██          ██     
                     ██          ██     
  ▒█████░  ██░████   ██  ▓██▒  ███████  
 ████████  ███████▓  ██ ▓██▒   ███████  
 ██▒  ░▒█  ███  ▒██  ██▒██▒      ██     
 █████▓░   ██    ██  ████▓       ██     
 ░██████▒  ██    ██  █████       ██     
    ░▒▓██  ██    ██  ██░███      ██     
 █▒░  ▒██  ██    ██  ██  ██▒     ██░    
 ████████  ██    ██  ██  ▒██     █████  
 ░▓████▓   ██    ██  ██   ███    ░████  

              v.01 manual


snkt is a static site generator focused on simplicity and efficiency.

snkt does a few things, but strives to do them coherently.

snkt generates my personal web site of ~2000 articles in under a second. It should be fast enough to completely regenerate even very large sites in near real-time if needed.


Every 5-10 years I throw out the software for my site and rewrite it.

This time it's in Go. Maybe you'll find it useful. It's 10x faster than the old version in Python.


It powers trenchant.org but is under active development and pieces may change. See TODO for future / in progress work.


The only dependency for building is go.

Install Go for your platform.

Download and build snkt with something like

$ go get adammathes.com/snkt

This will download dependencies, build snkt and place it in your $GOPATH/bin (by default, ~/go/bin/).

snkt is a self-contained binary, you can move it anywhere.

Quick Start

Creating a Site

Use the "-init" option to create the skeleton for a new site -

$ snkt -i myblog

This will create:

  • txt directory for posts
  • html directory for HTML output
  • tmpl directory for templates
    • base HTML structure wrapper
    • archive lists all posts
    • post single post page
    • home home page with recent posts
    • rss RSS 2.0
  • config.yml configuration file

First Post

A one line plaint text file is a valid post.

user@host:~/myblog$ echo "hello world" >> txt/hi

Build the site

$ snkt -b

Output should now be in the html directory and look like

  • html
    • hi/index.html hello world post
    • index.html
    • archive.html
    • rss.xml

Viewing the Results

snkt includes a simple web server to view the results with

$ snkt -p

Visiting http://localhost:8000 in a web browser should now show the site and the first post.

You can now copy this HTML anywhere and you're set.

Command Line Options

Usage of snkt:
  -b, --build
    	generates site from input files and templates
  -c, --config configuration
    	configuration file (default "config.yml")
  -h, --help
    	print usage information
  -i, --init directory
    	initialize new site at directory
  -s, --serve
    	serve site via integrated HTTP server
  -v, --verbose
    	log operations during build to STDOUT
  -w, --watch
    	watch configured input text dir, rebuild on changes


$ snkt -c site.yaml -b
$ snkt --config=myconfig.yml -v -w


Per site configuration is via a YAML file.

For most purposes, it should just be a listing of attribute : value

Configuration options --

name value default
input_dir absolute path of directory for text input files
output_dir absolute path of directory for html output files
tmpl_dir absolute path of directory for template files
site_title string for the site's title
site_url absolute URL for the site
filters list of search/replace regex's to run on posts
permalink_fmt format string for permalinks /%F/
post_file_fmt format string for post filenames /%F/index.html
show_future include posts with dates in the future false
preview_server host:port to spawn the preview server localhost:8000
preview_dir root directory of preview server output_dir


Post inputs are stored as plain text files. (I have only tested UTF-8 and ASCII.)

Posts have an optional metadata preamble, and a markdown formatted body. The preamble is just a series of name value pairings separated by a colon (:) character.

Minimal complete and valid post --

this is a totally valid post

Post with a preamble --

title: also a valid post
date: 2017-02-08
valid: totes

This post will have an explicitly set title (ooh! fancy!) 
instead of inferred from the filename. 

It will also have an explicitly set date instead of inferring 
it from the file creation/modification time.

`totes` will be stored in the post's `meta` map under `valid.` 
You don't have to worry about that right now. Honest.


Templates use the standard library Go text/template.

Entities in the templates --


	Title string
	URL   string
	Posts array of posts

See site/site.go for more.


	// Metadata
	Meta       map[string]string
	SourceFile string
	Title      string 
	Permalink  string
	Time       time.Time
	Year       int
	Month      time.Month
	Day        int
	InFuture   bool

	// Content text -- raw, unprocessed, unfiltered markdown
	Text string

	// Content text -- processed into HTML via markdown and other filters
	Content string

	// Content with sources and references resolved to absolute URLs
	AbsoluteContent string

	// Post following chronologically (later)
	Next *Post
	// Post preceding chronologically (earlier)
	Prev *Post

	// Precomputed dates as strings
	Date    string
	RssDate string


Displays recent posts and rendered to index.html in the output_dir.

  • {{.Site}} Site
  • {{.Posts}} Posts all posts on site in reverse chronological order


Each individual post uses this template

  • {{.Site}} Site
  • {{.Post}} Post the individual post


Lists all posts, showing only titles and links. Rendered to archive.html

  • {{.Site}} Site
  • {{.Posts}} Posts all posts, reverse chronological order


Displays recent posts as RSS 2.0 XML. Rendered to rss.xml

  • {{.Site}} Site
  • {{.Posts}} Posts all posts, reverse chronological order

Advanced Configuration Options

Permalink and filename formatter

Permalinks (URLs for individual posts) can be customized.

String Value Example
%Y Year 2017
%M Month 04
%D Day 14
%F Filename foo
%T Title bar

Filename is a cleaned version of the post's original filename with the extension removed. Filenames and titles will be "cleaned" of characters unsuitable for links, with whitespace replaced by -.


Arbitrary regular expressions can be executed on each post to create domain-specific and site-specific modifications.

Here are the real world examples of regular expressions that filter each post on my personal site -

  - s: <photo id="(.+)">
    r: <div class="photo"><img src="/img/$1" /></div>
  - s: <segue />
    r: <p class="segue">&middot; &middot; &middot;</p>
  - s: <youtube id="(.+)">
    r: <p class="video"><a href="https://www.youtube.com/watch?v=$1"><img src="/img/$1.jpg" /></a></p>
  - s: "amazon:(.+)"
    r: "http://www.amazon.com/exec/obidos/ASIN/$1/decommodify-20/"

Work in Progress Features

These features are working but less documented and potentially still in progress and subject to change.

Paged Archives

If a template named paged is present then paged archives (15 posts per page) are created at output_dir/page/%d.html

Template variables are the same as the archive template, but with .NextPage and .PrevPage as integers of the next and previous page.

See archive/paged.go for details.


There is preliminary support for tag style metadata per post.

Add a "tags" field to your post preamble. Tags should be comma separated.

    tags: TagOne, tag two, a third tag, fourth

Tags will be normalized to lowercase, with spaces replaced with underscores. So the above would have tagged a post with --

tagone tag_two a_third_tag fourth

Tags are accessible in each post via the Tags field.

To create pages by tag, create a template named tags.

This creates a file at OUTPUT_DIR/tag/tag_name/index.html for each tag.

It will have access to the same variables as an archive template with the additional .Tag for the tag name.

Binary Files as Posts

Preliminary support to treat binary files as standalone posts.

Drop image files with "jpg" or other image extensions into the "txt" dir.

  • post's ContentType will be set to "image"
  • text fields will be empty strings
  • metadata will be populated as it can via exif (maybe)

Video and audio files have preliminary support too -- see post/post.go

Example configurations, sites, themes

not done

Rebuild and deployment recipes

also not done


  • sample sites/templates
  • proper man pages for docs


Pull requests and issues are welcomed at https://github.com/adammathes/snkt