Written by: Mio Kanzaki(Security Analyst (AI))Reviewed by: 千葉圭介Updated
A source code assessment reads the written code itself, without running the program, to find ways of writing that lead to holes. It is meant for companies that hold the code of a web application or business system they built, or one that a contractor delivered. First check roughly how many lines you want assessed and which languages the code uses.
Cyber Shield combines static analysis (checking code automatically without running it) with a person's review. The price is quoted individually, and the guide is from JPY 50,000 per 10,000 lines (excl. tax). Below we cover what we check, how, what you keep and what it costs.
What does a source code assessment check?
An assessment that tries the screens from outside is like testing the locks on a finished building. A source code assessment is like reading the blueprint to find places where a lock may have been forgotten. It finds holes at the stage of writing, which are hard to see even when you run the system.
Our viewpoints rest on public documents. IPA's Secure Programming Course explains that ordinary programming techniques alone can introduce vulnerabilities unintentionally.
MITRE's CWE is another guide. It is a list that gives numbers to categories of software weakness. The 2025 Top 25 ranks the weaknesses that are seen often and have a large impact, drawn from just over 39,000 records of published vulnerabilities.
The table sets out our main viewpoints in everyday language. We do not publish attack procedures or the strings used in attacks.
| Viewpoint | In plain words | What can happen |
|---|---|---|
| Handling of input | Is a value typed by a user used as it is in a command to the database or the operating system? | Database contents are stolen, or the server runs unwanted processing |
| Authentication and permission checks | Do the login check and the "may this person see this?" decision cover every path? | Someone without rights reaches another person's data or an admin function |
| Where secrets are kept | Are passwords or connection keys written directly into the code? | If the code leaks, outside services can be entered too |
| Use of encryption | Is important information protected by an old method or a weak setting? | Stored data or the contents of traffic are read |
| Outside parts | Does the code keep using an old library with a published weakness? | The weakness in the part is used as it is |
| Restoring data | Is data that arrived from outside turned back into program objects without checks? | Unexpected processing is run |
| Error handling and records | Does a failure show internal information on screen, and can odd events be recorded? | The inner workings become visible, or intrusion is noticed late |
For individual flaws, see SQL injection, insecure deserialization and vulnerable dependencies.
Supported languages and what is out of scope
We support PHP (including Laravel), Node, Python, Ruby, Java and .NET. If your code uses another language, or a framework that is no longer supported, please ask us first whether we can assess it.
Three things fall outside this assessment. Checking behavior while the system runs is the web application assessment. A check limited to the doors used for connections is the API assessment. Testing whether an attacker can actually get in is the penetration test.
How do static analysis and a person's review divide the work?
The assessment moves through three steps: static analysis by tools, a check of the libraries you use, and a person's review. We do not stop at the tools because some holes cannot be judged without reading what the code means.
Tools are good at checking a large amount of code against one standard without gaps. We use the static analysis tool Semgrep together with deeper AI-assisted analysis to surface candidates for risky patterns. We also check whether the libraries you use have published weaknesses.
Tools have a weak side too. A question such as "can this processing be called by another route, skipping the permission check?" cannot be answered without reading the flow of the code and the business meaning. So an analyst reads each candidate and judges whether it can really be abused. False positives, which are findings flagged as holes that are not holes, are removed at this step.
The OWASP Code Review Guide gathers ways of reading code from a security viewpoint. The analyst uses it, together with IPA's documents, to decide the reading order and where to look.
Static analysis has limits. Server settings and values that are decided only when the program runs cannot be seen by reading code. If you have screens or APIs that are not yet public, we suggest pairing this assessment with a dynamic one.
What do you receive when the assessment ends?
You receive a report. For each hole it states which part of the code is affected, why it is a problem and how to fix it. The analyst confirms that the hole can be abused, and the report ranks what to fix first.
After you fix the code, we run one re-assessment at no charge. It confirms that the code you changed no longer has the problem, and the outcome is added to the report. The fixes themselves are charged separately, from JPY 10,000 per item (excl. tax).
What does it cost, and what makes it higher?
The amount of effort changes a great deal with the volume and structure of the code, so the assessment is quoted individually. All amounts exclude tax.
| Item | Price | Notes |
|---|---|---|
| Source code assessment | quoted individually (guide: from JPY 50,000 per 10,000 lines (excl. tax)) | Set after we see the code |
| Individual vulnerability fix | from JPY 10,000 per item (excl. tax) | Quoted by content |
| Re-assessment after fixes | Free | Once |
An example shows how to read the guide. For 30,000 lines of code, the starting point is from JPY 150,000 (excl. tax). The actual amount is decided after we see what the code contains.
Three things raise the amount. The first is the number of lines, because more lines mean more to read. The second is the number of languages and frameworks, because each kind adds viewpoints and tools. The third is the scope. Limiting it to code you wrote yourself reduces the effort, and including parts brought in from outside increases it.
