A prompt types itself and branches out into three agents, each with a role and overlapping tool logos
A prompt types itself and branches out into three agents, each with a role and overlapping tool logos
A split feature: on the left a serif headline that settles in word by word, an intro, a rounded pill that copies the install command on click and a sliding-marker pill; on the right a muted transcript panel where a short exchange with the agent appears message by message, one reply carrying a page preview that blurs in and drifts on scroll.
Example usage of Agent Tree.
Organizes missions, thinking, tool execution, human approval, and evidence into one continuous agent product page.
A Ferrowick agent workspace: sessions grouped by day with a search that go into a sheet on phones, a thread where thinking, tool calls that open to their input and result, result cards and a request waiting for Allow or Deny arrive as the agent works, and a composer that takes photos and files, tools and skills, queues what you write while the agent answers and sets how hard it thinks on a ramp of five dots you tap or drag. Plug your agent into onSend, onApprove and onDeny.
A complete, production-ready blog for Next.js 16 that AI search engines can actually read. Fully prerendered routes, a typed schema.org graph, sitemap and RSS and IndexNow, Open Graph images, answer capsules and question headings for retrieval, and an editorial policy page. Inherits your shadcn theme rather than imposing one.
Seven skills that take a blog from install to published: finish the setup for your site, research what to write with real search and AI citation data, plan the topic clusters, write posts to the GEO playbook, refresh them without falsifying dateModified, audit before publish, and publish with the IndexNow response code actually read. Installs into your project's .claude/skills directory, so any agent working in this repository picks them up. Also available as a Claude Code plugin and through npx skills add.
Posts as MDX files in your repository, which is the storage option a coding agent can genuinely maintain. Includes the content source adapter, the single MDX compilation point, two local remark plugins, and two seed posts written to the GEO playbook so they double as the format specification.
A publish endpoint that revalidates the metadata routes as well as the page routes, because sitemap.xml and feed.xml are cached route handlers that a naive webhook leaves stale, and then submits the URL to IndexNow with the response codes surfaced rather than swallowed.
The components a blog post is made of: post cards, byline with rel=author and machine-readable dates, answer capsule, server-rendered table of contents, author bio, related posts, breadcrumbs, category pills, and real anchor-based pagination. Styled entirely with your shadcn theme tokens, so it inherits your design system rather than replacing it.
Fully prerendered /blog index, post, category, tag, and author routes. generateStaticParams returns every slug, there is no client component in the article render path, and every segment spreads shared metadata defaults so og:site_name and max-snippet survive Next.js's shallow merge.
sitemap.xml built from real content dates rather than the build clock, an environment-guarded robots.txt that declares your AI access preferences, an RSS 2.0 feed, and a 404 that returns a real 404.
An editorial policy page with a real review and corrections process. This is the Trust surface a search quality rater looks for and almost no blog has one, which makes it unusually cheap to differentiate on.
The full blog, minus the two items a reader never sees: no /api/publish route and no secrets in .env, and nothing written into .claude/. Everything else matches @agentblog/blog. It cannot be smaller than that: the routes share one sitemap and one layout, app/sitemap.ts emits category, author, tag, and /editorial-policy URLs, and the blog layout links /editorial-policy and /feed.xml, so dropping those routes would publish a sitemap of 404s rather than a smaller blog. Install this when you want the blog without a publish endpoint to secure.
AI coding agents, IDEs, and copilots
AgentMail logo with icon variant and dark/light modes
Two agents edit a brief like teammates: a named pointer glides to a paragraph and its suggestion types itself in as a tracked change, a margin comment arrives after a typing mark, and a table lists every change.
A day of agents at a glance: totals that roll between Today and 7 days, requests approved in place, a fleet whose rows unfold their sub-agents, and when the last request leaves a check draws itself where the list was.
Showcase card for an AI agent or assistant with avatar, name, description, capability badges, sample conversation preview, stats row, and action buttons. Perfect for agent directories and AI product pages.
A composed, app-controlled AI agent workspace that composes Agent Run Timeline + Tool Call Activity + AI Response Stream + Source Citation Rail into one realistic full page; the app owns every state and it never simulates a model.
A Kivuline Roasters dashboard on the app shell: a sidebar with workspaces that goes into a sheet on phones, a top bar with search and a bell, a figures band, the agents that need you with Allow and Deny, recent activity and a usage ring. Allow a request and the whole page answers at once: the row folds away, the counts roll down, a line lands at the top of the activity and the ring grows. Plug your app into onApprove, onDeny and onWorkspaceChange.
A whole assistant screen: starter cards, reasoning, tool calls, an approval before it emails anyone, and a streamed answer, with past conversations beside it.
Floating AI assistant with document-aware writing actions and local chat history.
Showing 601–624 of 25,861