RenderMarkPublished withRenderMark

New Agent Features Launched in June 2026

By RenderMark TeamPublished July 17, 2026

In June 2026, RenderMark became much more than a browser-based Markdown renderer. We launched a complete agent-facing layer so AI assistants can render, export, publish, share, update, and manage Markdown documents without asking people to copy content between tools by hand.

The goal is simple: when an agent writes a good document, it should also be able to hand you a polished artifact. That might be a PDF, a DOCX file, a hosted RenderMark link, a temporary preview, a Google Doc, a rendered image, or a synchronized document from GitHub.

Here is everything that shipped in the June agent launch.

1. A RenderMark MCP server for AI agents

The biggest launch is the RenderMark MCP server, published as @rendermark/mcp-server.

MCP gives compatible agents a structured way to call RenderMark instead of improvising document conversion from scratch. Once connected, an agent can choose a specific RenderMark tool for the job: render a preview, export a file, publish a link, share a document, validate Markdown, or sync from GitHub.

The server now exposes 17 tools across rendering, publishing, sharing, management, sync, setup, and workflow guidance.

2. Rendering tools agents can call directly

Agents can now produce polished output from Markdown through dedicated rendering tools:

  • render_markdown converts Markdown into styled HTML with themes, templates, table of contents support, and syntax highlighting.
  • render_to_image creates PNG or JPEG previews when a chat, ticket, or review thread needs an image instead of a document link.
  • render_diff creates visual comparisons between two Markdown versions.
  • validate_markdown checks important documents for common formatting and quality issues before they are published.

These tools are built for the content agents already create: reports, specs, proposals, meeting notes, README files, changelogs, and internal documentation.

3. Export tools for real document files

The June launch also added agent-accessible exports:

  • export_markdown saves a Markdown document as PDF, DOCX, or HTML.
  • export_batch lets an agent export multiple Markdown files in one workflow.

This matters for practical agent work. If an agent creates a project brief, it can now hand you a PDF. If it updates a documentation folder, it can batch export the whole set. If a stakeholder needs an editable version, the same Markdown can become a Word document.

4. Publishing and sharing tools

RenderMark now gives agents first-class publishing workflows:

  • publish_to_rendermark creates a hosted RenderMark share link.
  • share_live_preview creates a temporary preview URL with an expiration window.
  • share_document shares an already published document with specific email addresses.
  • publish_to_google_docs publishes Markdown as a Google Doc for teams that review and comment in Google Workspace.

This closes the gap between "the agent drafted it" and "the team can use it." The agent does not have to stop at raw Markdown. It can produce a shareable, readable document and route it to the right destination.

5. Document management from the agent

Agents can now work with existing RenderMark documents instead of only creating new ones.

The new management tools include:

  • read_document to fetch a document by URL, slug, or ID.
  • update_document to update content, title, or settings.
  • list_documents to search and page through a user's documents.
  • delete_document to remove a document when explicitly confirmed.

This makes agent workflows iterative. An agent can retrieve a published spec, update it after a meeting, preserve the same share link, and keep the document current.

6. GitHub sync for repository docs

The sync_from_github tool connects repository Markdown to RenderMark.

Agents can import or refresh a Markdown file from GitHub, including README files and documentation pages. RenderMark also understands GitHub context, which helps relative links and images resolve correctly when repository content becomes a polished RenderMark document.

For teams that keep docs near code, this is the agent workflow: update the Markdown in GitHub, sync it to RenderMark, and share a clean document with people who do not want to read source files.

7. Browser-based API key setup

Agent publishing needs authentication, so June also added setup support.

The setup_api_key tool helps users configure RenderMark credentials through browser authentication. The CLI and MCP server can also use RENDERMARK_API_KEY or a local ~/.rendermark/config.json file.

That gives teams a practical path for both local agent workflows and scripted environments.

8. Built-in workflow guidance for agents

Agents are better when they know which tool to choose. RenderMark now includes get_rendermark_instructions, a guidance tool that explains capabilities, tool selection, workflows, defaults, and best practices.

