[{"data":1,"prerenderedAt":829},["ShallowReactive",2],{"/en-us/blog/our-step-by-step-guide-to-evaluating-runtime-security-tools":3,"navigation-en-us":40,"banner-en-us":450,"footer-en-us":460,"blog-post-authors-en-us-Hiroki Suezawa|Mitra Jozenazemian":700,"blog-related-posts-en-us-our-step-by-step-guide-to-evaluating-runtime-security-tools":726,"blog-promotions-en-us":767,"next-steps-en-us":819},{"id":4,"title":5,"authorSlugs":6,"body":9,"categorySlug":10,"config":11,"content":15,"description":9,"extension":27,"isFeatured":13,"meta":28,"navigation":29,"path":30,"publishedDate":22,"seo":31,"stem":35,"tagSlugs":36,"__hash__":39},"blogPosts/en-us/blog/our-step-by-step-guide-to-evaluating-runtime-security-tools.yml","Our Step By Step Guide To Evaluating Runtime Security Tools",[7,8],"hiroki-suezawa","mitra-jozenazemian",null,"security",{"slug":12,"featured":13,"template":14},"our-step-by-step-guide-to-evaluating-runtime-security-tools",false,"BlogPost",{"title":16,"description":17,"authors":18,"heroImage":21,"date":22,"body":23,"category":10,"tags":24},"Our step-by-step guide to evaluating runtime security tools","Key learnings from the GitLab Security team’s runtime security tool evaluation on Kubernetes clusters and Linux servers using real-world attack simulations.",[19,20],"Hiroki Suezawa","Mitra Jozenazemian","https://res.cloudinary.com/about-gitlab-com/image/upload/v1750097534/Blog/Hero%20Images/Blog/Hero%20Images/AdobeStock_1097303277_6gTk7M1DNx0tFuovupVFB1_1750097534344.jpg","2025-05-13","Choosing the right runtime security tool is critical for protecting modern cloud-native environments.  We recently undertook a rigorous evaluation process using real-world attack simulations on our Kubernetes clusters and Linux servers. Why? Because traditional cloud audit logs do not provide enough detail, leaving critical gaps in threat detection, incident response, and forensic analysis. Our evaluation meticulously examined each critical stage from initial access to lateral movement and data exfiltration.\n\nWhile we won't be naming the specific vendor in this post, we want to share our detailed methodology and key learnings, providing a blueprint you can adapt for your own security tool evaluations.\n\n## Why are runtime security tools necessary?\n\nWithout runtime security tools, detecting “suspicious activities” and understanding “what actually happened” during an attack can become extremely challenging.\n\n### Limitations of cloud audit logs\n\n- **Lack of runtime details**  \n  Cloud audit logs primarily record operations and data access within the cloud. However, they do not capture runtime-level activities on systems such as Kubernetes servers – overlooking fine-grained command executions, process behaviors, and transient network activities.  \n\n- **Gaps in investigation and forensics**  \n    In Kubernetes environments, the absence of continuous, real-time logging can lead to the loss of critical activity records once a container terminates.\n\nAlthough well-known open-source runtime security tools are available, we decided to evaluate a commercial product to assess additional capabilities and enterprise-level support through attack simulation testing.\n\n### The role and purpose of runtime security tools\n\nRuntime security tools address these cloud audit log limitations by continuously monitoring systems in real time, offering the following functionalities:\n\n- **Threat detection**  \n  They monitor command executions, system calls, and network events in real-time to instantly detect abnormal behaviors, which enables the security team to respond rapidly. While some public cloud providers now offer limited runtime monitoring capabilities, these native solutions typically lack the depth and comprehensive coverage of dedicated security tools.  \n\n- **Incident response**  \n  By maintaining detailed chronological records of system activities, these tools provide security teams with the evidence needed to reconstruct attack timelines, determine the full scope of compromise, and conduct thorough forensic investigations after an incident occurs.  \n\n- **Scalability in investigations**  \n  Unlike traditional endpoint-by-endpoint forensic analysis, runtime security tools allow teams to collect, store, and analyze data centrally across the entire environment. This enables the efficient investigation of incidents without manually correlating disparate data sources.  \n\n(**Note:** Products that also offer container information or server vulnerability monitoring are outside the scope of this discussion.)\n\n## Key evaluation points\n\nOur primary objective in evaluating a runtime security tool was to determine its effectiveness in real-world security investigations. While evaluations often focus on the volume of detections or overall coverage, in actual operations, an overload of false positives – or tens of alerts for a single attack chain – can paralyze incident response teams. Therefore, our in-depth investigation centered on whether the tool could be used to support security operations with understanding and responding to actual attacks.\n\n- **Detection capability**  \n\n  - **Built-in rule**  \n    We assessed whether the built-in rule sets could effectively detect a variety of attack techniques and provide the necessary detail for accurate detection.\n\n  - **Custom detection capabilities**  \n    We evaluated the ease with which additional rules could be integrated and considered the quality of telemetry data delivered by the product, which enabled us to build our own monitoring solutions leveraging our unique understanding of our environment.\n\n  - **Alert quality**  \n    We also verified the rate of false positives. We confirmed that it effectively focuses on genuine security threats requiring action while minimizing noise that could cause alert fatigue.\n\n- **Incident response**  \n\n  - **Richness of logs**  \n    We evaluated whether the logs capture sufficient details – including executed commands, network connections, DNS queries, and process information – to fully reconstruct the incident. The ability to piece together the entire attack scenario and determine the full impact is crucial during incident response.  \n\n  - **Log searchability**  \n    We assessed how effectively the tool allowed us to search, filter, and correlate events across multiple systems. The ability to quickly query massive volumes of data is essential for timely investigations during security incidents. \n\n## Evaluation process\n\nWe divided our evaluation process into four major phases:\n\n1. **Development of attack scenarios**  \n   We designed scenarios that mimicked real-world attack flows. These scenarios, developed in collaboration with our Red Team, included the following elements:  \n   - attacks exploiting GitLab-specific vulnerabilities (e.g., CVE-2021-22205)  \n   - attacks leveraging the compromise of developer laptops  \n   - detailed step-by-step attack procedures  \n2. **Infrastructure setup**  \n   We deployed two parallel environments:  \n   - Kubernetes environment  \n   - Virtual machine (VM) environment \n\n   We installed an older version of GitLab to test known vulnerabilities and carried out similar evaluation flows in both the Kubernetes and VM environments.\n\n3. **Execution of attacks**  \n   We executed the attack flow for each scenario and meticulously recorded the timeline – from initial access to lateral movement and data exfiltration.  \n\n4. **Analysis of results**  \n   We conducted a comprehensive evaluation of detection capabilities, log richness, and areas for improvement, clearly outlining the strengths and weaknesses of the tools.\n\n### Attack scenarios\n\n**Scenario 1: Exploitation of a known GitLab vulnerability**\n\n![Scenario 1: Exploitation of a known GitLab vulnerability](https://res.cloudinary.com/about-gitlab-com/image/upload/v1750097560/Blog/Content%20Images/Blog/Content%20Images/image1_aHR0cHM6_1750097560795.png)\n\n- **Attack flow**  \n  1. **Initial access**  \n     We simulated an attack by exploiting CVE-2021-22205, a known GitLab vulnerability that allows remote code execution. This granted us unauthorized access to the target system.  \n  2. **Command execution**  \n     After gaining access, we executed a reverse shell to interact remotely with the compromised machine and take control.  \n  3. **Deployment of a C2 agent**  \n     We installed a Command and Control (C2) agent to evaluate persistence techniques, enabling us to execute further commands and manage the system remotely.  \n  4. **Lateral movement**  \n     We then moved laterally within the environment, accessing Kubernetes API secrets and PostgreSQL databases.  \n  5. **Data exfiltration**  \n     We exfiltrated sensitive data via a dedicated C2 channel.\n\nThe following table summarizes the attack techniques used at each phase:\n\n| Initial access | Command and control | Enumeration | Credential access | Lateral movement | Collection | Exfiltration |\n| :---- | :---- | :---- | :---- | :---- | :---- | :---- |\n| Exploit GitLab application using known RCE vulnerability | Execute known reverse shell command | Harvesting info on the box | Get environment variables | Get secret from Kubernetes API | Get data from Cloud Storage | Exfiltration over C2 channel |\n|  | Install post-exploitation C2 agent |  | Get K8s token | Access to database | DNS exfiltration |  |\n|  | SOCKS proxy |  | Get cloud token via Cloud metadata server |  |  |  |\n\n\u003Cbr>\u003C/br>\n\n**Scenario 2: Compromise of a developer’s laptop**\n\n![Scenario 2: Compromise of a developer’s laptop](https://res.cloudinary.com/about-gitlab-com/image/upload/v1750097561/Blog/Content%20Images/Blog/Content%20Images/image2_aHR0cHM6_1750097560796.png)\n\n- **Attack flow**  \n  1. **Initial compromise**   \n     We simulated an attacker compromising a developer’s laptop and abusing legitimate credentials to gain unauthorized access to internal resources.  \n  2. **Privilege escalation**  \n     Using the compromised credentials, we escalated privileges within the Kubernetes environment. \n  3. **Container manipulation**  \n     We deployed a privileged container to extract sensitive information.  \n  4. **Data exfiltration and persistence**  \n     We exfiltrated sensitive data while maintaining persistent access.\n\n      The following table summarizes the attack techniques used at each phase:\n\n| Initial access | Execution | Privilege escalation | Credential access | Lateral movement | Exfiltration |\n| :---- | :---- | :---- | :---- | :---- | :---- |\n| Valid account (kubectl) | Create a new container | Create a privileged container | Get K8s secrets via privilege of the node | Enter a container in the same node | Upload credential data to the attacker’s server |\n|  |  |  | Get an environment variable in the containers via `crictl` command on the node |  |  |\n\n\u003Cbr>\u003C/br>\n\n### Execution of the attacks\nDuring the execution of the attack scenarios, we followed these processes to obtain detailed records:\n\n- **Verification of detections:** We confirmed whether each attack command was detected and if the key points of each scenario were properly flagged.\n\n- **Timeline recording:** Every event was logged in sequence to assess how well command executions and network communications were captured.\n\n- **Scoring and analysis:** We scored each event based on detection effectiveness to quantitatively evaluate the tool’s performance.\n\n## What we learned\n\n### Don't overestimate – test commercial products yourself\n\n- **Identifying and addressing detection gaps (collaboration with vendors)**  \n  Our evaluation revealed that several critical scenarios and events were not detected or not logged. Consequently, we held meetings with the vendor and submitted multiple improvement requests. As a result, the vendor enhanced the product by adding new features and improving detection capabilities, with many issues identified during our evaluation subsequently addressed.  \n- **Understanding the limitations**  \n  Many modern runtime security tools use eBPF to monitor Linux system calls for detection. However, because commands executed within a C2 framework do not generate new processes, tracing these attack events proved challenging.  \n\n- **Recognizing tool boundaries**  \n  Our findings highlighted that, during incident response, relying solely on runtime security tools is insufficient. It is essential to combine them with other logs, such as Kubernetes audit logs and cloud logs, to gain a comprehensive view.\n\n### The importance of continuous runtime event logging in Kubernetes\n\nIn Kubernetes environments, there is a risk of losing forensic data when containers terminate, making continuous logging indispensable. Our evaluation confirmed that establishing a scalable, persistent logging infrastructure is crucial. Without proper runtime security tools, a significant amount of critical information could be lost post-attack.\n\n## Summary\n\nWe do not simply install security tools – we evaluate their utility to help ensure that our customers can safely use GitLab.com. Thorough product assessments like the one outlined above not only reveal unique use cases and areas for improvement that vendors might overlooks, but also provide valuable insights that benefit both the vendor and internal teams in organizing how the tool is best utilized.