ProFunc

The diagnostic layer for AI-assisted software development

ProFunc is becoming the diagnostic layer for AI-assisted software development. It gives engineers runtime evidence, route verification, diagnose workflows, and blast-radius context before they accept, review, or ship code changes.

Today ProFunc is strongest in Python backend systems, especially Flask and FastAPI. It can also surface route and architecture context in mixed Python, JavaScript, and TypeScript repositories when analyzer or runtime evidence exists, and the product is moving toward a broader workflow for teams building with AI.

See pricing · Support · Terms · Documentation

Why Teams Use ProFunc

Keep AI-assisted changes grounded in runtime evidence instead of guesswork.
Investigate hotspots, history, alerts, reports, call graphs, and complexity signals in one place.
Diagnose endpoint behavior against the local runtime first, then fall back to HTTP when needed.
Inspect routes, handlers, request and response behavior, and profiling evidence before risky edits.
Trace blast radius through Impact Map before refactors spread into unknown code paths.
Export reports, pin compact cloud snapshots, and sync diagnostic artifacts across machines.
Maintain a durable local evidence cache, run maintenance, and clear heavy artifacts without wiping useful history.

What You Can Do Today

Profile Python functions, classes, and executable endpoints from CodeLens, hover actions, or the command palette.
Diagnose selected endpoints and generate structured reports from real executions.
Review hot paths, trends, alerts, history, and function detail inside Investigate.
Inspect discovered routes, request and response behavior, logs, and history diffs inside Routes.
Open Impact Map in trace, explore, or system views to move from one endpoint into downstream dependencies and architecture context.
Export function reports, pin cloud snapshots, and export pinned snapshots when evidence needs to travel.
Sign in, manage subscription access, and sync cloud data for connected workflows.

Core Workspaces

Diagnose

Run selected endpoints through the local runtime bridge or HTTP fallback, watch execution progress, review findings, and export a Diagnose report. Verify generated request payloads with a preview before execution, run robust validation of Python enums and Pydantic constraints, and execute batch-based diagnose runs in parallel or sequential modes.

Investigate

Follow hotspots through overview, functions, investigation, alerts, and history views so you can understand what changed and why it matters.

Routes

Inspect route inventory, handler evidence, request and response behavior, logs, history diffs, and profile workbench flows from one route-centered surface.

Impact Map

Trace blast radius from routes into services and dependencies, browse saved analyzer-backed graphs, and switch between focused trace and broader system views.

Account And AI Workflows

Cloud sync for profiling history and saved project artifacts across machines
Pinned cloud snapshots that keep raw payloads local-only while preserving shareable evidence
Hosted AI notes, docstring generation, and model-backed bottleneck guidance
Account-aware billing, sign-in, and entitlement flows inside the extension

MCP Server

ProFunc includes a local MCP (Model Context Protocol) server that exposes diagnostic and profiling data to AI coding agents like Claude Code, Cursor, Windsurf, and other MCP-compatible clients.

Available Tools

Tool	Description
`get_blast_radius`	Trace downstream dependencies and impact of a function or endpoint
`get_investigation`	Retrieve hotspot investigations with alerts, trends, and function detail
`get_endpoint_catalog`	List discovered routes with handler signatures and request configuration
`get_complexity`	Get algorithmic complexity analysis for functions in scope
`get_run_history`	Fetch profiling run history with timing and metadata

Setup

The MCP server starts automatically as a sidecar when the extension activates. To connect an external MCP client:

Stdio (recommended for Claude Code, Cursor, etc.):

Run ProFunc: Copy MCP Stdio Config from the command palette, then add the output to your MCP client configuration.

Streamable HTTP (for network-based clients):

The server listens on http://127.0.0.1:37842/mcp by default. Configure the port via profunc.mcpServer.port in VS Code settings.

Auto-Discovery

The server automatically resolves project identity from git remotes and discovers the profunc database across editor storage locations. No manual configuration is needed for typical single-project workflows.

Works Best With

Python
Flask
FastAPI
Mixed Python, JavaScript, and TypeScript repositories with backend entrypoints
Teams that want deterministic evidence before merging AI-generated changes

Quick Start

Open a workspace in VS Code.
Run ProFunc: Open Diagnostic Layer.
Start in Investigate for hotspots, Diagnose for endpoint execution, Routes for route behavior, or Impact Map for blast-radius context.
Profile or diagnose from CodeLens, hover actions, or the command palette.
Configure your OpenAI key or sign in when you want connected AI and cloud workflows.

Docs

License And Terms

This repository is proprietary software and not open source. Source code rights are described in LICENSE, and installation or commercial use of packaged copies is governed by TERMS.md.

For support, contact support@profunc.dev.

ProFunc - Diagnostic Layer for Python Backends

Dawanse