symfony-runtime-mcp
MCP server that exposes Symfony profiler runtime data (requests, queries, logs) to AI agents like Claude Code for debugging and optimization.
README
skrepr/ai-performance-mate-extension
An AI Mate extension that exposes runtime data from the Symfony profiler to AI agents (Claude Code, Cursor, …). Where code intelligence looks at what sits on disk, this looks at what actually happens: request timing, Doctrine queries, N+1 patterns and query plans.
This is the successor to the standalone Node MCP server symfony-runtime-mcp: the same analysis tools, but now in-process as a Mate extension. That eliminates docker exec, a copied extractor script and host↔container path mapping — the tools run where the code and the database live.
For local dev environments only. Never point this at production.
Tools
| Tool | What it does |
|---|---|
slow_queries |
Slowest query shapes over the last N requests, with total time, execution count and origin (file:line + source code) |
detect_n_plus_one |
Repeated identical queries within a single request, with the likely parent query (1+N) |
request_breakdown |
Wall clock of a single request broken down per category (DB / external HTTP / rendering / listeners …) |
profile_diff |
Before/after comparison of the same endpoint: duration, memory, query count and disappeared/appeared query shapes |
explain_query |
Runs EXPLAIN and tells you why a query is slow (full scan, unused index, filesort). analyze=true for SELECT/WITH only |
Complementary to the official symfony/ai-symfony-mate-extension (which exposes raw profiler and container data): install both side by side.
Installation
composer require --dev skrepr/ai-performance-mate-extension
Already running AI Mate in your project (a mate/ directory exists)? Then you are done: the Mate composer plugin runs mate discover automatically after every composer require/update.
No AI Mate yet? The package pulls in symfony/ai-mate automatically; initialize it once:
vendor/bin/mate init # creates mate/ + mcp.json (Claude Code and friends pick those up automatically)
vendor/bin/mate discover # registers the extension; happens automatically from then on
The four profiler tools work right away: they read var/cache/dev/profiler. Make a few requests to your app and call request_breakdown, for example. Check with vendor/bin/mate mcp:tools:list that the tools are registered.
Configuration
All configuration is optional and goes through mate/config.php in your project.
explain_query — database connection
The extension does not know your app connection (Mate does not boot the kernel), so explain_query builds its own connection from a DSN. By default it falls back to DATABASE_URL from the environment; load that by setting mate.env_file:
// mate/config.php
return static function (ContainerConfigurator $container): void {
$container->parameters()
->set('mate.env_file', '.env') // loads .env and .env.local
;
};
mate.env_file expects a string (not an array) and requires symfony/dotenv in your project — already present in a standard Symfony app.
Using a different env var or multiple connections? Point at the DSN explicitly:
$container->parameters()
->set('mate.env_file', '.env')
->set('skrepr_mate.database_url', '%env(resolve:POSTCODE_DB_URL)%')
;
Query origins (strongly recommended)
Without backtraces, slow_queries/detect_n_plus_one know that a query is slow, but not where it comes from. Set in config/packages/dev/doctrine.yaml:
doctrine:
dbal:
profiling_collect_backtrace: true
With that, origin points to the exact file:line in your project code, including the surrounding source.
Custom profiler location
$container->parameters()
->set('skrepr_mate.profiler_dir', '%mate.root_dir%/var/cache/dev/profiler')
;
Known limitation: the Symfony bridge on PHP < 8.4
If you also use symfony/ai-symfony-mate-extension, its symfony-profiler-* tools fail on PHP < 8.4 with "Cannot generate lazy proxy": the collector formatters are final and ->lazy(), and before PHP 8.4 Symfony creates a lazy proxy through subclassing. On PHP 8.4+ this does not apply. Workaround without patching the vendor — redefine the formatters non-lazy in mate/config.php:
foreach (['Request','Exception','Mailer','Translation','Doctrine','Time','Logger','Memory'] as $name) {
$fqcn = "Symfony\\AI\\Mate\\Bridge\\Symfony\\Profiler\\Service\\Formatter\\{$name}CollectorFormatter";
$container->services()->set($fqcn)->tag('ai_mate.profiler_collector_formatter');
}
Requirements
- PHP ≥ 8.2
- Symfony 6.4 / 7.x / 8.x with the profiler enabled in dev (
symfony/web-profiler-bundle) - Doctrine DBAL 3.6+ / 4.x
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.