\n",[10,25,26],"DevSecOps","inside GitLab","yml",{},true,"/en-us/blog/our-step-by-step-guide-to-evaluating-runtime-security-tools",{"title":16,"description":17,"ogTitle":16,"ogDescription":17,"noIndex":13,"ogImage":21,"ogUrl":32,"ogSiteName":33,"ogType":34,"canonicalUrls":32},"https://about.gitlab.com/blog/our-step-by-step-guide-to-evaluating-runtime-security-tools","https://about.gitlab.com","article","en-us/blog/our-step-by-step-guide-to-evaluating-runtime-security-tools",[10,37,38],"devsecops","inside-gitlab","bWvULW1inrCUyD_c0dKZj141uqYV4WHqIeFBypoFm5E",{"data":41},{"logo":42,"freeTrial":47,"sales":52,"login":57,"items":62,"search":370,"minimal":401,"duo":420,"switchNav":429,"pricingDeployment":440},{"config":43},{"href":44,"dataGaName":45,"dataGaLocation":46},"/","gitlab logo","header",{"text":48,"config":49},"Get free trial",{"href":50,"dataGaName":51,"dataGaLocation":46},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com&glm_content=default-saas-trial/","free trial",{"text":53,"config":54},"Talk to sales",{"href":55,"dataGaName":56,"dataGaLocation":46},"/sales/","sales",{"text":58,"config":59},"Sign in",{"href":60,"dataGaName":61,"dataGaLocation":46},"https://gitlab.com/users/sign_in/","sign in",[63,90,185,190,291,351],{"text":64,"config":65,"cards":67},"Platform",{"dataNavLevelOne":66},"platform",[68,74,82],{"title":64,"description":69,"link":70},"The intelligent orchestration platform for DevSecOps",{"text":71,"config":72},"Explore our Platform",{"href":73,"dataGaName":66,"dataGaLocation":46},"/platform/",{"title":75,"description":76,"link":77},"GitLab Duo Agent Platform","Agentic AI for the entire software lifecycle",{"text":78,"config":79},"Meet GitLab Duo",{"href":80,"dataGaName":81,"dataGaLocation":46},"/gitlab-duo-agent-platform/","gitlab duo agent platform",{"title":83,"description":84,"link":85},"Why GitLab","See the top reasons enterprises choose GitLab",{"text":86,"config":87},"Learn more",{"href":88,"dataGaName":89,"dataGaLocation":46},"/why-gitlab/","why gitlab",{"text":91,"left":29,"config":92,"link":94,"lists":98,"footer":167},"Product",{"dataNavLevelOne":93},"solutions",{"text":95,"config":96},"View all Solutions",{"href":97,"dataGaName":93,"dataGaLocation":46},"/solutions/",[99,123,146],{"title":100,"description":101,"link":102,"items":107},"Automation","CI/CD and automation to accelerate deployment",{"config":103},{"icon":104,"href":105,"dataGaName":106,"dataGaLocation":46},"AutomatedCodeAlt","/solutions/delivery-automation/","automated software delivery",[108,112,115,119],{"text":109,"config":110},"CI/CD",{"href":111,"dataGaLocation":46,"dataGaName":109},"/solutions/continuous-integration/",{"text":75,"config":113},{"href":80,"dataGaLocation":46,"dataGaName":114},"gitlab duo agent platform - product menu",{"text":116,"config":117},"Source Code Management",{"href":118,"dataGaLocation":46,"dataGaName":116},"/solutions/source-code-management/",{"text":120,"config":121},"Automated Software Delivery",{"href":105,"dataGaLocation":46,"dataGaName":122},"Automated software delivery",{"title":124,"description":125,"link":126,"items":131},"Security","Deliver code faster without compromising security",{"config":127},{"href":128,"dataGaName":129,"dataGaLocation":46,"icon":130},"/solutions/application-security-testing/","security and compliance","ShieldCheckLight",[132,136,141],{"text":133,"config":134},"Application Security Testing",{"href":128,"dataGaName":135,"dataGaLocation":46},"Application security testing",{"text":137,"config":138},"Software Supply Chain Security",{"href":139,"dataGaLocation":46,"dataGaName":140},"/solutions/supply-chain/","Software supply chain security",{"text":142,"config":143},"Software Compliance",{"href":144,"dataGaName":145,"dataGaLocation":46},"/solutions/software-compliance/","software compliance",{"title":147,"link":148,"items":153},"Measurement",{"config":149},{"icon":150,"href":151,"dataGaName":152,"dataGaLocation":46},"DigitalTransformation","/solutions/visibility-measurement/","visibility and measurement",[154,158,162],{"text":155,"config":156},"Visibility & Measurement",{"href":151,"dataGaLocation":46,"dataGaName":157},"Visibility and Measurement",{"text":159,"config":160},"Value Stream Management",{"href":161,"dataGaLocation":46,"dataGaName":159},"/solutions/value-stream-management/",{"text":163,"config":164},"Analytics & Insights",{"href":165,"dataGaLocation":46,"dataGaName":166},"/solutions/analytics-and-insights/","Analytics and insights",{"title":168,"items":169},"GitLab for",[170,175,180],{"text":171,"config":172},"Enterprise",{"href":173,"dataGaLocation":46,"dataGaName":174},"/enterprise/","enterprise",{"text":176,"config":177},"Small Business",{"href":178,"dataGaLocation":46,"dataGaName":179},"/small-business/","small business",{"text":181,"config":182},"Public Sector",{"href":183,"dataGaLocation":46,"dataGaName":184},"/solutions/public-sector/","public sector",{"text":186,"config":187},"Pricing",{"href":188,"dataGaName":189,"dataGaLocation":46,"dataNavLevelOne":189},"/pricing/","pricing",{"text":191,"config":192,"link":194,"lists":198,"feature":278},"Resources",{"dataNavLevelOne":193},"resources",{"text":195,"config":196},"View all resources",{"href":197,"dataGaName":193,"dataGaLocation":46},"/resources/",[199,232,250],{"title":200,"items":201},"Getting started",[202,207,212,217,222,227],{"text":203,"config":204},"Install",{"href":205,"dataGaName":206,"dataGaLocation":46},"/install/","install",{"text":208,"config":209},"Quick start guides",{"href":210,"dataGaName":211,"dataGaLocation":46},"/get-started/","quick setup checklists",{"text":213,"config":214},"Learn",{"href":215,"dataGaLocation":46,"dataGaName":216},"https://university.gitlab.com/","learn",{"text":218,"config":219},"Product documentation",{"href":220,"dataGaName":221,"dataGaLocation":46},"https://docs.gitlab.com/","product documentation",{"text":223,"config":224},"Best practice videos",{"href":225,"dataGaName":226,"dataGaLocation":46},"/getting-started-videos/","best practice videos",{"text":228,"config":229},"Integrations",{"href":230,"dataGaName":231,"dataGaLocation":46},"/integrations/","integrations",{"title":233,"items":234},"Discover",[235,240,245],{"text":236,"config":237},"Customer success stories",{"href":238,"dataGaName":239,"dataGaLocation":46},"/customers/","customer success stories",{"text":241,"config":242},"Blog",{"href":243,"dataGaName":244,"dataGaLocation":46},"/blog/","blog",{"text":246,"config":247},"Remote",{"href":248,"dataGaName":249,"dataGaLocation":46},"https://handbook.gitlab.com/handbook/company/culture/all-remote/","remote",{"title":251,"items":252},"Connect",[253,258,263,268,273],{"text":254,"config":255},"GitLab