Prerender Buddy developer tools
Choose public diagnostics, read-only workspace MCP, managed rendering or the self-hosted engine.
On this page
Overview
Choose the tool for the evidence or operation you need: inspect a public HTTP response, read saved workspace findings, use managed browser rendering or operate the rendering engine yourself.
Managed Prerender Buddy
Use the managed product to review website Health, recorded crawler activity and AI Visibility in an eligible workspace. Where a public page has a confirmed JavaScript delivery gap, managed rendering provides crawler-readable HTML through the configured integration.
Account features and rendering capacity depend on the effective plan and configuration. Managed rendering, AI monitoring and creative generation are separate capabilities; access to one does not imply every other feature is included.
Chrome extension
Run a crawler-readability check for the current public page without leaving Chrome. Use it as a quick entry point for investigation, then review the technical evidence before changing the website.
Public HTTP diagnostics: CLI
Use the CLI in a terminal or CI workflow to inspect returned HTML, compare normal and crawler-profile HTTP responses, and check discovery files.
npx @prerenderbuddy/cli check https://example.comPublic diagnostics require no PB account. They do not execute JavaScript or provide managed browser rendering. A response comparison shows what those HTTP requests returned; it does not establish indexing, a citation or an operator's identity.
Local MCP: public checks and workspace evidence
Let a compatible MCP client run the public diagnostics through a local stdio process:
npx --yes @prerenderbuddy/mcp@0.2.0The published package requires Node.js 20 or newer. Without a key it provides public-page checks. With an eligible scoped key passed through PRERENDER_BUDDY_API_KEY, it also provides read-only workspace tools for sites, Health, crawler activity, AI Visibility, recommendations and content status.
Workspace mode returns bounded evidence and metadata, not full provider answers or article bodies. It does not start runs, change settings or publish content. It is a local process, not a hosted PB MCP endpoint.
Keep credentials in a trusted process/client configuration rather than prompts or repository files.
Agent plugin
The PB agent plugin adds reusable audit guidance alongside the same local MCP server. It helps an assistant gather evidence and turn it into prioritized recommendations.
Compatibility is documented per host: the current repository records a locally tested Codex alpha, portable packaging and Claude Code packaging awaiting native validation. Check the maintained compatibility and installation notes before use. A packaged manifest does not establish marketplace availability in every client.
Self-hosted rendering engine
Use the open rendering engine when your team wants to operate the browser-rendering layer. The supported release uses Docker. Your team owns routing, networking, capacity, security, monitoring, updates and incident response.
The engine is the rendering core; it does not include the managed workspace's AI monitoring, billing, support Agent or Ads workflows.
Choose the next step
| Need | Start here |
|---|---|
| Inspect one public page | Public checker, Chrome extension or CLI |
| Give an assistant public HTTP diagnostics | Local MCP without a key |
| Read saved workspace evidence in an assistant | Eligible scoped workspace MCP |
| Use a repeatable agent audit workflow | Agent plugin, subject to client support |
| Serve browser-rendered content through a maintained service | Managed rendering after confirming the need |
| Operate the rendering core yourself | Docker engine and self-hosting guide |
For endpoint-level integration, use the Developer API. Keep its authenticated capabilities and scopes distinct from the narrower read-only MCP tool set.