StackLift
All playbooks

Playbook

Security Review Checklist for Software Projects

A working checklist for reviewing a web or SaaS application before release. Each row names the risk, the control, who owns it and what evidence shows it is done.

Best reader

CTOs, engineering leads and founders preparing a release or responding to a customer security questionnaire

Outcome

A completed review with findings sorted into fix now, schedule and accept, each with an owner.

Use this sequence

1

Draw the system and its data flows, and list what an attacker would want.

2

Threat-model the riskiest workflows: sign-in, payments, file upload, admin and integrations.

3

Check authentication, authorization and tenant isolation against that model.

4

Review secrets, dependencies, configuration and infrastructure access.

5

Test logging, alerting and the response path.

6

Record findings, fix the blocking ones and set a recheck date.

Review the design before the code

Most serious flaws are design flaws. A short threat-modeling session finds them while they are cheap to change.

Data flow diagram

Trust boundaries

Abuse cases for each critical workflow

Use a recognized baseline

The OWASP Top 10 and the Application Security Verification Standard give you a shared vocabulary. Pick a version and a level and write them down.

Pin the standard version

Map each requirement to an owner

Keep evidence with the checklist

Decide before you start what blocks release

Agree the release rule in advance so findings lead to decisions, not debate.

Blocking: data exposure, broken access control

Scheduled: hardening gaps with compensating controls

Accepted: named risk, named approver, review date

Start in 60 seconds

Turn this checklist into a scoped plan.

Answer a few questions and get an AI-assisted, architect-reviewed scope, cost range, and timeline for your software project.

Frequently asked questions

Is a security review the same as a penetration test?

No. A review checks design, code, configuration and process against a checklist. A penetration test tries to break the running system. Many teams do the review first and test afterward.

How often should we repeat it?

Before major releases, after significant architecture changes and on a regular schedule. Add a recheck whenever a new integration or data type is introduced.

The complete guide

Security Review Checklist for Software Projects: the complete guide

Updated November 18, 2026 ยท Published by , Founder, StackLift AI. Written with AI assistance; how we make content.

Short answer

A security review checklist lets you check a software project against a fixed list before release. It covers the design and threat model, authentication and access control, data protection, secrets, dependencies, infrastructure, logging and the response plan. Each item needs an owner, evidence that it is done, and a release decision.

What is a security review, and what is it not?

A security review is a structured look at a system's design, code, configuration and operations to find weaknesses before attackers do. It is a human process guided by a checklist. It produces findings, owners and decisions.

It is not a penetration test, which tries to break the running system, and it is not a compliance audit, which checks you against a named standard. Many teams use all three at different times. No checklist proves a system is secure; it shows that you looked at the right things and wrote down what you found. This guide describes common practice and is not a certification of anything.

What are the steps in a security review?

  1. Scope the system. Output: a diagram of components, data stores, users and external services. Involve the lead engineer and product owner.
  2. Threat-model the critical workflows. Output: a list of threats and the controls that address them. Involve engineering and a security reviewer.
  3. Check identity and access. Output: findings on authentication, authorization and tenant isolation. Involve the security reviewer.
  4. Review data, secrets and dependencies. Output: findings on storage, encryption, key handling and known vulnerable components. Involve engineering and operations.
  5. Review infrastructure and delivery. Output: findings on cloud settings, network access and the build pipeline. Involve DevOps.
  6. Triage and decide. Output: each finding marked fix now, schedule or accept, with an owner and a date. Involve the CTO or founder.

The sections below give you checks for each step. For the wider launch picture, pair this with the production readiness checklist.

How do you threat-model before reviewing code?

Threat modeling asks four plain questions: what are we building, what can go wrong, what are we doing about it, and did we do a good enough job. Start with a rough diagram of how data moves, mark the trust boundaries where data crosses from one level of trust to another, and list what an attacker would want.

Workflows to model first
WorkflowWhat could go wrongQuestion to ask
Sign-in and recoveryStolen or guessed credentials, account takeover through resetCan someone reach an account without the owner?
PaymentsTampered amounts, replayed requests, leaked card dataDoes the server decide the price, not the client?
File uploadMalicious files, oversized uploads, exposed storageWhat is checked and where do files land?
Admin actionsPrivilege abuse, missing audit trailWho can do this and is it logged?
Integrations and webhooksForged callbacks, leaked tokensHow is each caller verified?
AI featuresPrompt injection, data leakage between usersWhat can the model read and do on a user's behalf?

Repeat the session when the design changes or you add a new integration. A fifteen-minute update is better than a one-off workshop that goes stale.

Which standard should the checklist follow?

Do not invent a list from scratch. The OWASP Top 10 describes the most common web application risks in general terms. The OWASP Application Security Verification Standard, known as ASVS, is a more detailed set of requirements organized by area and by verification level. Teams often use the ASVS as the base for a checklist and add their own items.