Services",{"href":256,"dataGaName":257,"dataGaLocation":46},"/services/","services",{"text":259,"config":260},"Community",{"href":261,"dataGaName":262,"dataGaLocation":46},"/community/","community",{"text":264,"config":265},"Forum",{"href":266,"dataGaName":267,"dataGaLocation":46},"https://forum.gitlab.com/","forum",{"text":269,"config":270},"Events",{"href":271,"dataGaName":272,"dataGaLocation":46},"/events/","events",{"text":274,"config":275},"Partners",{"href":276,"dataGaName":277,"dataGaLocation":46},"/partners/","partners",{"backgroundColor":279,"textColor":280,"text":281,"image":282,"link":286},"#2f2a6b","#fff","Insights for the future of software development",{"altText":283,"config":284},"the source promo card",{"src":285},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758208064/dzl0dbift9xdizyelkk4.svg",{"text":287,"config":288},"Read the latest",{"href":289,"dataGaName":290,"dataGaLocation":46},"/the-source/","the source",{"text":292,"config":293,"lists":295},"Company",{"dataNavLevelOne":294},"company",[296],{"items":297},[298,303,309,311,316,321,326,331,336,341,346],{"text":299,"config":300},"About",{"href":301,"dataGaName":302,"dataGaLocation":46},"/company/","about",{"text":304,"config":305,"footerGa":308},"Jobs",{"href":306,"dataGaName":307,"dataGaLocation":46},"/jobs/","jobs",{"dataGaName":307},{"text":269,"config":310},{"href":271,"dataGaName":272,"dataGaLocation":46},{"text":312,"config":313},"Leadership",{"href":314,"dataGaName":315,"dataGaLocation":46},"/company/team/e-group/","leadership",{"text":317,"config":318},"Team",{"href":319,"dataGaName":320,"dataGaLocation":46},"/company/team/","team",{"text":322,"config":323},"Handbook",{"href":324,"dataGaName":325,"dataGaLocation":46},"https://handbook.gitlab.com/","handbook",{"text":327,"config":328},"Investor relations",{"href":329,"dataGaName":330,"dataGaLocation":46},"https://ir.gitlab.com/","investor relations",{"text":332,"config":333},"Trust Center",{"href":334,"dataGaName":335,"dataGaLocation":46},"/security/","trust center",{"text":337,"config":338},"AI Transparency Center",{"href":339,"dataGaName":340,"dataGaLocation":46},"/ai-transparency-center/","ai transparency center",{"text":342,"config":343},"Newsletter",{"href":344,"dataGaName":345,"dataGaLocation":46},"/company/contact/#contact-forms","newsletter",{"text":347,"config":348},"Press",{"href":349,"dataGaName":350,"dataGaLocation":46},"/press/","press",{"text":352,"config":353,"lists":354},"Contact us",{"dataNavLevelOne":294},[355],{"items":356},[357,360,365],{"text":53,"config":358},{"href":55,"dataGaName":359,"dataGaLocation":46},"talk to sales",{"text":361,"config":362},"Support portal",{"href":363,"dataGaName":364,"dataGaLocation":46},"https://support.gitlab.com","support portal",{"text":366,"config":367},"Customer portal",{"href":368,"dataGaName":369,"dataGaLocation":46},"https://customers.gitlab.com/customers/sign_in/","customer portal",{"close":371,"login":372,"suggestions":379},"Close",{"text":373,"link":374},"To search repositories and projects, login to",{"text":375,"config":376},"gitlab.com",{"href":60,"dataGaName":377,"dataGaLocation":378},"search login","search",{"text":380,"default":381},"Suggestions",[382,384,388,390,394,398],{"text":75,"config":383},{"href":80,"dataGaName":75,"dataGaLocation":378},{"text":385,"config":386},"Code Suggestions (AI)",{"href":387,"dataGaName":385,"dataGaLocation":378},"/solutions/code-suggestions/",{"text":109,"config":389},{"href":111,"dataGaName":109,"dataGaLocation":378},{"text":391,"config":392},"GitLab on AWS",{"href":393,"dataGaName":391,"dataGaLocation":378},"/partners/technology-partners/aws/",{"text":395,"config":396},"GitLab on Google Cloud",{"href":397,"dataGaName":395,"dataGaLocation":378},"/partners/technology-partners/google-cloud-platform/",{"text":399,"config":400},"Why GitLab?",{"href":88,"dataGaName":399,"dataGaLocation":378},{"freeTrial":402,"mobileIcon":407,"desktopIcon":412,"secondaryButton":415},{"text":403,"config":404},"Start free trial",{"href":405,"dataGaName":51,"dataGaLocation":406},"https://gitlab.com/-/trials/new/","nav",{"altText":408,"config":409},"Gitlab Icon",{"src":410,"dataGaName":411,"dataGaLocation":406},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203874/jypbw1jx72aexsoohd7x.svg","gitlab icon",{"altText":408,"config":413},{"src":414,"dataGaName":411,"dataGaLocation":406},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1758203875/gs4c8p8opsgvflgkswz9.svg",{"text":416,"config":417},"Get Started",{"href":418,"dataGaName":419,"dataGaLocation":406},"https://gitlab.com/-/trial_registrations/new?glm_source=about.gitlab.com/get-started/","get started",{"freeTrial":421,"mobileIcon":425,"desktopIcon":427},{"text":422,"config":423},"Learn more about GitLab Duo",{"href":80,"dataGaName":424,"dataGaLocation":406},"gitlab duo",{"altText":408,"config":426},{"src":410,"dataGaName":411,"dataGaLocation":406},{"altText":408,"config":428},{"src":414,"dataGaName":411,"dataGaLocation":406},{"button":430,"mobileIcon":435,"desktopIcon":437},{"text":431,"config":432},"/switch",{"href":433,"dataGaName":434,"dataGaLocation":406},"#contact","switch",{"altText":408,"config":436},{"src":410,"dataGaName":411,"dataGaLocation":406},{"altText":408,"config":438},{"src":439,"dataGaName":411,"dataGaLocation":406},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1773335277/ohhpiuoxoldryzrnhfrh.png",{"freeTrial":441,"mobileIcon":446,"desktopIcon":448},{"text":442,"config":443},"Back to pricing",{"href":188,"dataGaName":444,"dataGaLocation":406,"icon":445},"back to pricing","GoBack",{"altText":408,"config":447},{"src":410,"dataGaName":411,"dataGaLocation":406},{"altText":408,"config":449},{"src":414,"dataGaName":411,"dataGaLocation":406},{"title":451,"button":452,"config":457},"See how agentic AI transforms software delivery",{"text":453,"config":454},"Watch GitLab Transcend now",{"href":455,"dataGaName":456,"dataGaLocation":46},"/events/transcend/virtual/","transcend event",{"layout":458,"icon":459,"disabled":29},"release","AiStar",{"data":461},{"text":462,"source":463,"edit":469,"contribute":474,"config":479,"items":484,"minimal":689},"Git is a trademark of Software Freedom Conservancy and our use of 'GitLab' is under license",{"text":464,"config":465},"View page source",{"href":466,"dataGaName":467,"dataGaLocation":468},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/","page source","footer",{"text":470,"config":471},"Edit this