Features & Pricing

Everything you need to catch what AI breaks

Seven specialized risk-hunting MCP tools for your editor, instant web checks, and automated GitHub PR reviews — backed by transparent monthly pricing.

Editor Integration

All Seven MCP Risk-Hunting Tools

Available directly inside Claude Code, Cursor, Antigravity, and Windsurf via the standard Model Context Protocol.

explain_change

Plain-language walkthrough of the diff citing exact file:line locations so you understand what AI wrote.

// explain_change output 1. src/cart.py:14 initializes CheckoutSession with customer id 2. src/stripe.py:28 creates payment intent without idempotency key 3. src/orders.py:44 updates order status to pending confirmation

business_risk_review

Hunts lost money, double charges, missing auth, and state corruption with concrete business impact.

// business_risk_review output ### [HIGH] Double-charge condition on network retry - Where: src/checkout.py:42 - What happens: Network timeout retries charge without idempotency key - Fix: Pass idempotency_key=f"order_{cart.id}"

scenario_questions

Generates targeted failure-scenario quiz questions and suggested answers citing exact risky lines.

// scenario_questions output Q: What happens if redis.acquire_lock() fails at lock.py:54? <suggested answer citing lock.py:54> Fails open, allowing concurrent workers to double-process the payout.

risk_tests

Produces edge-case test specifications and runnable code targeting the specific failure modes in the diff.

// risk_tests output def test_concurrent_checkout_prevents_duplicate_charge(client): # Targets race condition in src/checkout.py:42 responses = run_parallel([client.post("/pay") for _ in range(2)]) assert sum(r.status_code == 200 for r in responses) == 1

flow_check

Traces cross-file data flows (payment, auth, data) and flags where the change breaks business invariants.

// flow_check output [payment] billing.py:18 creates invoice -> charge.py:33 charges card BREAK at charge.py:38: status updated before card authorization succeeds Consequence: Unpaid orders are marked fulfilled on card decline

pre_deploy_checklist

Actionable operational checklist covering database migrations, environment keys, timeouts, and rollback steps.

// pre_deploy_checklist output - [ ] Run migration 0042_add_idempotency_keys.sql (models.py:19) - [ ] Verify STRIPE_WEBHOOK_SECRET is set in worker env (billing.py:104) - [ ] Rollback plan: Revert worker image; migration is backwards-compatible

security_audit

Audits code for leaked credentials, tokens, private keys, and personal data with severity, line citations, and fixes.

// security_audit output DiffVet security audit: Found 2 potential secret(s) or PII item(s). ### [Stripe Key] Line(s): 14 - Danger: Exposes customer payment details and direct charge ability - Fix: Remove secret and inject via STRIPE_SECRET_KEY env var

Everywhere You Ship

Web Checks & GitHub PR Reviews

Check code interactively in your browser or let DiffVet automatically guard every incoming pull request.

WEB INTERFACE (/CHECK)

Interactive Web Code Check

Paste a unified git diff or raw source snippet directly into DiffVet for instant analysis. Perfect for quick reviews, terminal-free inspections, and verifying snippets before committing. Features automatic in-memory secret masking so credentials never touch the LLM or storage.

// Web check execution result banner We found and masked 2 secret(s)/personal data item(s) before analysis (1 api key, 1 email) src/checkout.py:42 - High risk: Double-charge condition on timeout retry Counts 1 check on your unified monthly meter

GITHUB APP INTEGRATION

Automated GitHub PR Reviews

Connect DiffVet to your GitHub repositories with one click. When pull requests are opened or synchronized, DiffVet automatically inspects the changed files, reconstructs execution paths, and posts clear, advisory-only risk findings directly as PR comments. Never blocks merges.

// Automated GitHub PR review comment ## What changed: Added Stripe webhook idempotency handling ## Execution path: 3 steps citing exact PR lines (billing.py:104) ## What could break: 2 failure-scenario questions

Transparent Pricing

Simple plans for solo engineers and teams

Start free or upgrade to Pro to unlock DiffVet inside your AI coding assistant.

Free

For developers exploring AI-assisted development and solo projects.

$0 / month

20 checks/month: web code checks and GitHub PR reviews
Paste or upload code on the web (/check)
Automated GitHub PR review comments
Automatic in-memory secret & personal data masking
Exact file:line citation guarantees
Standard community documentation

Get started free

Recommended

Pro

For professional engineers and teams shipping production code.

$15 / month

2,000 checks/month: everything in Free, plus DiffVet inside your AI tools
DiffVet inside Claude Code, Cursor, Antigravity, and Windsurf via MCP
All 7 hosted MCP risk-hunting tools (including security_audit)
Generate and manage private API keys
Unlimited connected GitHub repositories
Automatic secret & personal data masking
Priority model execution pipeline and support

Get started with Pro

FAQ

Frequently asked questions

Clear, direct answers about checks, privacy, and subscriptions.

What counts as a check?

Each web submission on /check, each automated GitHub PR review comment, and each MCP tool invocation (e.g. business_risk_review in Cursor or Claude Code) consumes exactly 1 check from your shared monthly meter. Unparseable inputs or syntax errors are never counted.

Is my code stored?

DiffVet does not store your code. Diffs and code snippets are analyzed in memory and never written to our database or disk. Code is processed by our AI provider (Google Gemini) under their API terms, and DiffVet itself never uses your code for training. Known secret formats are masked before analysis.

Can I cancel anytime?

Yes. You can cancel your Pro subscription anytime directly from your dashboard billing settings with a single click. You retain full Pro access until the end of your current monthly billing period with zero cancellation fees.