Make Your Site Machine-Readable for AI Agents with AgentMarkup

Where this fits

You've just deployed a marketing site or a docs portal, and you notice AI crawlers are fetching pages but missing key context—pricing, FAQs, and product details scattered across unstyled HTML. Your team manually updates llms.txt and JSON-LD, but it drifts out of sync with every content change. You need a repeatable way to generate, validate, and repair machine-readable metadata directly from your JavaScript build pipeline.

Why agents benefit

• Automated configuration: The skill detects your framework (Vite, Astro, Next, Nuxt) and installs the correct AgentMarkup plugin without manual setup.
• Audit and repair loop: Runs a CLI audit against a deployed URL, flags missing or malformed metadata, and patches source files or output until the surface is coherent.
• Multi-format coverage: Handles llms.txt, llms-full.txt, JSON-LD, markdown mirrors, robots.txt AI rules, Content-Signal headers, and A2A Agent Cards in one pass.
• Dynamic content handling: Generates metadata from your actual routes and content, so it stays fresh after builds instead of requiring hand-maintained files.
• Fix-first workflow: When an audit fails, the skill proposes concrete edits (e.g., adding missing JSON-LD schema to a component) and applies them with your approval.

Practical scenarios

Auditing a legacy site before AI crawler access

Your client's e-commerce site has been live for years, but you discover AI bots are reading paginated category pages without structured data. You run the skill with the audit command, it scans the deployed site, and reports that llms.txt is missing and product pages lack Product JSON-LD. It then scaffolds the missing files and updates the robots.txt to allow AI crawlers on relevant paths—all within your Next.js repo.

Adding AgentMarkup to a fresh Astro docs site

You're building a new API docs site with Astro. The skill detects the framework, installs @agentmarkup/astro, and asks if you want to generate markdown mirrors for every page. It configures the plugin to output llms.txt and a full-text mirror, then runs a post-build audit to confirm the headers and JSON-LD are correct before you merge.

Fixing broken Content-Signal headers after a migration

After moving a Nuxt app to a new hosting provider, you notice Content-Signal headers aren't being served. The skill checks your _headers file, identifies the missing rules, and updates it to include the proper header syntax. It also verifies that the llms.txt URL resolves and that the A2A Agent Card points to the correct endpoint.

Add it to your agent workflow

Install the skill from OpenAgentSkill, which helps you discover and manage agent skills like this one:

npx skills add agentmarkup/agentmarkup --skill agentmarkup

Then ask your agent to "audit agentmarkup for the current repo" after a build. The skill will run the CLI, parse the output, and present a list of issues with suggested fixes—e.g., "Add @agentmarkup/vite to vite.config.js and generate llms.txt."

Compare before adopting

Before making this a permanent part of your toolchain, compare against alternatives like manual metadata scripts or generic SEO auditing tools. Look for:

• Quality signals: Does the audit catch real gaps like missing JSON-LD types or malformed llms.txt? Check the GitHub issues and release cadence.
• Maintenance freshness: This project is actively developed (24 stars, recent commits). Verify the skill stays compatible with your framework's major versions.
• Alternatives: Consider single-purpose tools (e.g., an llms.txt generator) if you don't need full coverage. The skill's value is in the unified audit/fix loop.
• Workflow fit: It requires shell and package-manager access. In sandboxed environments, it can still draft config, but it won't install or run audits.

Why it is worth tracking

AgentMarkup addresses a growing need: making websites understandable to AI agents. The skill's ability to go from audit to automated repair is a differentiator, and its framework‑specific plugins reduce friction. As AI crawlers become more prevalent, this type of machine-readable metadata will become a baseline expectation. Track it if you manage content-heavy sites or work with clients who care about AI visibility. Evaluate it when you next need to update your site's structured data or prepare for an AI crawler audit. You can also review the skill's page on OpenAgentSkill for community examples and version history.