page",{"href":472,"dataGaName":473,"dataGaLocation":468},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/content/","web ide",{"text":475,"config":476},"Please contribute",{"href":477,"dataGaName":478,"dataGaLocation":468},"https://gitlab.com/gitlab-com/marketing/digital-experience/about-gitlab-com/-/blob/main/CONTRIBUTING.md/","please contribute",{"twitter":480,"facebook":481,"youtube":482,"linkedin":483},"https://twitter.com/gitlab","https://www.facebook.com/gitlab","https://www.youtube.com/channel/UCnMGQ8QHMAnVIsI3xJrihhg","https://www.linkedin.com/company/gitlab-com",[485,532,584,628,655],{"title":186,"links":486,"subMenu":501},[487,491,496],{"text":488,"config":489},"View plans",{"href":188,"dataGaName":490,"dataGaLocation":468},"view plans",{"text":492,"config":493},"Why Premium?",{"href":494,"dataGaName":495,"dataGaLocation":468},"/pricing/premium/","why premium",{"text":497,"config":498},"Why Ultimate?",{"href":499,"dataGaName":500,"dataGaLocation":468},"/pricing/ultimate/","why ultimate",[502],{"title":503,"links":504},"Contact Us",[505,508,510,512,517,522,527],{"text":506,"config":507},"Contact sales",{"href":55,"dataGaName":56,"dataGaLocation":468},{"text":361,"config":509},{"href":363,"dataGaName":364,"dataGaLocation":468},{"text":366,"config":511},{"href":368,"dataGaName":369,"dataGaLocation":468},{"text":513,"config":514},"Status",{"href":515,"dataGaName":516,"dataGaLocation":468},"https://status.gitlab.com/","status",{"text":518,"config":519},"Terms of use",{"href":520,"dataGaName":521,"dataGaLocation":468},"/terms/","terms of use",{"text":523,"config":524},"Privacy statement",{"href":525,"dataGaName":526,"dataGaLocation":468},"/privacy/","privacy statement",{"text":528,"config":529},"Cookie preferences",{"dataGaName":530,"dataGaLocation":468,"id":531,"isOneTrustButton":29},"cookie preferences","ot-sdk-btn",{"title":91,"links":533,"subMenu":542},[534,538],{"text":535,"config":536},"DevSecOps platform",{"href":73,"dataGaName":537,"dataGaLocation":468},"devsecops platform",{"text":539,"config":540},"AI-Assisted Development",{"href":80,"dataGaName":541,"dataGaLocation":468},"ai-assisted development",[543],{"title":544,"links":545},"Topics",[546,551,556,561,566,569,574,579],{"text":547,"config":548},"CICD",{"href":549,"dataGaName":550,"dataGaLocation":468},"/topics/ci-cd/","cicd",{"text":552,"config":553},"GitOps",{"href":554,"dataGaName":555,"dataGaLocation":468},"/topics/gitops/","gitops",{"text":557,"config":558},"DevOps",{"href":559,"dataGaName":560,"dataGaLocation":468},"/topics/devops/","devops",{"text":562,"config":563},"Version Control",{"href":564,"dataGaName":565,"dataGaLocation":468},"/topics/version-control/","version control",{"text":25,"config":567},{"href":568,"dataGaName":37,"dataGaLocation":468},"/topics/devsecops/",{"text":570,"config":571},"Cloud Native",{"href":572,"dataGaName":573,"dataGaLocation":468},"/topics/cloud-native/","cloud native",{"text":575,"config":576},"AI for Coding",{"href":577,"dataGaName":578,"dataGaLocation":468},"/topics/devops/ai-for-coding/","ai for coding",{"text":580,"config":581},"Agentic AI",{"href":582,"dataGaName":583,"dataGaLocation":468},"/topics/agentic-ai/","agentic ai",{"title":585,"links":586},"Solutions",[587,589,591,596,600,603,607,610,612,615,618,623],{"text":133,"config":588},{"href":128,"dataGaName":133,"dataGaLocation":468},{"text":122,"config":590},{"href":105,"dataGaName":106,"dataGaLocation":468},{"text":592,"config":593},"Agile development",{"href":594,"dataGaName":595,"dataGaLocation":468},"/solutions/agile-delivery/","agile delivery",{"text":597,"config":598},"SCM",{"href":118,"dataGaName":599,"dataGaLocation":468},"source code management",{"text":547,"config":601},{"href":111,"dataGaName":602,"dataGaLocation":468},"continuous integration & delivery",{"text":604,"config":605},"Value stream management",{"href":161,"dataGaName":606,"dataGaLocation":468},"value stream management",{"text":552,"config":608},{"href":609,"dataGaName":555,"dataGaLocation":468},"/solutions/gitops/",{"text":171,"config":611},{"href":173,"dataGaName":174,"dataGaLocation":468},{"text":613,"config":614},"Small business",{"href":178,"dataGaName":179,"dataGaLocation":468},{"text":616,"config":617},"Public sector",{"href":183,"dataGaName":184,"dataGaLocation":468},{"text":619,"config":620},"Education",{"href":621,"dataGaName":622,"dataGaLocation":468},"/solutions/education/","education",{"text":624,"config":625},"Financial services",{"href":626,"dataGaName":627,"dataGaLocation":468},"/solutions/finance/","financial services",{"title":191,"links":629},[630,632,634,636,639,641,643,645,647,649,651,653],{"text":203,"config":631},{"href":205,"dataGaName":206,"dataGaLocation":468},{"text":208,"config":633},{"href":210,"dataGaName":211,"dataGaLocation":468},{"text":213,"config":635},{"href":215,"dataGaName":216,"dataGaLocation":468},{"text":218,"config":637},{"href":220,"dataGaName":638,"dataGaLocation":468},"docs",{"text":241,"config":640},{"href":243,"dataGaName":244,"dataGaLocation":468},{"text":236,"config":642},{"href":238,"dataGaName":239,"dataGaLocation":468},{"text":246,"config":644},{"href":248,"dataGaName":249,"dataGaLocation":468},{"text":254,"config":646},{"href":256,"dataGaName":257,"dataGaLocation":468},{"text":259,"config":648},{"href":261,"dataGaName":262,"dataGaLocation":468},{"text":264,"config":650},{"href":266,"dataGaName":267,"dataGaLocation":468},{"text":269,"config":652},{"href":271,"dataGaName":272,"dataGaLocation":468},{"text":274,"config":654},{"href":276,"dataGaName":277,"dataGaLocation":468},{"title":292,"links":656},[657,659,661,663,665,667,669,673,678,680,682,684],{"text":299,"config":658},{"href":301,"dataGaName":294,"dataGaLocation":468},{"text":304,"config":660},{"href":306,"dataGaName":307,"dataGaLocation":468},{"text":312,"config":662},{"href":314,"dataGaName":315,"dataGaLocation":468},{"text":317,"config":664},{"href":319,"dataGaName":320,"dataGaLocation":468},{"text":322,"config":666},{"href":324,"dataGaName":325,"dataGaLocation":468},{"text":327,"config":668},{"href":329,"dataGaName":330,"dataGaLocation":468},{"text":670,"config":671},"Sustainability",{"href":672,"dataGaName":670,"dataGaLocation":468},"/sustainability/",{"text":674,"config":675},"Diversity, inclusion and belonging (DIB)",{"href":676,"dataGaName":677,"dataGaLocation":468},"/diversity-inclusion-belonging/","Diversity, inclusion and belonging",{"text":332,"config":679},{"href":334,"dataGaName":335,"dataGaLocation":468},{"text":342,"config":681},{"href":344,"dataGaName":345,"dataGaLocation":468},{"text":347,"config":683},{"href":349,"dataGaName":350,"dataGaLocation":468},{"text":685,"config":686},"Modern Slavery Transparency