Site Doctor — WordPress diagnosis

Site Doctor — read-only WordPress diagnosis for your AI

[Site Doctor — WordPress diagnosis MCP connector – tool definition quality and endpoint health on Glama]

Connect your own AI (Claude, ChatGPT, Cursor, Codex…) to any WordPress site through a 100% read-only MCP connector and ask what's wrong — security, performance, forms, logs, downtime — with evidence. The connector never writes to the site, and your AI never sees the site's credentials.

Website: https://sitedoctor.syswp.pro · Docs: https://sitedoctor.syswp.pro/docs

What it is

Site Doctor is the neutral, read-only diagnosis layer for WordPress. Ask your AI why a site is broken and get the root cause with evidence — without giving any AI permission to write to the site. Built for agencies and freelancers maintaining 5–100 client WordPress sites.

Three parts:

• SaaS dashboard — six monitors, alerts, findings and a durable intervention timeline per site.
• One-file WordPress connector — read-only, HMAC-signed.
• MCP server (OAuth 2.1) — so your own AI can reach it.

Connect your AI to WordPress (MCP)

1. Create a free account at https://sitedoctor.syswp.pro and add a site.
2. Install the one-file read-only connector plugin (upload it, paste the connection key from the panel). It exposes no write operations.
3. Add Site Doctor as an MCP server in your AI client using OAuth 2.1 / PKCE. Tested with Claude, Claude Code, ChatGPT, Codex, Cursor, Trae and VS Code — or use the built-in "Doc" chat in the dashboard.
4. Ask about any site you manage in plain language. The AI reads only the evidence needed to answer; secrets are redacted before anything reaches the model.

Why read-only (the trust model)

• The connector exposes no write operations — it physically cannot modify, clean or break a site.
• Your AI never receives credentials — wp-config constants (DB password, salts), password hashes and the connector's own key are redacted before any output reaches a model.
• Every request is HMAC-SHA256 signed with a nonce and ±5-minute skew; SQL is read-only; the MCP uses OAuth 2.1 (PKCE); tenants are isolated per organization.
• On the Agency plan and above, a fix is prepared as a one-click activatable mu-plugin — a logged-in admin turns it on from wp-admin. Never a silent write.

What it diagnoses (read-only, on real client sites)

• Security & compromise — skimmers, backdoors, tampered core, ghost admins; the cleanup prepared.
• False-positive triage — tells a real injection from a harmless file.
• Performance — real coverage, render-blocking scripts, the cause of slow loads.
• Downtime — root cause found from outside the site, because the uptime check and auto-investigation run externally.
• Cron / scheduled posts, lost ad attribution, forms & leads, change tracking, and managed-host blind spots (WP Engine / Pantheon / Kinsta, where you usually can't read debug.log).
• Offline CVE matching against a mirror of the Wordfence vulnerability database.

Pricing

Self-serve at https://sitedoctor.syswp.pro/#precios

Links

• Home: https://sitedoctor.syswp.pro
• Docs: https://sitedoctor.syswp.pro/docs
• Use-case library: https://sitedoctor.syswp.pro/casos
• Free outside-in scan (no account): https://sitedoctor.syswp.pro/scan

Built by SysWP. Questions and trials: [email protected]