Get free audit

Startup growth agent

Blog posting

Finds the day’s most relevant AI news and publishes an SEO/AEO-friendly blog post.

Running since Jun 20, 2026·claw-world.app·Risk: Low
Have us run itSee all agents

What this solves

A blog that only gets a post when someone remembers to write one goes stale fast, and stale blogs are a worse signal to both readers and search/AI crawlers than no blog at all.

Who it’s for: Teams who want a steadily updated blog without assigning someone to write on a schedule.

How we use it: This agent has published our own daily post since June 2026. It is also the one place our own usage changed the design: search data showed the posts were read as general AI news rather than content about AI visibility, so the topic-picking step now explicitly favors stories about AI search and answer engines over bigger but less relevant news.

How it works

What happens on every run.

  1. Pulls the day’s top AI news from a news API, with a fallback source if the primary one is down or empty.
  2. Ranks the candidates and picks the one closest to what the blog is actually about, rather than simply the highest-scoring story.
  3. Checks recent posts so the same story is not covered twice in a short window.
  4. Writes the post with the required frontmatter (title, slug, date, excerpt, and the rest the site’s blog list reads) and a body in the site’s established voice.
  5. Commits and pushes the new post, which deploys the site.

Skills and scripts

What it’s built from.

A Claude Code workflow plus a news-fetching helper. Not packaged for public download yet — the setup guide below describes what it does; the managed tier runs it for you today.

Setup guide

Running it yourself.

Runtime

  • An agent runner that can follow a written workflow, fetch URLs, and write files — we use Claude Code.
  • Git, and push access to the repository the site builds from.

Browser setup

  • None needed — this agent only fetches data and writes files, no browser automation.

Account setup

  • A news source to pull from (an API or a page you can fetch); a fallback source is worth having for the days the primary one is unavailable.
  • Push access to whatever repository your site deploys from.

Configuration

  • The exact frontmatter fields your blog list expects, so new posts render correctly without a manual fix.
  • A short editorial brief: what the blog is about, what to prefer when several stories qualify, what voice to write in.
  • A duplicate-coverage window (we use five days) so the same story does not run twice.

First run

  • Run it once and read the draft post before it is pushed.
  • Confirm the live site built the new post with the frontmatter intact and nothing missing from the blog list.

Common questions

The news source returned nothing for today.

Widen the time window once, then fall back to the backup source. If both come up empty, skip the day rather than publish something thin.

It picked a story that is not really relevant to us.

Tighten the ranking brief with explicit examples of what counts as relevant, and fall back to the next-highest story when nothing in the top results qualifies.

Risk notice

What to watch for.

  • Lowest-risk of the three: no automation against a third-party platform, so there is nothing to get an account flagged for.
  • The main failure mode is editorial drift — a feed can start optimizing for whatever is easiest to source rather than what is actually relevant, which is why the ranking step exists and gets revisited.

Run it

Do it yourself, or have us run it.

Have us run itQuestions about setting it up yourself

Run log for this agent: coming soon