Trust / Evidence

Security claims need evidence, not adjectives.

This page distinguishes public documentation from independent verification and identifies gaps rather than inventing reassuring statistics. Status snapshot: .

Evidence status

What a reader can verify

Published repository material is a primary source for project intentions and development records. It is not automatically proof that a user's device is protected against every threat.

PUBLIC SOURCE

Source and security policy

Core source, defensive test conventions and responsible reporting policy are public. The core README describes bounded static inspection, integrity controls, and fail-closed behavior. Validate those features against a particular version and release before relying on them.

Review source → · Report vulnerabilities →

PROJECT-REPORTED

Engineering test records

Release metadata records automated Linux and Windows check totals, setup checks, and a Windows verification flag. These are project-reported engineering checks, not an independently supervised penetration test or a representative real-world malware detection study. Counts and pass flags must be tied to the exact build and CI run.

Inspect release evidence metadata →

NOT ESTABLISHED

Independent audit or certification

No external audit report or third-party security certification has been substantiated here. The project should only announce such an assessment after publishing the auditor, scope, date, result, and limitations.

NO PUBLISHED RATE

Detection accuracy

Representative sensitivity, false-positive rates, and benchmark methodology are not available as independently validated public metrics. We do not claim “100% secure,” “zero false negatives,” or protection from every attack.

NO INSURANCE CLAIM

Insurance and compensation

These public project materials do not establish a consumer insurance policy, insurer, coverage amount, eligibility rules or claims process. MalGuard must not advertise that it insures users or guarantees reimbursement without a real, independently verifiable policy and terms.

How to read a scan

Unknown remains unknown.

A static inspection may identify suspicious structures, but unsupported formats, opaque archives, insufficient evidence or unavailable isolation cannot become a safety guarantee. Inconclusive findings require caution, even when no known malicious signature is detected.