.sh -d <directory>              # Transform all .md files in a directory
#   ./transform_links.sh -i <file.md> [...]          # Edit files in-place (overwrites originals)
#
# Examples:
#   ./transform_links.sh README.md
#   ./transform_links.sh -d ./docs
#   ./transform_links.sh -i README.md CHANGELOG.md

set -euo pipefail

INPLACE=false
DIR_MODE=false
TARGET_DIR=""

usage() {
    echo "Usage:"
    echo "  $0 <file.md> [file2.md ...]   Transform specific files (output to stdout)"
    echo "  $0 -d <directory>             Transform all .md files in a directory"
    echo "  $0 -i <file.md> [...]         Edit files in-place (overwrites originals)"
    exit 1
}

# Parse flags
while getopts ":id:" opt; do
    case $opt in
        i) INPLACE=true ;;
        d) DIR_MODE=true; TARGET_DIR="$OPTARG" ;;
        *) usage ;;
    esac
done
shift $((OPTIND - 1))

# The core sed expression:
# Matches: [Any Text](https://anything/wiki/PAGE-SLUG)
# Captures: link text and the last path segment (page slug)
# Replaces: [Any Text](PAGE-SLUG.md)
SED_EXPR='s|\(\[[^]]*\]\)(https\?://[^))]*/wiki/\([^)]*\))|\1(\2.md)|g'

transform_file() {
    local file="$1"

    if [[ ! -f "$file" ]]; then
        echo "Warning: '$file' not found, skipping." >&2
        return
    fi

    if $INPLACE; then
        # BSD (macOS) and GNU sed both support -i with an empty string for in-place
        if sed --version 2>/dev/null | grep -q GNU; then
            sed -i "$SED_EXPR" "$file"
        else
            sed -i '' "$SED_EXPR" "$file"
        fi
        echo "Transformed: $file" >&2
    else
        sed "$SED_EXPR" "$file"
    fi
}

if $DIR_MODE; then
    [[ -d "$TARGET_DIR" ]] || { echo "Error: '$TARGET_DIR' is not a directory."; exit 1; }
    INPLACE=true  # Directory mode always edits in-place
    find "$TARGET_DIR" -name "*.md" | while read -r f; do
        transform_file "$f"
    done
elif [[ $# -gt 0 ]]; then
    for file in "$@"; do
        transform_file "$file"
    done
else
    usage
fi