AI Error Diagnosis & MCP Server for PHP

AI diagnosis

Errors that explain themselves

Paste an OpenAI or Anthropic API key in the dashboard and PostRequest can diagnose any captured error with an LLM — on demand from the wand button, or automatically in the background as new faults arrive.

Because the diagnosis is grounded in your actual code — the stack trace, request context and the real source around the fault — it reads like a senior engineer who already knows your codebase triaged the ticket.

Root-cause analysis, suggested fix and a confidence score
Ask follow-up questions in a chat on any error
API key stored encrypted at rest — never leaves your host
Fully optional — the tool works without it

AI diagnosis

TypeError: null given 92% confident

Root cause. getUser() returns null for logged-out sessions; the caller dereferences it without a guard.

Fix. Add a null check before ->profile, or make the route require auth.

Model Context Protocol

Connect Claude to your fleet

PostRequest ships a built-in MCP server. Point Claude Desktop, Claude Code or any MCP client at your dashboard and it can investigate your monitored sites directly — no copy-pasting stack traces.

One endpoint, your whole estate

Add the server once. The AI reaches every client through a single authenticated endpoint and always checks for an existing diagnosis before re-reading source — fast, and cheap on tokens.

Pretty endpoint at https://your-host/mcp
Bearer prq_… token from Settings → API Access, or OAuth 2.1
Streamable HTTP (JSON-RPC 2.0) — works with any MCP client
Client-bound tokens scope the AI to a single site

claude_desktop_config.json

{
  "mcpServers": {
    "postrequest": {
      "url": "https://monitor.example.com/mcp",
      "headers": {
        "Authorization": "Bearer prq_your_token"
      }
    }
  }
}
// Now: "Which client has new errors, and why?"

Grounded in your real code

Decisions from your actual codebase — not guesses

PostRequest already retrieves each client's source for inspection. Over MCP, your assistant can read across that whole codebase — the real files behind the real errors — and reason about fixes with full context instead of hallucinating from a stack trace alone.

Ask it to explain a fault, propose a patch or weigh a refactor, and every answer is anchored to the code you actually run — if you choose to let it look.

You decide — reading source is opt-in and strictly read-only
Source stays encrypted at rest, on your own server
Your key, your model — nothing is used to train a third party
Suggestions cite the exact file and line they came from

Grounded suggestion

src/Http/UserController.php line 42

“Your show() method calls $this->users->find($id) and uses the result without a null check. Add a guard, or a findOrFail() — here's the patch for your code.”

What the AI can see

A read-only investigation toolkit

Every tool is read-only. The AI can look, correlate and explain — it can never change a client, run code, or write to your data.

Errors & diagnoses

Browse & search failures

List and search faults across one client or the whole fleet, and pull the existing AI diagnosis first.

Health & lifecycle

Heartbeats & status history

Read the latest heartbeat — PHP version, extensions, disk, TLS — and the client's status history.

Database schema

Tables, columns & indexes

Inspect a client's structure to reason about data-shape and schema-drift issues.

DNS & sitemap

Records & discovered URLs

Check DNS records and the discovered URL map when triaging reachability or routing problems.

Source code

Read & search retrieved source

Pinpoint the exact line behind a fault in the retrieved source — only when needed.

Fleet overview

Every monitored client

List monitored clients so the AI can pick the right one — or sweep across all of them.

Private by design

AI on your terms, on your server

The AI features are opt-in and use your own API key. Nothing calls an LLM unless you enable it, the key is stored encrypted at rest, and the MCP tools are strictly read-only — so pointing an assistant at PostRequest never puts your clients at risk.

Get in touch

Want this watching your PHP sites?

Leave your email and we'll reach out to get you set up — AI diagnosis, MCP and all. No account, no obligation.

We'll only use your email to reply — no spam, never shared.