Blog contents

2026MAY

Podlite 2.0 — eight new blocks, and a CSV that finally renders itself I shipped Podlite 2.0 this week. The release is mostly about what a single block can carry, and the one I already reach for daily is =data-table: I point a block at a CSV or TSV file and it renders as a table — instead of laying the table out by hand and re-spacing every column each time a value changes. The other seven additions follow the same idea. =set pre-configures attributes for the next block, so a long caption or a shared setting lives in one place. G<> and :masked keep content hidden until you choose to show it — useful for spoilers, exercise answers, or notes meant for one audience. =boundary is a typed section divider a tool can read, not just a line on the page. And =table now warns about malformed rows instead of quietly corrupting the output — a small thing that has bitten me before. Underneath the feature list there is the thread I keep pulling on: a document should mean the same thing whether a person or an agent reads it. Each block in 2.0 is a contract — explicit type, named attributes — so the parser and the pod6.in editor can rely on it, and so can whatever agent opens the file next. That is the part I care about most. Two things parse differently now: a few legacy string-attribute formats were removed, and =include is classified as a directive. Both are parser-side — a well-formed v1.0 document renders unchanged, so there is nothing to migrate in your own text. Full write-up · Spec · Release notes · Try it live

26th

MARCH

podlite-markup skill — structured output for AI agents The podlite-markup skill is now on the skills.sh marketplace. One command installs Podlite support into Claude Code, GitHub Copilot, Cursor, Codex CLI, and 35+ other agents: npx skills add podlite/podlite-skills The problem it solves: AI agents write well, but their output doesn't hold structure between sessions. Markdown frontmatter is a convention — parsers may or may not read it, and agents that haven't been told what to look for usually don't. Podlite uses explicit typed blocks with named attributes. When an agent writes =begin pod :pubdate('2026-03-12') :type('page'), that metadata is a structured field, not prose — any tool can read it without guessing. I've been using this in my own workflow for specs, decision logs, and blog posts. What changed: a second agent can open a file written by a first agent and immediately understand what it is, what state it's in, and what to do next. No re-explaining context. The file carries the instructions. Podlite natively supports Markdown, so existing files don't need migration. Full write-up · Source

11th

2025NOVEMBER

Podlite Desktop Editor 0.6.1 is out Just released a new version of Podlite Desktop Editor - a free markup editor that works with Podlite files. What's new in 0.6.1Rebuilt the editor from scratch with updated libraries. Here's what changed:Syntax highlighting that actually works wellInline formatting: B<bold>, I<italic>, C<code>, L<links>, O<strikethrough>Click links with Cmd/Ctrl to open themLink to other files with file: schema - they open in new windowsText search (finally!)Fixed a bunch of window resize bugsmacOS Tahoe supportYou can toggle between half-screen and full-screen preview with keyboard shortcuts.DownloadAvailable free on all platforms:Github releases: https://github.com/podlite/podlite-desktop/releasesMicrosoft Store, Snap Store, Mac App StoreLearn more about PodliteCheck out the Podlite Quick Tour: https://podlite.org/quick-tourTry the web version first if you want: https://pod6.inFull details: https://podlite.org/2025/11/1/1/podlite-desktop-editor-0-6-1-releasedSource code: https://github.com/podlite/podlite-desktop

1st