Crypto-View

Frequently asked questions

Using the tool

Is my code stored anywhere

No source code is kept. A scan fetches one commit into a temporary directory, reads it, then deletes the directory before it answers. What is kept is the findings and the CBOM: file paths, line numbers, the matched line and the algorithm. Only public repositories can be scanned here.

The command-line tool keeps everything on your machine. Scanning makes no external call at all. Two things can use the network and neither happens on its own: supplying a QComply token turns on reporting to your account, and crypto-view update fetches a new release. Without a token the tool contacts nothing.

Can I scan a private repository

Yes, with the command-line tool: run crypto-view . in your checkout and nothing leaves the machine unless you supply a QComply token, which is what turns reporting on. The website accepts public repositories only, because it fetches them without credentials.

How often is a repository re-scanned

Whenever someone asks. Every report has a Run a fresh scan button that scans the latest commit of the same branch with the current engine; the new report joins the repository's history.

Can we be certified as quantum-ready

QComply issues a quantum-readiness certification for a codebase you own: an independent scan, a reviewed CBOM and a label you can show customers and auditors. Write to

A scan on its own is not that. It reports what the source contains; certification is a reviewed judgement about a system.

Why does the scan appear in a public history

So a repository scanned twice can be compared over time. Every entry concerns a public repository at a public commit. The scan adds an assessment, it does not disclose anything.

The threat

What does quantum-safe actually mean here

One specific thing: no known quantum algorithm brings that primitive below its stated strength. It does not mean the code using it is correct, the key is well managed, or the implementation is sound.

When does this actually happen

Nobody can give a date. What is fixed is this. NIST published the replacements in August 2024 as FIPS 203, FIPS 204 and FIPS 205, and added HQC in March 2025. US federal guidance uses 2030 and 2035 as planning dates. Those are recommendations, not legal deadlines.

The date that binds you is earlier, for the reason below.

Why is the deadline already past for some data

harvest-now-decrypt-later. An attacker who records encrypted traffic today does not need a quantum computer today. They need one before the data stops being sensitive. If a payload must stay confidential for ten years and a capable machine arrives in twelve, you are fine. If it must stay confidential for twenty, the traffic you send this afternoon is already exposed.

That is why key establishment is ranked above signatures throughout this tool, and why the report counts it separately.

Why is a signature finding treated differently from a key-exchange finding

A broken signature scheme lets an attacker forge signatures from the moment they have the machine. It does not work backwards. A signature verified in 2026 was genuine in 2026.

A broken key exchange is retroactive. Traffic recorded in 2026 becomes readable the moment the exchange falls, without touching your systems again. That asymmetry decides what to do first.

Is symmetric encryption affected

Much less than public-key cryptography. Grover's algorithm halves the effective strength of a symmetric key, so AES-128 weakens while AES-256 stays comfortable. Hash functions behave the same way. SHA-256 keeps a workable margin, and SHA-384 restores the full one.

Would RSA-4096 buy us time

No. Doubling the key costs performance today and buys nothing against Shor's algorithm. The move is to ML-KEM and ML-DSA, usually alongside the classical algorithm rather than instead of it.

Reading the report

What is a CBOM, and how is it not an SBOM

An SBOM lists the components software is built from. A CBOM lists the cryptography it uses: which algorithms, with which parameters, in which files. An SBOM says you depend on BouncyCastle. A CBOM says you call it with RSA in eleven places.

What happened to the readiness score

It was withdrawn in 5.10.0. It was a number out of 100, produced by a weighted deduction table, and a reader could not derive it from the findings in front of them. It also read as a grade in front of a client, which is more than a source scan can support: what the report can say is which algorithms are present, where, and which of them protect traffic that is being recorded today.

Everything else is unchanged. The inventory, the postures, the key establishment count and the CBOM are all still here, and the API and SARIF output carry the same counts they always did, minus that one field.

Why does it say "could not be determined"

The algorithm could not be established from the source. Usually it is chosen while the program runs: the argument to getInstance() is a variable, or the key type comes from configuration. It may also be decided somewhere this scan does not reach. A guess would put a wrong entry in the inventory, so the gap is reported instead.

Does renaming an import hide a finding

No. Import aliases are resolved before matching, so from lib import encrypt as enc followed by enc(...) is found, and the finding is marked as reached through a renamed import. Python, JavaScript, TypeScript and Go aliases are covered, along with Java's static imports. See the documentation.

Why is an import listed separately from the code that uses it

Because they are different facts. Importing a library shows what a file can reach. It is not evidence that anything is called. Imports are listed on their own, and are counted apart from the locations where cryptography is actually used.

Why are my tests not in the report

Since 5.14, test, fixture and example files are not scanned. A test that generates a 1024-bit RSA key to check an error message is not a decision anyone has to make, and in most repositories those files outnumbered the production findings. The report says how many test files were left out. The command-line tool scans them with --include-tests.

Comments are treated the same way: an algorithm named only in a comment or in documentation is not a use of it. --include-comments reports those too, marked as such.

Why does one algorithm appear more than once in the CBOM

Because it is used for different things. Each component is one algorithm and one set of operations, so RSA used to generate keys, RSA used to sign and RSA used to decrypt are three components, and every location in each one does exactly what its cryptoFunctions say. That is standard CycloneDX: an asset is identified by its properties, and what it is used for is one of them.

This finding is wrong. What do I do

If a rule matched something that is not cryptography, that is a bug. Please send us the line at contact@qcomply.tech.

If the match is right but not a problem in your case, record the decision in the code:

# crypto-view: ignore[py.cryptography.rsa] fixture key, never used in production

Suppressed findings are still counted and shown in the summary, so the number is never quietly reduced.

The scan found nothing. Are we safe

Nothing was found in the files analysed. Cryptography reached through a compiled dependency, a container image, a managed service or another team's API is outside what a source scan can see.

Migrating

What should we do first

In this order, and the report is arranged to support it:

  1. Anything already broken. MD5, SHA-1, DES, RC4, PKCS#1 v1.5 encryption. These fail today and have nothing to do with quantum computers.
  2. Key establishment protecting long-lived data. This is the harvest-now-decrypt-later exposure. It is the part where waiting costs you something you cannot get back.
  3. Certificates and keys with long lives. Not because they are more broken, but because reissuing them is slow. Shortening certificate lifetimes now makes the eventual change routine.
  4. Signatures. Quantum-vulnerable too, but the exposure begins when the attacker has the machine, not before.
  5. AES-128 and SHA-256 in long-lived roles. A parameter change, scheduled rather than urgent.

For key exchange the usual move is hybrid: run the classical algorithm and ML-KEM together, so the result is no weaker than either one. TLS 1.3 with X25519MLKEM768 and OpenSSH's sntrup761x25519 both do this today.

How does this relate to FIPS 203, 204 and 205

They are the NIST standards for the replacements, all published in August 2024. FIPS 203 is ML-KEM for key establishment. FIPS 204 is ML-DSA for signing. FIPS 205 is SLH-DSA for cases that need a hash-based signature. Every remediation here points at one of them.

Can we get help with the migration itself

Yes. QComply works on cryptographic inventory and migration planning beyond what a scanner can do on its own. See enterprise, or write to contact@qcomply.tech.

Something is not answered here

Write to contact@qcomply.tech.