Software verification

Find what breaks, what leaks,
and what nobody knows first

Deploy anxiety comes from not knowing what will break. Prober finds where features break, where permissions and personal data leak, and the parts of an unfamiliar system nobody understands yet. Every finding comes with evidence, severity, and how to fix it.

Three products

Split by when you use them and who uses them. All three work the same way: find it first, attach evidence, and suggest a fix.

Before you deploy

prober QA

Find what breaks before you deploy

Before merge, see what changed and what to check. Before deploy, run it in a real browser and keep the evidence.

  • Requirements vs. implementation table · PR handoff
  • Reviews by role, published only after a person approves
  • Test scope pulled from the change
  • Test cases · regression · pre-deploy smoke
  • Before-and-after data as evidence

Good fit for Teams where developers and PMs handle QA without a dedicated QA team

Start for free
Before a breach, before it becomes a problem

prober Security & Privacy

Check who can see what, and whether you keep your promises

Prober follows the code and API responses to find where permissions leak and where personal data or tokens show up. It also checks whether the product behaves the way your terms and privacy policy say.

  • Whether other users' data opens (IDOR · privilege escalation)
  • Personal data and tokens in responses, logs, and bundles
  • Input boundaries such as injection and SSRF
  • Terms and privacy policy vs. actual behavior
  • Required notices · consent flows · hidden cancellation paths

This is not a penetration test, a certification audit, or legal advice. We compare and find; where a legal call is needed, we mark it as such.

Good fit for Teams facing due diligence, a security review, or app store review

Start for free
When you take over a system

prober System Map

See an inherited system's structure on one page

CI collects scattered database schemas into one map, and every PR gets a comment listing changed tables, columns, and risks. Built for FDEs who step into a client's system and need to understand it.

  • Service CI collects DB schemas into one map
  • Changed structure and risk on every PR (dropped columns, NOT NULL added without a default, and more)
  • Layer concepts and flows on top of tables and zoom in
  • Extract on your own machine before anything goes to the server
  • AI agents find tables over MCP

Good fit for FDEs, SI teams, and consultants who analyze and verify client systems

Start for free

Wherever a finding comes from, the report looks the same

A finding from a feature and a finding from your terms go into the same four fields. That way you can decide what to fix first in one pass.

A security & privacy example, based on a fictional service

Finding
When a user deletes their account, their payment history is deleted right away. But the privacy policy says it is kept for 5 years.
Evidence
privacy-policy.md L42 "Kept for 5 years as required by law" ↔ user.service.ts L118 hardDelete(payments)
Severity
HIGH May violate a legal retention requirement, and contradicts what users were told
Suggested fix
Anonymize payment history instead of deleting it, and add a separate job that deletes it when the retention period ends. Update the policy so it lists what you actually keep.

Use a tool, an agent, or our team

For QA and Security & Privacy, pick what fits your team. The tool and the agent read the same verification records.

SaaS

Console

Your team reads and edits handoffs, reviews, and test results in one place. Free to start.

Agent

Agent

A Claude plugin runs scans, smoke tests, and reviews. Results are imported into the console over MCP. We help you install it when you contact us.

Service

Consulting

Our people work with you on pre-launch audits, QA setup, and security assessments. Tell us the scope and we'll send a quote.

Six lenses behind QA and Security & Privacy

If every test passes but nobody can use it, the product still failed. So we look beyond features to the people using it, your terms, and your code.

Angle 01 · QA

Functional QA

Does it work as specified?

  • Test cases and regression suites
  • Pre-deploy smoke tests in a real browser
  • Cases designed with boundary values and equivalence partitioning
Angle 02 · QA

Usability and personas

Can a first-time user get to the end?

  • Each persona's scenario, followed to the end
  • Where people get stuck or misread the text
  • Onboarding, empty states, and the steps to a first success
Angle 03 · QA

UI, UX, and design

Does the screen make sense on its own?

  • Whether the same concept looks the same on every screen
  • Empty, loading, error, and overflow states
  • Accessibility basics, responsive layout, and information hierarchy
Angle 04 · Security & Privacy

Security

Where would an attacker try first?

  • Whether one user can open another user's data (IDOR, privilege escalation)
  • Injection, XSS, file uploads, and SSRF
  • Tokens and personal data leaking through responses, logs, and bundles
Angle 05 · Security & Privacy

Policy and legal

Does the product behave the way your terms say?

  • Terms of service and privacy policy vs. actual behavior
  • Required notices and consent flows
  • Dark patterns and hidden cancellation paths
Angle 06 · QA

Code and structure

What will give way first?

  • Paths where missing error handling loses data
  • Transactions, concurrency, and consistency
  • Performance traps, structural debt, and dead paths

Also covered across the six angles

Performance and loadToo slow to use is a defect too
CompatibilityBrowsers, devices, screen sizes
Data integrityMigrations, duplicates, data loss
Observability and recoveryLogs, alerts, rollback paths
InternationalizationLanguages, time zones, currencies
SEO and sharingMeta tags, OG cards, crawling

How it works

Checklists live as markdown files in git. Run records are kept in the console.

1. Scope

Decide what to check. We map the scope and point out the gaps.

2. Write

Write checklist items, or pull them from the code that changed.

3. Run

Run them in a real browser and keep the before and after state as evidence.

4. Report

Add evidence and a severity to each finding, and suggest a fix.

Free for now

When you sign up, you can try Team features for 14 days. No card needed. When the trial ends you move to Free and your data stays. We don't take payments yet.

Free1 project · MCP connection · Team invites
14-day Team trialUnlimited projects · Starts when you sign up
ConsultingTell us the scope and we'll send a quote · Contact us

Contact us

Ask about adoption, consulting quotes, or anything else. We'll reply by email. Prefer email? Write to support@prober.im.