How to read a malware analysis report

Understand malware verdicts, evidence and incomplete analysis. Learn why a risk score is not a guarantee that a file is safe or malicious.

Start with completeness

Before reading the verdict, check whether analysis finished. A failed, timed-out or unavailable check is missing information, not evidence that a file is clean. Offline checks and server-assisted analysis can cover different parts of a file.

Read the evidence behind the label

A report can combine known-threat matches, file structure, suspicious strings and model scores. Ask what each finding actually establishes. An imported Windows function can have legitimate uses; its presence alone does not prove that a program performed a malicious action.

For example, a file with networking imports and a packed section needs context. A known-malware hash match is a different kind of evidence from a model finding unusual file features. Read the individual findings instead of treating every warning as equivalent.

A risk score is not a probability of infection

Nyxgard presents a risk score with supporting evidence. A score of 80 out of 100 should not be read as an independently measured 80 percent chance of infection. The interpretation depends on the scoring system, available checks and file type.

Separate explanation from observation

An AI explanation helps you read the evidence. It does not add a new observation of what the file did. Static analysis reads a file without running it; observing behaviour requires a separate execution environment and still only covers the observed run.

Choose the next action

Do not open an unexpected attachment simply because no warning appeared. Verify its source through a separate trusted channel, review the complete findings and seek help if the report is incomplete or contradictory. Detection can miss threats and can flag legitimate software.

In Nyxgard, keep the report context when contacting support. Do not attach suspicious files to a support message; use the supported analysis workflow when appropriate.

Technical references

For examples of why file contents and observed behaviour can be incomplete evidence, see MITRE ATT&CK on obfuscated files and sandbox evasion.

Continue learning

How Nyxgard works · Download availability · Ask for help