Statement",{"href":687,"dataGaName":688,"dataGaLocation":468},"https://handbook.gitlab.com/handbook/legal/modern-slavery-act-transparency-statement/","modern slavery transparency statement",{"items":690},[691,694,697],{"text":692,"config":693},"Terms",{"href":520,"dataGaName":521,"dataGaLocation":468},{"text":695,"config":696},"Cookies",{"dataGaName":530,"dataGaLocation":468,"id":531,"isOneTrustButton":29},{"text":698,"config":699},"Privacy",{"href":525,"dataGaName":526,"dataGaLocation":468},[701,714],{"id":702,"title":19,"body":9,"config":703,"content":705,"description":9,"extension":27,"meta":709,"navigation":29,"path":710,"seo":711,"stem":712,"__hash__":713},"blogAuthors/en-us/blog/authors/hiroki-suezawa.yml",{"template":704},"BlogAuthor",{"name":19,"config":706},{"headshot":707,"ctfId":708},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749662370/Blog/Author%20Headshots/hiroki_suezawa.png","cw6ZIj0yjr1uw2LAFr23h",{},"/en-us/blog/authors/hiroki-suezawa",{},"en-us/blog/authors/hiroki-suezawa","5SalhkDKL3j3O2JyjOofwTFjgEeNhoGH8nLQDeI7N7c",{"id":715,"title":20,"body":9,"config":716,"content":717,"description":9,"extension":27,"meta":721,"navigation":29,"path":722,"seo":723,"stem":724,"__hash__":725},"blogAuthors/en-us/blog/authors/mitra-jozenazemian.yml",{"template":704},{"name":20,"config":718},{"headshot":719,"ctfId":720},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1749662373/Blog/Author%20Headshots/Screenshot_2024-10-25_at_8.23.56_AM.png","4suqsutT8w5ZmkIvSVrmWQ",{},"/en-us/blog/authors/mitra-jozenazemian",{},"en-us/blog/authors/mitra-jozenazemian","DzsXAQdHz5kyKwnJ-xGAs_Wr_lt8XWi_J9LQijwLiVY",[727,740,755],{"content":728,"config":738},{"title":729,"description":730,"authors":731,"date":733,"body":734,"category":10,"tags":735,"heroImage":737},"Prepare your pipeline for AI-discovered zero-days","AI is finding vulnerabilities faster than teams can patch. Learn how pipeline enforcement, automated triage, and AI remediation close the gap.",[732],"Omer Azaria","2026-04-20","Anthropic's [Mythos Preview model](https://red.anthropic.com/2026/mythos-preview/) recently identified thousands of zero-day vulnerabilities across every major operating system and web browser, including an OpenBSD bug that went undetected for 27 years. In testing, Mythos autonomously chained four vulnerabilities into a working browser exploit that escaped its sandbox. Anthropic is restricting access to Mythos, but the company’s head of offensive cyber research expects threats to have comparable tooling within six to twelve months.\n\nThe defender side of the equation hasn't kept pace. One third of exploited Common Vulnerabilities and Exposures (CVEs) in the first half of 2025 showed activity on or before disclosure day, before most teams even know there's something to patch. AI is compressing that window further, accelerating attackers and flooding teams with whitehat disclosures faster than they can triage. Defender tooling has improved, but most organizations can't operationalize it fast enough to close the gap between discovery and exploitation.\n\nWhen the window between disclosure and exploitation is measured in hours, the security team can't be the last line of defense. Security has to run where code enters the system: in the pipeline, on every merge request, enforced by policy. The fixes that can be automated should be. The ones that can't need to reach the right human faster than they do today.\n\n## Known vulnerabilities are already outpacing remediation\n\nThe bottleneck isn't detection, it's acting at scale on what teams already know. Sixty percent of breaches in the 2025 Verizon DBIR involved exploiting known vulnerabilities where a patch was already available. Teams couldn’t close them in time.\n\nThe backlog was untenable before Mythos. Developers spend [11 hours per month remediating vulnerabilities](https://about.gitlab.com/resources/developer-survey/) post-release instead of shipping new work. Over half of organizations have at least one open internet-facing vulnerability, and the median time to close half of those is 361 days. Exploitation takes hours, while remediation takes months.\n\nAI-assisted development is widening the gap, and stakeholders know it. By June 2025, AI-generated code was adding over 10,000 new security findings per month across Fortune 50 repositories, a 10x jump from six months earlier. Georgia Tech identified 34 [CVEs attributable to AI-generated code](https://research.gatech.edu/bad-vibes-ai-generated-code-vulnerable-researchers-warn) in March 2026, up from 6 in January, and that count reflects only the ones where AI authorship is clear. AI coding assistants hallucinate package names, reach for outdated patterns, and copy insecure examples from training data. More code, more dependencies, and more vulnerabilities per line are generated faster than security teams can review them.\n\nDefenders need to harness frontier AI models, too — not bolted onto the SDLC as external tooling, but running inside the same policies, approvals, and audit trail as the rest of the team. \n\n## Security at the speed of AI coding\n\nWhen a critical CVE drops, how quickly can your team confirm which projects are affected? How many tools does an alert cross before a developer can submit a fix?\n\nThe teams that benefit most from AI already have policies, enforcement, and controls embedded in their development workflows. AI amplifies that foundation. It doesn't replace it.\n\n**Enforcement at the point of change.** As exploitation windows compress, every line of code entering a repository needs to pass through a defined set of controls. Not a separate review, in a different tool, by a different team. Organizations need the ability to enforce security policies across every group and project, with the merge request as the enforcement point. Policies defined once, applied everywhere, with exceptions reviewed, approved, and logged.\n\n**Simple issues caught before the merge request, not during.** Hardcoded secrets, known-vulnerable imports, and deprecated API calls can be flagged in the IDE before a developer pushes a commit. Catching them at authoring time means fewer findings blocking the MR, so review cycles go to the findings that require cross-component context: reachability, exploitability, and architectural risk.\n\n**Triage automated by default, not by exception.** Embedding security into every merge request creates a volume problem. More scans, more findings, more noise reaching developers who aren’t trained to distinguish a reachable critical from a theoretical one. AI must handle false positive detection, reachability, exploitability context, and severity assessment before a developer sees the finding, so the findings they see actually warrant their time.