Skip to main content

What are you looking for?

Explore our services and discover how we can help you achieve your goals

Client Workspace

Secure development lifecycle: threat modelling, SAST, DAST and dependency checks in CI

A secure development lifecycle puts one security control at each stage of delivery: threat modelling at design, static analysis and dependency checks on every change, dynamic testing against a running staging build, and a release gate that reads all of them. Netbase sequences these controls this way, because each one catches what the others miss.

Book a solution review See the related solution

By Netbase Delivery and Quality Desk · Reviewed by David (CEO) · Updated · 11 min read

star

This guide is for CTOs, engineering managers and product owners who already run a CI pipeline and want to know which security check belongs where, what it can and cannot find, and which findings should stop a merge. If you want the testing and review done for you, see application security assurance.

In this guide

What does a secure development lifecycle add to a normal CI pipeline?

A secure development lifecycle adds security decisions to stages that already exist: design reviews get a threat model, pull requests get static analysis and dependency checks, staging gets a dynamic scan, and the release decision gets a written rule for open findings. It does not add a separate security phase at the end.

The NIST Secure Software Development Framework makes the same point: most delivery models do not say much about security, so the practices are meant to be added to whatever lifecycle a team already uses. A mid-market team needs no new process, only four or five checks placed where the information to act on them already exists.

The placement is the real decision. A control that runs too early has nothing to inspect; a control that runs too late finds problems when fixing means reopening finished work. A dynamic scan run only before a yearly audit finds issues live for months; after release, they belong in one patch, scan and recovery cadence.

Which control runs at which stage?

Each control looks at a different thing, so each has a natural place in the pipeline and a set of problems it cannot see.

Stage and control What it looks at What it misses Blocks
Design: threat model Data flows, trust boundaries, roles and abuse cases Coding mistakes; anything not drawn on the diagram The design, until each high threat has an owner and a planned control
Pull request: static analysis (SAST) Source code patterns such as injection and unsafe calls Access-control, authentication and configuration problems The merge, for new high-severity findings in changed code
Pull request and nightly: dependency checks Third-party libraries against publicly disclosed vulnerabilities Flaws in your own code; libraries that are risky but have no published entry The merge, for a new dependency with a known critical issue
Staging: dynamic testing (DAST) The running application from outside, through HTTP Code paths the scanner cannot reach; business-logic and multi-tenant flaws The release, for confirmed high findings on changed features
Release: release gate Open findings from all of the above, with owners and dates Nothing new; it only enforces the rules The release, when a rule is broken without a recorded exception

Read the "misses" column before buying tools. The OWASP page on source code analysis tools lists authentication problems, access-control issues and configuration problems among the things static analysis struggles to find, and warns about high numbers of false positives. Those gaps are exactly where business applications tend to fail, which is why the threat model and human review stay in the lifecycle even when scanning is fully automated.

How do you add the controls in order?

Add one control at a time and make each one trusted before the next. A pipeline that fails on every run teaches the team to ignore it.

  1. Start with dependency checks

    They are cheap to run, their findings are concrete, and every library an application pulls in is code the team did not review. A tool such as OWASP Dependency-Check identifies project dependencies and checks them for known, publicly disclosed vulnerabilities. Block only on new critical findings at first.

  2. Add static analysis on changed code only

    Scan the pull request diff and block on new high-severity findings. Leave the existing backlog as a tracked list rather than a gate, or the first run will stop all work.

  3. Write a threat model for the next significant feature

    Use the four questions of the OWASP threat modelling cheat sheet: what are we working on, what can go wrong, what are we going to do about it, and did we do a good enough job. Keep the output to one page with owners.

  4. Add a dynamic scan against staging

    Run an authenticated scan after each deploy to staging, so the scanner reaches screens behind the login. Start in report-only mode until the team has tuned out the noise.

  5. Write the release rule

    One short policy: which severities block a release, who may sign an exception, and how long an exception lasts. Without the rule, scan results become a report nobody reads.

  6. Review the misses every quarter

    Compare what reached production, including what a penetration test found, with what the pipeline caught, and move a control earlier where a class of issue keeps slipping through.

What should block a merge, and what should only be reported?

A merge should be blocked only by findings that are new, in code the change touched, confirmed or highly likely to be real, and above an agreed severity. Everything else belongs on a tracked list with an owner and a date.

Static analysis and dynamic scanners both produce false positives, and a gate that blocks on unconfirmed or inherited findings gets bypassed within weeks. A narrow gate that is always right changes behaviour; a wide gate that is often wrong does not.

Report-only findings still need a home. Put them in the same backlog as other defects, with severity adjusted for what the affected feature does: a medium finding on a payment or permissions screen may outrank a high finding on an internal report page. The release gate then checks that nothing above the line is open without a recorded, time-limited exception.

Secrets are the exception to the narrow rule. A credential committed to a repository should block the merge every time, and the fix is to rotate the credential, not only to delete the line, because the old value stays in the history.

When is a full secure development lifecycle the wrong next step?

A full lifecycle is the wrong next step when the basics underneath it are missing. In three situations, a smaller move gives more security for the same effort.

  • No reliable CI yet. If builds are manual or tests rarely pass, security gates have nothing stable to attach to. Fix the pipeline first; a DevOps and platform engineering review is the better spend.
  • No access-control design. If roles and permissions were never designed, scanners will report symptoms of one missing design. Design the roles, then scan.
  • A short-lived prototype. For a throwaway proof of concept, a dependency check and a secrets check are enough. Add the rest when the product is going to hold customer data.

