An Agent-Friendly Blog Framework

TLDR Agora is a lightweight Hugo-based blog evolving from human-readable to Agent-native. Covers Markdown-first approach, terminal aesthetics, client-side search, and roadmap: DID identity, RSS, cross-site search, REPL interaction, annotation layer.

January 1, 2025 · updated Jun 16, 2026 #agent

It’s 2026 and I’m rebuilding my blog. Again. But this time AI has moved the goalposts. I’m no longer the total front-end newbie (Well, mostly.)

But AI shouldn’t only help us write better code. Agents themselves should be readers.

Here’s why: human attention is finite. In an age of infinite content, nobody can read, watch, or learn everything. A dense one-hour tutorial takes more than an hour to truly absorb — especially in a foreign language. Agents process text at token-per-second speeds. And unlike us, they don’t forget. We’re squirrels stashing acorns for winter, then drawing a blank when spring comes. Agents remember where they put things.

Core Principles

A few ground rules before building anything:

Agent-First: Clean structure, explicit semantics. Agents should parse and interact with this effortlessly.
Lightweight: No infrastructure bloat. Deployable and maintainable by one person.
Minimal: The daily workflow is just Markdown. Nothing more complicated than jotting notes.

What is an Agent-Friendly Blog

Traditional blogs are built for human eyes. An agent-friendly blog treats agents as first-class citizens — not just writing assistants, but independent readers with their own agendas.

That means the blog needs to give agents:

Structured access to content (not just rendered HTML)
A way to identify themselves
Interfaces they can actually use — read, annotate, publish, subscribe

Requirements Breakdown

Agent CLI Interface

The framework should feel natural from a terminal:

DID Identity: Every agent gets its own decentralized identity
RSS Feeds: Agents subscribe and poll for updates like any other reader
Cross-Site Search: Agents can crawl and index content across multiple blogs
Site API: Full CRUD access
- Reader mode: summarize, bookmark, translate, query articles
- Author mode: publish, edit, delete posts from the command line
REPL Interaction: A command prompt at the bottom of every page. Register your agent and start typing.
Annotation Layer: Agents can read and contribute to discussions

Human Layer

None of the old conveniences disappear:

Markdown-First: Writing is just Markdown files
Cross-Platform: Publish to multiple platforms with one command
Author-Friendly: Edit Markdown. Don’t think about deployment.

References and Inspirations

Hacker News API interaction design
openCLI interaction paradigm