Start with the free consultation and tell us the size of the code and the languages. The formal quotation follows the free assessment. The pricing page shows all plans side by side. To see what is visible from outside first, you can also use the free URL check.
How does a request go from first contact to delivery?
The general flow applies here. The list notes what is particular to a code assessment.
- In the free consultation we ask about the target system, the languages and roughly how many lines there are.
- We sign a non-disclosure agreement at no cost. Code is information you do not want to leave your hands, so we also agree at this stage how it will be handed over.
- With your signed consent form, we run the free assessment (tools plus an analyst's check) and return an issue report.
- The fix cost for the issues found is quoted. If your code needs a full rewrite, that gets a separate quotation.
- After ordering, we go through every item and fix the holes.
- Before delivery we send a PDF of the changes made and the reasons.
- You try the application's behavior on the actual screens.
- We hand over the result, and support stays free for one month after delivery.
If you want to keep watching after the assessment, maintenance and advisory plans and our AI monitoring service, Guard, are available.
Common objections
A web application assessment is done, so is reading the code unnecessary?
An assessment from outside matters. Holes in processing that is hard to reach from outside, or that runs only under certain conditions, can be hard to find that way. Reading the code directly lets us confirm those places too.
A contractor built our system and we do not know what is in the code
Code delivered by a contractor can be assessed even when another company wrote it. In the free consultation we first ask whether we can receive the code and whether we support the language.
The check of the libraries you use is part of the assessment. It can help you find parts that nobody has updated.
We are uneasy about handing code to an outside party
That concern is natural. So the NDA is signed before the assessment, and the target and period are put in writing in the consent form. We agree the way the code is handed over with you, in line with your internal rules.
If the code truly cannot leave your hands, one option is to limit the assessment to some of the functions.
How is this different from running a static analysis tool ourselves?
Running a tool yourselves is a good habit. Tool output, though, mixes false positives with holes the tool cannot read, and the more findings there are, the harder it is to decide which to fix first.
Our analysts read the candidates, pick those that can be abused and give you a report ranked by priority. We do not hand over raw tool output.
Frequently asked questions
What is a source code assessment?
It is a check that reads the written code itself, without running the program, for ways of writing that lead to holes. Static analysis tools scan broadly, and an analyst follows the flow of the code to judge which findings can really be abused. The results go into a report.
How much does it cost?
It is quoted individually. The guide is from JPY 50,000 per 10,000 lines (excl. tax), and the amount changes with the number of lines, the number of languages and the scope. An individual vulnerability fix is from JPY 10,000 per item (excl. tax), quoted by content, and one re-assessment after the fixes is free.
Which languages do you support?
We support PHP (including Laravel), Node, Python, Ruby, Java and .NET. For other languages, or code written on an old framework, please ask us first whether we can assess it.
If we already had a web application assessment, do we still need a source code assessment?
Their roles differ. A web application assessment tries the running screens from outside to look for holes. A source code assessment confirms from the inside the processing that is hard to reach from a screen and unsafe coding patterns that never show on a screen. Combining them leaves fewer gaps.
Does an assessment prove the system is safe?
No. An assessment tells you which holes the code has at that moment. A code change or a parts update can bring new holes, so check again after a large revision.
Next step
If you want to know how your public site looks from outside before the assessment, use the free URL check. It looks only at what a visitor can see, runs no attacks and cannot read your code.
If the amount of code, the languages or your relationship with a contractor are unclear, contact us by form, phone or online meeting. The NDA costs nothing and is in place before we look at any code.
Written by Mio Kanzaki (AI security analyst). Supervised by 千葉圭介. Updated 19 November 2026.
Frequently asked questions
What is a source code assessment?
It is a check that reads the written code itself, without running the program, for ways of writing that lead to holes. Static analysis tools scan broadly, and an analyst follows the flow of the code to judge which findings can really be abused. The results go into a report.
How much does it cost?
It is quoted individually. The guide is from JPY 50,000 per 10,000 lines (excl. tax), and the amount changes with the number of lines, the number of languages and the scope. An individual vulnerability fix is from JPY 10,000 per item (excl. tax), quoted by content, and one re-assessment after the fixes is free.
Which languages do you support?
We support PHP (including Laravel), Node, Python, Ruby, Java and .NET. For other languages, or code written on an old framework, please ask us first whether we can assess it.
If we already had a web application assessment, do we still need a source code assessment?
Their roles differ. A web application assessment tries the running screens from outside to look for holes. A source code assessment confirms from the inside the processing that is hard to reach from a screen and unsafe coding patterns that never show on a screen. Combining them leaves fewer gaps.
Does an assessment prove the system is safe?
No. An assessment tells you which holes the code has at that moment. A code change or a parts update can bring new holes, so check again after a large revision.