Logseq-Blogger

Use Logseq as your blog editor. Write posts in your graph and mark them with type:: blog property. Logseq-Blogger discovers them and when you set the property status:: online invokes syndicator.

Context

                logseq-blog                         syndicate
[Logseq Graph]───────○ )───────[Logseq-Blogger]───────( ○───────[Syndicator]

Logseq-Blogger uses the logseq-blog interface to retrieve blog posts from a Logseq Graph and forwards these blog posts to Syndicator using the syndicate interface.

• The logseq-blog interface is specified below.
• The syndicator interface is specified in syndicator.

Commands

uv run logseq-blogger syndicate                 # full syndication of all ::online blog posts that have no ::syndicated-at
uv run logseq-blogger list                      # list slug of all blog posts with there ::syndicated-at status
uv run logseq-blogger syndicate --post <slug>   # full syndication of the blog post with the given slug
uv run logseq-blogger redeploy --post <slug>    # regenerate the static hugo page for the blog post with the given slug
uv run logseq-blogger version                   # print version

logseq-blog interface

The logseq-blog interface is the contract between a Logseq Graph and Logseq-Blogger. A blog post is a Logseq markdown fragment marked with type:: blog.

Discovery

Under the Logseq graph root (logseq_dir in config.local.yaml):

Path	Role
journals/	Journal files (nested-list format)
pages/	Page files (top-level metadata format)

Every *.md file in those directories is scanned. A post is found wherever a property block contains the substring type:: blog.

Selection and gates

Rule	Behaviour
type:: blog	Required discovery marker
status:: online	Only online posts are considered for hand-off
syndicated-at:: present	Skipped by syndicate (already handed off)

Properties

Logseq properties use key:: value syntax.

Property	Required	Notes
type	yes	Must be blog
status	yes (for publish)	Only online is selected
date	yes	YYYY-MM-DD; used in the slug
title	yes	Used in the slug (spaces → _)
author	yes	Passed through to syndicate meta.author
header	yes for hand-off	Markdown image, e.g. . File must exist on disk
summary	no	Explicit summary; otherwise first body text block
language	no	Word form: german, english, spanish, french, italian → de/en/es/fr/it (default de)
position	no	Free text or coordinates; passed through to syndicate meta.position
syndicated-at	written by tool	ISO-8601 timestamp set after successful syndicate

Example 1 — Nested list (journals)

Metadata lives in the first list item under a parent (commonly [[Blog]]). Sibling list items are the body.

- [[Blog]]
	-
	  type:: blog
	  status:: online
	  language:: german
	  date:: 2026-05-19
	  position:: 38.83388, 20.69757
	  title:: Charly Superstar
	  author:: Benno
	  header:: ![charly-superstar.jpg](../assets/charly-superstar_1779210265357_0.jpg)
	  syndicated-at:: 2026-08-03T11:27:19+02:00
	- Wie die meisten Hündeler glauben auch wir: unser Hund ist der schönste und liebste der Welt.
	- ![charly-drinkt.mp4](../assets/charly-drinkt_1785678123966_0.mp4)
	- Charly scheint aber wirklich eine besondere Wirkung auf die Menschen zu haben.
	- {{video https://youtube.com/shorts/_yNn9MUXBaw}}

• Properties are indented under the type:: blog bullet
• Each subsequent list item becomes a content block
• Multiple blog posts may appear in one journal file
• Non-blog journal content is ignored

Example 2 — Top-level metadata (pages)

Properties at the top of the file; body as list items below.

type:: blog
status:: online
language:: german
date:: 2026-05-19
position:: 38.83388, 20.69757
title:: Charly Superstar
author:: Benno
header:: ![charly-superstar.jpg](../assets/charly-superstar_1779210265357_0.jpg)
syndicated-at:: 2026-08-03T11:27:19+02:00
- Wie die meisten Hündeler glauben auch wir: unser Hund ist der schönste und liebste der Welt.
- ![charly-drinkt.mp4](../assets/charly-drinkt_1785678123966_0.mp4)
- Charly scheint aber wirklich eine besondere Wirkung auf die Menschen zu haben.
- {{video https://youtube.com/shorts/_yNn9MUXBaw}}

• One blog post per page file

Content blocks

Body list items are classified as:

Kind	How it is recognised
title	Line starts with # … ######
text	Ordinary prose (nested bullets become * … lines)
media	First line is  — image or local video by extension
youtube	First line is {{video https://…}} with a YouTube URL

Setup

git clone git@github.com:bbaumgartner/logseq-blogger.git
cd logseq-blogger
curl -LsSf https://astral.sh/uv/install.sh | sh   # if uv is missing
uv sync
cp config.local.yaml.example config.local.yaml       # paths, SFTP host, webhook URLs

Test

uv sync --group dev
uv run pytest

Architecture

This is a cli built with python. Simplicity and Maintainability are most important NFRs. It should just wire together various programs that do the complex things and hence, act as an orchestrator. The most complex thing this project does is parse Logseq Markdown files. That's at the very limit of acceptable complexity in this project. It was considered to pull that out into a dedicated project, however, with LLM support this complexity should be manageable, if not we can still pull it out or use a library. The runtime environment for this is Linux, Mac support is nice to have, Windows is out-of-scope.

Software Design

classDiagram
    direction TB

    class cli {
        entry point
        parse commands, dispatch
    }
    class config {
        load YAML → Config
    }
    class extract {
        scan journals/pages → BlogPost
    }
    class marker {
        read/write syndicated-at::
    }
    class trigger {
        stage, upload, webhooks, mark done
    }
    class sftp {
        resumable SFTP upload
    }
    class webhook {
        HTTP POST /publish, /reel
    }

    cli --> config : load_config
    cli --> trigger : syndicate / redeploy / list
    trigger --> config
    trigger --> extract : scan_blog_posts
    trigger --> marker : is_syndicated / set_syndicated_at
    trigger --> sftp : sftp_session / upload
    trigger --> webhook : post_webhook
    marker --> extract : BlogPost
    sftp --> config : host, key

Loading

Module	Role
cli	Entry point: parse commands, load config, dispatch
config	YAML → Config dataclasses
extract	Scan journals/pages → BlogPost domain model
marker	Read/write syndicated-at:: on source files
trigger	Stage, upload, fire webhooks, mark done
sftp	Resumable SFTP upload to syndicator staging
webhook	HTTP POST to n8n /publish and /reel