Where AI fits

AI helps most with the volume around these controls, not with the decision. A model can draft a first threat list from an architecture description, group duplicate findings from several scanners, explain a static-analysis finding in plain language and draft a fix and a regression test. Engineers still decide whether a finding is real, set its severity, review every generated fix and own the threat model, because a confident wrong answer in security costs more than a slow right one.

AI-assisted code also needs the same pipeline as any other code. The security review gate for AI-generated code adds dependency verification and named approval on top of the stages described here. Netbase works with the major commercial and open-source AI tools and models, chosen per project, and runs AI-assisted engineering under human review.

What delivery record exists, and what does not?

Netbase's published security practices are secure code review and version control, TLS in transit and AES at rest, role-based access control, MFA for admin dashboards, vulnerability scanning and penetration testing, and disaster recovery; NDAs, data processing agreements and SLAs are available on request. Security and API design from the start is part of how Netbase delivers its own projects.

Netbase holds ISO 27001 certification and a SOC 2 Type II attestation for its own operations. These describe how Netbase works; they do not certify a client's application or pipeline, and they are not scan results.

Delivery follows a six-step lifecycle from discovery and architecture planning to ongoing support, with weekly reviews and a named project manager, which is where the threat model, the gate rules and open findings are reviewed. Netbase also works as offshore development partner on a multi-tenant cloud ERP for a US client, a SaaS product where tenant isolation is the first abuse case a threat model has to cover.

Most Netbase projects are delivered on fixed-price contracts. No Netbase record publishes scan results, finding counts, false-positive rates, fix times or tool comparisons, and none is offered here as evidence that a particular control works.

Alternatives: who runs the controls, and how to choose

The tools are rarely the hard part; ownership is. There are three common set-ups.

  • The product team owns everything. Best when the team is stable and has time in each sprint for findings. Risk: security work loses to feature work without a written release rule.
  • A central platform or security function provides the pipeline. Best for organisations with several products; the teams inherit the same gates. Risk: gates tuned for one product produce noise on another.
  • A development or assurance partner runs the controls with the team. Best when there is no in-house security skill, or for an outside view before an audit or customer review. Ask how findings are triaged, who fixes them and how fixes are proven. See quality engineering and testing for testing that runs alongside each release.

Choose by three criteria: who can tell a real finding from a false one fastest, who can change the pipeline without breaking delivery, and who will still own the gate rules a year from now.

Limits of this guide

  • It covers application code and its dependencies in a CI pipeline, not network, endpoint or cloud-account security.
  • It names tool categories and one open-source example, not a tool ranking; the right tool depends on languages and hosting.
  • It is not compliance advice; an auditor or the standard you report against decides what evidence is required.

Plan the next step with a Netbase consultant

Frequently asked questions

Static application security testing reads source code without running it and points to the file and line of a suspected flaw. Dynamic application security testing attacks the running application from outside through HTTP. They find different problems, so most teams run static analysis on each pull request and dynamic scans against staging.

Yes, but only for new findings above an agreed severity in code the change touched, plus any committed secret. Blocking on inherited or unconfirmed findings makes the gate noisy and teaches the team to bypass it.

Threat modelling should happen during design, before the code exists, and again when a feature changes data flows, roles or trust boundaries. A threat model written after release only describes risks it is now expensive to fix.

Most small teams should add dependency checks first, because they are cheap to run and their findings are concrete. Static analysis on changed code comes next, then a threat model for the next significant feature and a dynamic scan against staging.

Next step

If you want the controls in your pipeline placed, tuned and owned by the team that will fix the findings, book a solution review.

Application security assurance for web, SaaS and AI features Application security assurance for web, SaaS and AI features

Netbase provides application security testing for product and engineering leaders to find and fix vulnerabilities before release, including in AI features. We review code with AI-assisted triage, scan dependencies and test running applications, then rank findings and retest fixes. Netbase's own alignment with ISO 27001 and SOC 2 Type II covers Netbase's operations, never your application.

Learn More
line
AI-assisted QA and software testing that run inside delivery AI-assisted QA and software testing that run inside delivery

Netbase provides software testing and QA services for product and commerce teams to release often without breaking what customers use. QA runs inside delivery, not after it: acceptance criteria, automated regression with AI-generated test drafts, performance budgets and release readiness checks. Page speed shows the result: USticker's page load fell 21%, and PrintLeo's improved 35%.

Learn More
line
AI-assisted DevOps and platform engineering for routine releases AI-assisted DevOps and platform engineering for routine releases

Netbase provides DevOps and platform engineering for SaaS and commerce teams to make releases routine instead of risky. We set up CI/CD pipelines, repeatable environments and observability on AWS, Google Cloud, DigitalOcean or Cloudflare, with AI assistance for tests, reviews and incident triage, so every change ships the same way and problems surface before customers notice.

Learn More
line
SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules SaaS product accelerator: launch an AI-ready SaaS on proven Netbase modules

A SaaS product accelerator is a set of reusable Netbase modules for accounts, billing, roles and integrations that helps founders and product teams launch subscription software faster, with room for in-product AI from the first release. Reusing these modules can shorten development time, and the approach is proven on Printcart, Netbase's own web-to-print SaaS.

Learn More
line

Work with Netbase remotely

Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.

Contact Netbase

Discuss a project

Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.
Project enquiries

project@netbasejsc.com

WhatsApp

+84 937 869 689

Office address

91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam

Get in touch

Tell us what you want to build, modernize, or operate.

Tell us what you want to build, modernize, or operate.

Contact Netbase