\n\n**Remediation governed like any other change.** AI-based remediation compresses the timeline for closing vulnerabilities, but every generated fix must move through the same governance as a human-authored change: policies enforce scans, the right reviewers approve, and evidence is recorded. GitLab’s automated remediation capability proposes each fix in a merge request with a confidence score. The MR records which policy applied, which scans ran, what they found, and who approved. Human code and AI-generated code move through the same process, with the same audit trail.\n\n## What a ready pipeline looks like\n\nHere's how these pieces work together when a high-severity vulnerability is discovered and the clock is running.\n\nA proof-of-concept exploit for a vulnerability in a popular open-source package appears on a security mailing list. There’s no CVE, no National Vulnerability Database (NVD) entry, and no scanner signature yet. The security team finds out the usual way: someone shares it in Slack.\n\nA security engineer asks the security agent if the package is in use, which projects have affected versions, and whether any vulnerable call paths are reachable in production. The agent checks the dependency graph for every project, matches the affected versions and entry points from the disclosure, and returns a ranked list of exposed projects with details about reachability. There’s no need to search through repositories by hand or wait for a scanner update. The question, \"Are we exposed?\" is answered in minutes.\n\nThe engineer starts a remediation campaign for every exposed project. The remediation agent suggests fixes: version updates where a patched release is available, and targeted call-path patches where it is not. Scan execution policies are already in place for projects tagged SOC 2. The engineer hardens the rules to block merges on any merge request that introduces or keeps the affected dependency, and an approval policy now requires security sign-off on every fix. The agent's first proposed patch fails the pipeline when an integration test catches a regression. The agent revises the patch based on the test failure, and the second attempt passes. Developers review the changes, security signs off under the stricter policy, and merges proceed across the campaign.\n\nAt the next audit review, the security team presents a report showing how policies were enforced and risks were reduced during the campaign. It includes scan results, policies applied, approvers, and merge timestamps for every MR in every affected project. The evidence was automatically generated in flight, not assembled after the fact.\n\n## Close the gaps now\n\nMythos exists today, and comparable models will be in attacker hands within a year. Every month between now and then is a chance to strengthen your software supply chain.\n\nAsk these questions about your pipeline:\n\n* How do you enforce that security scans run on every merge request, not just the projects where teams configured them?\n\n* If a compromised package entered your dependency tree today, would your pipeline catch it before build?\n\n* When a scanner flags a critical finding, how many tool boundaries does it cross before a developer starts the fix?\n\n* If an AI agent proposed a code fix for a vulnerability, what process would that fix go through before reaching production, and is that process auditable?\n\n* When auditors ask for evidence that a specific policy was enforced on a specific change, how long does it take to produce?\n\nIf the answers expose gaps, address them now. [Talk to a GitLab solutions architect](https://about.gitlab.com/sales/) about the role of security governance in your development lifecycle.",[736,10,535],"AI/ML","https://res.cloudinary.com/about-gitlab-com/image/upload/v1772195014/ooezwusxjl1f7ijfmbvj.png",{"featured":29,"template":14,"slug":739},"prepare-your-pipeline-for-ai-discovered-zero-days",{"content":741,"config":753},{"title":742,"description":743,"authors":744,"heroImage":746,"date":747,"category":10,"tags":748,"body":752},"Manage vulnerability noise at scale with auto-dismiss policies","Learn how to cut through scanner noise and focus on the vulnerabilities that matter most with GitLab security, including use cases and templates.",[745],"Grant Hickman","https://res.cloudinary.com/about-gitlab-com/image/upload/v1774375772/kpaaaiqhokevxxeoxvu0.png","2026-03-25",[10,749,25,750,751],"tutorial","features","product","Security scanners are essential, but not every finding requires action. Test code, vendored dependencies, generated files, and known false positives create noise that buries the vulnerabilities that actually matter. Security teams waste hours manually dismissing the same irrelevant findings across projects and pipelines. They experience slower triage, alert fatigue, and developer friction that undermines adoption of security scanning itself.\n\nGitLab's auto-dismiss vulnerability policies let you codify your triage decisions once and apply them automatically on every default-branch pipeline. Define criteria based on file path, directory, or vulnerability identifier (CVE, CWE), choose a dismissal reason, and let GitLab handle the rest.\n\n## Why auto-dismiss?\nAuto-dismiss vulnerability policies enable security teams to:\n- **Eliminate triage noise**: Automatically dismiss findings in test code, vendored dependencies, and generated files.\n- **Enforce decisions at scale**: Apply policies centrally to dismiss known false positives across your entire organization.\n- **Maintain audit transparency**: Every auto-dismissed finding includes a documented reason and links back to the policy that triggered it.\n- **Preserve the record**: Unlike scanner exclusions, dismissed vulnerabilities remain in your report, so you can revisit decisions if conditions change.\n\n## How auto-dismiss policies work\n\n1. **Define your policy** in a vulnerability management policy YAML file. Specify match criteria (file path, directory, or identifier) and a dismissal reason.\n\n2. **Merge and activate.** Create the policy via **Secure > Policies > New  policy > Vulnerability management policy**. Merge the MR to enable it.\n3. **Run your pipeline.** On every default-branch pipeline, matching vulnerabilities are automatically set to \"Dismissed\" with the specified reason. Up to 1,000 vulnerabilities are processed per run.\n4. **Measure the impact.** Filter your vulnerability report by status \"Dismissed\" to see exactly what was cleaned up and validate that the right findings are being handled.