Get 50% off Pro

See plans
profuncbeta
Trusted by 5000+ Developers worldwide
For Python and API backends — inside VS Code.Python

Diagnostic Intelligence for modern backend software

AI generates code faster than you can reason about it. Profunc traces runtime dependencies, maps affected endpoints, and gates commits with evidence-backed risk analysis — so what breaks is visible before you ship.

Use Cases

What breaks if I change this?

In AI-assisted codebases, the gap between what was changed and what was affected is invisible. Profunc answers the questions that matter before something reaches production.

Use case
What breaks if I change this?
You merged a change that touched one route handler. Two days later, three other endpoints regressed — and nobody knew they were connected. Profunc traces the runtime call graph before you push, mapping every endpoint, service, and dependency the change actually touches.

What changes

You see the full impact before the PR is open — not after the incident is filed.

Use case
AI wrote the endpoint. Now it crashes.
The endpoint returns 500. There's no test, no mental model, and no one on the team who wrote it. Profunc traces the live request through every route handler, model, and database call — surfaces exactly where it broke, and generates a failing test from the real execution so the same bug can't hide again.

What changes

From error log to root cause and test in minutes — not hours of reading generated code.

Use case
How do I know my functions still behave correctly?
AI rewrites functions faster than you can watch them. Profunc checks every run against your defined contracts and last-known-good baseline — flagging timing regressions, unexpected outputs, and contract violations the moment something deviates, in the dashboard, before deployment.

What changes

Silent regressions don't reach production. They get caught in the run.

How It Works

Select a function. See the blast radius.

Point Profunc at any function or endpoint — it traces the execution path, checks regressions against your baseline, and produces a runtime evidence report. From selection to full diagnostic in seconds.

Step 01
Run it. See what it actually does.
Select any Python function or API endpoint. Profunc generates AI payloads, runs them against your live code, and captures the full execution path — route handlers, queries, timing, memory. No fixtures. No configuration. Real data, in seconds.
Step 02
Catch what changed before it ships.
Every run is automatically compared to your contracts and baseline. Profunc surfaces timing regressions, unexpected outputs, and contract violations — showing you exactly what changed, what broke, and how far the impact spreads.
Step 03
Every commit is evidence-backed.
Profunc generates tests from the real execution trace and runs an AI risk assessment against your diff. The git hook blocks the commit if risk is high — with the evidence attached, not just a warning.
Plans

Start free. Investigate deeper with Pro.

Both plans give you the full diagnostic toolkit. The difference is who reads the output.

Free
The complete diagnostic workflow.
The complete diagnostic toolkit, running locally. No subscription. Bring your own provider key.

Free forever

OpenAI, Anthropic, Gemini, or Bedrock — your key, your cost.

Know which functions, routes, and downstream callers are affected before you push — blast-radius tracing and dependency mapping on every change, automatically.

Regression detection flags behavioral changes automatically. AI generates test payloads from real traces — no fixtures, no config.

Evidence-backed commit gating via git hook — risk analysis runs before every push, with your own provider key (OpenAI, Anthropic, Gemini, or Bedrock).

ProHosted AI
Your AI investigator. Always on.
A hosted AI model that reads your execution traces and explains what it finds. Root cause, blast radius, and the exact call path that broke — no API key to manage.

$7.99 monthly • $79.99 annual

Hosted AI included. No provider key, no config, no token bills. Open the extension and it works.

A hosted AI model that reads your execution traces and explains what it finds — root cause, blast radius, and the exact call path that broke. No API key to manage.

The Diagnose run: feed any endpoint a request and get a full execution walkthrough — route handler, models, queries, timings, and what deviated from baseline.

Unlimited run history synced across sessions — track how your system's behavior drifts across every deployment, not just the last ten runs.

Upgrade to Pro

Open the extension, run a Diagnose, and get a full execution walkthrough — route handler, models, queries, timings — with root cause surfaced automatically.

FAQ

Questions developers ask before they install.

Straight answers on what Profunc is, why it exists, and what changes when you move to Pro.

What is Profunc?
Profunc is a diagnostic layer that helps developers understand how backend systems behave, change, and break — especially in AI-assisted codebases where architectural understanding disappears quickly.
Why do developers need a diagnostic layer?
AI generates code faster than your team can build a mental model of it. You didn't write these dependencies — so you don't know which endpoints they touch, what contracts they break, or how the timing changes under load. Profunc answers those questions from live execution data. Before any of it ships.
How is this different from a debugger?
A debugger shows you one execution, once, when you're already looking for the problem. Profunc runs continuously: it traces every profiled function, compares the result to your baseline, flags the regression before deployment, and generates a test so the same failure can't recur silently. It also shows you the blast radius — the other parts of the system that same change could affect.
Which AI providers does Profunc support?
On Free, bring your own provider key (OpenAI, Anthropic, Google Gemini, or AWS Bedrock) — Profunc uses it for payload generation, test output, and commit risk analysis. On Pro, a hosted model is included: no key to configure, no token bills, no vendor to choose. You open the extension and it works.
What is free and what requires Pro?
Free runs the complete local diagnostic workflow — blast-radius tracing, payload generation, contract assertions, regression detection, test generation, and evidence-backed commit gating. It's not a trial or a feature-limited demo. Pro adds a hosted AI model that reads your execution traces and explains what it finds — root cause analysis, a Diagnose run that walks any endpoint from request to response, the full API workbench, and unlimited run history across every deployment.
I'm on monthly. How do I switch to annual?
Go to your profile page, open the billing section, and select the annual plan. You'll see a prorated charge preview before anything is confirmed. Do not purchase a second subscription — just switch cadence from the account page.
See the blast radius before your next deploy.

Stop shipping what you can't explain.

Start free — full diagnostics, your machine, no subscription. When you want the root cause handed to you instead of hunting for it, that's Pro.