Choosing a baseline
OptionGood forWatch out for
OWASP Top 10Awareness and a quick first passIt is a risk list, not a full requirement set
OWASP ASVSA detailed, testable checklist by areaPin the version and level so results are comparable
Customer questionnaireAnswering a specific buyerIt reflects their concerns, not necessarily yours
Regulation or contractMeeting a stated obligationCheck with counsel what applies to you

Whatever you choose, write down the version, the level and the date. When someone asks later what you checked, you can answer exactly.

What should you check on access, data and secrets?

Broken access control is among the most common serious findings, so spend real time here. Test with accounts of different roles and different customers, not just with an admin. Then look at how data and keys are stored.

  • Authentication uses a proven provider or library, with protection against repeated guessing and a safe reset flow. See the Auth0 and Clerk integration guide and the SSO guide.
  • Authorization is enforced on the server for every request, not hidden in the interface.
  • One customer cannot read or change another customer's records, tested by changing identifiers in requests.
  • Data is encrypted in transit and at rest, and backups are encrypted with limited access.
  • Secrets live in managed configuration, never in the repository, logs or client code, and they can be rotated.
  • Personal data is collected only when needed and can be deleted or exported on request.
  • Admin and support tools are logged and limited to the people who need them.

Agents and automation need the same discipline. At StackLift, agents work on feature branches and never touch production secrets. Apply an equivalent rule to your own tools.

How do you review dependencies, infrastructure and the pipeline?

Modern applications are mostly other people's code. Scan what you depend on, and review how your software gets built and deployed, because the pipeline is part of the attack surface.

Dependency and infrastructure checks
AreaCheckEvidence
DependenciesScanned for known vulnerabilities; serious ones fixed or mitigatedScan report with dates
LicensesOpen source licenses reviewed against your useDependency inventory
Containers and imagesMinimal, patched and from trusted sourcesImage scan results
Cloud configurationNo public storage by accident, least-privilege roles, separate environmentsConfiguration review notes
NetworkOnly needed ports exposed; databases not publicNetwork diagram
PipelineBranch protection, reviewed changes, restricted deploy credentialsPipeline settings
BackupsEncrypted, separate from production, restore testedRestore test record

The CI/CD glossary entry and our delivery operations page describe how review and release controls fit together.

What belongs in logging, alerting and incident response?

A review should confirm you would notice an attack and know what to do. Logs should record security events such as sign-ins, failed attempts, permission changes and admin actions, without storing secrets or more personal data than needed. Alerts should reach a person who can act.

  • Security events are logged with time, actor and target, and logs are protected from tampering.
  • Alerts exist for repeated failures, unusual admin activity and error spikes.
  • A named incident owner and a one-page response plan exist: contain, assess, notify, recover, review.
  • You know who to notify if personal data is exposed and by when your obligations require it; check this with counsel.
  • A contact route exists for people who find a vulnerability.

See the observability glossary entry for background.

How do you record findings and decide whether to release?

Agree the release rule before you start. Then every finding leads to one of three outcomes. Use one row per control: the risk, the owner, the required state, the evidence, the result and the recheck trigger.

Finding triage
OutcomeTypical examplesNeeds
Fix before releaseExposed customer data, broken access control, secrets in the repositoryOwner and a verified fix
ScheduleMissing hardening, outdated but unexploited dependencyOwner, date and interim control
AcceptLow impact, high cost to fixNamed approver, written reason, review date

Keep the checklist with the release record so the next reviewer starts from it. If you want an independent view of an existing codebase, StackLift's Production Rescue Sprint audits and hardens AI-coded prototypes; see vibe-coded app rescue. You can also scope work with the free estimate or contact us.

Questions people ask

How long does a security review take?

It depends on the size of the system and the depth you choose. A focused review of a small application is short, while a large platform with many integrations takes longer. Scope it by the critical workflows first.

Do we need a penetration test as well?

Often yes for products that hold sensitive data or face the public. A review finds design and process gaps; a test shows what an attacker can do to the running system. Many teams do the review first.

What is the OWASP ASVS?

The Application Security Verification Standard is an OWASP project listing security requirements for web applications, grouped by area and by level of rigor. Teams use it as a base for review checklists.

Who should run the review?

Someone who did not write the code, with security experience and access to the architecture. Internal reviewers work for early checks; an outside reviewer adds independence before important releases.

How should we handle AI-generated code?

Review it as you would any code, with extra attention to missing validation, hard-coded secrets and absent tests. AI-built prototypes often work in a demo but lack these basics.

How often should we repeat the review?

Before major releases, after significant architecture changes, when you add integrations or data types, and on a regular schedule. Each repeat should start from the last checklist.

Sources

Sources and further reading

The standards, rules, research and official documentation this page draws on.