The MCP server also exposes RenderMark-specific instructions at connection time. That guidance tells agents when to use RenderMark, when not to use it, and how to choose between publishing, exporting, rendering, validation, sharing, and GitHub sync.

9. MCP resources and prompts

The server now includes MCP resources for passive discovery:

  • rendermark://capabilities summarizes available tools and supported features.
  • rendermark://templates lists templates and themes.
  • rendermark://examples provides example workflows.

It also includes prompts for common tasks:

  • Render and preview Markdown.
  • Export a batch of Markdown files.
  • Publish and optionally share a document.
  • Sync a GitHub file to RenderMark.

These are small additions, but they make RenderMark easier for agents to discover and use correctly.

10. A CLI fallback for agents and humans

Not every environment connects MCP tools directly. The June launch added a rendermark CLI in the MCP package so agents and people can use RenderMark from a terminal too.

The CLI supports workflows like exporting Markdown to PDF or publishing a file and returning JSON. It gives coding agents a reliable fallback when MCP is unavailable, and it gives developers a simple manual path for the same rendering and publishing pipeline.

11. Agent Skill packaging

RenderMark now ships an open-standard Agent Skill in the MCP package.

The skill teaches compatible agents when to use RenderMark and how to choose the right workflow. It covers tasks like publishing a README, exporting Markdown to PDF, rendering a preview image, validating important docs, syncing GitHub content, and falling back to the CLI when MCP is not connected.

The skill does not replace MCP. It complements it. MCP provides the tools; the skill teaches the agent how and when to use them.

12. Claude Desktop MCPB packaging

For Claude Desktop users, RenderMark added MCPB packaging support.

The package includes a build script that creates a Claude Desktop bundle from the RenderMark MCP server. That makes setup easier for users who prefer installing a bundled server instead of editing MCP configuration by hand.

13. Hosted and local agent access

RenderMark now supports multiple agent connection paths:

  • Hosted remote MCP at https://mcp.rendermark.app/mcp for compatible remote clients.
  • Local MCP through @rendermark/mcp-server over stdio.
  • REST API access at https://rendermark.app/api/v1.
  • CLI fallback through the rendermark command.
  • Browser workflows for people who want to paste, preview, export, and publish manually.

That flexibility is intentional. Some agents run locally and need filesystem exports. Some run in hosted environments and need a remote endpoint. Some teams want API calls. RenderMark now supports each path.

14. Agent discovery files

June also added machine-readable discovery files:

These files make RenderMark easier for agents, catalogs, and developer tools to understand without scraping the whole site.

15. Agent-aware VS Code workflows

The VS Code extension also gained agent-focused behavior.

RenderMark for VS Code can automatically refresh previews when external processes modify Markdown files. That includes AI coding agents, build scripts, and other tools writing to disk. When an agent updates a file, the preview can stay current without manual reloads.

For developers reviewing agent-generated docs, this turns Markdown review into a live workflow: let the agent edit, watch the rendered document update, then export or publish when it is ready.

Why this launch matters

AI agents produce a lot of Markdown. They write implementation plans, incident summaries, product briefs, release notes, architecture docs, README updates, and handoff notes. But raw Markdown is not always the final deliverable.

The June 2026 launch gives agents the missing delivery layer:

  • Render the Markdown so people can read it comfortably.
  • Validate it before it goes out.
  • Export it to the format the audience needs.
  • Publish it to a stable link.
  • Share it with teammates.
  • Sync it from source-controlled docs.
  • Keep existing documents updated over time.

RenderMark now works as both a product for people and a document layer for agents.

Getting started

If you use an MCP-compatible agent, connect the local server with @rendermark/mcp-server or use the hosted endpoint when your client supports remote MCP. If you prefer direct integration, use the RenderMark REST API. If you want portable agent instructions, install the RenderMark Agent Skill from the package or follow the setup links in the rendermark-agents repository.

And if you just want to try the workflow yourself, paste Markdown into RenderMark, preview it, and publish or export the finished document from the browser.