Capabilities
What STVI verifies, and what it does not.
Each method below is marked by what has actually been shown on a real project. Nothing is advertised that has not been done.
What STVI can verify
From one small function to an entire product journey.
Not every assessment performs every method. Which methods apply depends on your project, your evidence and the environment STVI is permitted to run in. Where something cannot be run, the report says so and says what would settle it.
Unit Testing
Verify focused behaviour in individual functions and modules.
Shown Run on a real project, on small self-contained parts.
Module & Integration Testing
Check whether components work correctly together.
Planned Where the environment permits. Not yet shown on a real project.
API Testing
Verify inputs, outputs, errors and contract behaviour.
Planned Where the environment permits. Not yet shown on a real project.
End-to-End Testing
Follow complete user journeys where the environment permits.
Planned Where the environment permits. Not yet shown on a real project.
Regression Testing
Check whether existing behaviour remains intact.
Planned Needs an earlier build to compare against. Not yet shown on a real project.
UI Testing
Assess observable interface behaviour and user-facing flows where the environment permits.
Planned Where the environment permits. Not yet shown on a real project.
Negative & Boundary Testing
Look for failure conditions, edge cases and boundary behaviour.
Shown Run on a real project, on small self-contained parts.
Security & Access-Control Checks
Check input handling, signatures and access rules where the evidence and environment permit.
Limited Input-handling and signature checks only. This is not a security assessment.
What can be run today
Honest limits, from the one real assessment so far and the safety rules STVI runs under.
| Project type | What STVI can run |
|---|---|
| Python code | Tests STVI writes run in isolation with the network blocked. |
| JavaScript and TypeScript code | Only parts that use nothing but built-in modules can be run. The network is not blocked for these runs and the run record says so. |
| Anything needing installed packages, a database, a running server or a browser | Cannot be run yet. In the one real assessment, 12 of 15 claims could not be checked for this reason. STVI says so for each claim and what would unblock it. |
The verification gap
Your test count isn’t your verification coverage.
- Requirements
- Claims
- Existing Tests
- Coverage
- Missing Tests
- Execution
- Evidence
- Verdict
STVI looks across the entire chain instead of treating the existence of tests as proof that the product is verified.
What STVI does not do
- A report does not establish that the software is correct, secure, accessible or compliant.
- It does not perform penetration testing or usability testing.
- It does not test accessibility.
- It is not independently validated: results are produced and checked by AI sessions, and no person reviews them.