\n\n## Use cases with ready-to-use configurations\n\nEach example below includes a policy configuration you can copy, customize, and apply immediately.\n\n### 1. Dismiss test code vulnerabilities\n\nSAST and dependency scanners flag hardcoded credentials, insecure fixtures, and dev-only dependencies in test directories. These are not production risks.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Dismiss test code vulnerabilities\"\n    description: \"Auto-dismiss findings in test directories\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: file_path\n            value: \"test/**/*\"\n      - type: detected\n        criteria:\n          - type: file_path\n            value: \"tests/**/*\"\n      - type: detected\n        criteria:\n          - type: file_path\n            value: \"spec/**/*\"\n      - type: detected\n        criteria:\n          - type: directory\n            value: \"__tests__/*\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: used_in_tests\n\n```\n\n### 2. Dismiss vendored and third-party code\n\nVulnerabilities in `vendor/`, `third_party/`, or checked-in `node_modules` are managed upstream and not actionable for your team.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Dismiss vendored dependency findings\"\n    description: \"Findings in vendored code are managed upstream\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: directory\n            value: \"vendor/*\"\n      - type: detected\n        criteria:\n          - type: directory\n            value: \"third_party/*\"\n      - type: detected\n        criteria:\n          - type: directory\n            value: \"vendored/*\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: not_applicable\n\n```\n\n### 3. Dismiss known false positive CVEs\n\nCertain CVEs are repeatedly flagged but don't apply to your usage context. Teams dismiss these manually every time they appear. Replace the example CVEs below with your own.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Dismiss known false positive CVEs\"\n    description: \"CVEs confirmed as false positives for our environment\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CVE-2023-44487\"\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CVE-2024-29041\"\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CVE-2023-26136\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: false_positive\n\n```\n\n### 4. Dismiss generated and auto-created code\n\nProtobuf, gRPC, OpenAPI generators, and ORM scaffolding tools produce files with flagged patterns that cannot be patched by your team.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Dismiss generated code findings\"\n    description: \"Generated files are not authored by us\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: directory\n            value: \"generated/*\"\n      - type: detected\n        criteria:\n          - type: file_path\n            value: \"**/*.pb.go\"\n      - type: detected\n        criteria:\n          - type: file_path\n            value: \"**/*.generated.*\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: not_applicable\n\n```\n\n### 5. Dismiss infrastructure-mitigated vulnerabilities\n\nVulnerability classes like XSS (CWE-79) or SQL injection (CWE-89) that are already addressed by WAF rules or runtime protection. Only use this when mitigating controls are verified and consistently enforced.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Dismiss CWEs mitigated by WAF\"\n    description: \"XSS and SQLi mitigated by WAF rules\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CWE-79\"\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CWE-89\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: mitigating_control\n\n```\n\n### 6. Dismiss CVE families across your organization\n\nA wave of related CVEs for a widely-used library your team has assessed? Apply at the group level to dismiss them across dozens of projects. The wildcard pattern (e.g., `CVE-2021-44*`) matches all CVEs with that prefix.\n\n```yaml\nvulnerability_management_policy:\n  - name: \"Accept risk for log4j CVE family\"\n    description: \"Log4j CVEs mitigated by version pinning and WAF\"\n    enabled: true\n    rules:\n      - type: detected\n        criteria:\n          - type: identifier\n            value: \"CVE-2021-44*\"\n    actions:\n      - type: auto_dismiss\n        dismissal_reason: acceptable_risk\n\n```\n\n## Quick reference\n\n| Parameter | Details |\n|-----------|---------|\n| **Criteria types** | `file_path` (glob patterns, e.g., `test/**/*`), `directory` (e.g., `vendor/*`), `identifier` (CVE/CWE with wildcards, e.g., `CVE-2023-*`) |\n| **Dismissal reasons** | `acceptable_risk`, `false_positive`, `mitigating_control`, `used_in_tests`, `not_applicable` |\n| **Criteria logic** | Multiple criteria within a rule = AND (must match all). Multiple rules within a policy = OR (match any). |\n| **Limits** | 3 criteria per rule, 5 rules per policy, 5 policies per security policy project. Vulnerabilty management policy actions process 1000 vulnerabilities per pipeline run in the target project, until all matching vulnerabilities are processed. |\n| **Affected statuses** | Needs triage, Confirmed |\n| **Scope** | Project-level or group-level (group-level applies across all projects) |\n\n## Getting started\nHere's how to get started with auto-dismiss policies:\n\n1. **Identify the noise.** Open your vulnerability report and sort by \"Needs triage.\" Look for patterns: test files, vendored code, the same CVE across projects.\n\n2. **Pick a scenario.** Start with whichever use case above accounts for the most findings.\n\n3. **Record your baseline.** Note the number of \"Needs triage\" vulnerabilities before creating a policy.\n\n4. **Create and enable.** Navigate to **Secure > Policies > New policy > Vulnerability management policy**. Paste the configuration from the use case above, then merge the MR.\n\n5. **Validate results.** After the next default-branch pipeline, filter by status \"Dismissed\" to confirm the right findings were handled.\n\nFor full configuration details, see the [vulnerability management policy documentation](https://docs.gitlab.com/user/application_security/policies/vulnerability_management_policy/#auto-dismiss-policies).\n\n> Ready to take control of vulnerability noise? [Start a free GitLab Ultimate trial](https://about.gitlab.com/free-trial/) and configure your first auto-dismiss policy today.\n",{"slug":754,"featured":29,"template":14},"auto-dismiss-vulnerability-management-policy",{"content":756,"config":765},{"title":757,"description":758,"authors":759,"heroImage":761,"date":762,"body":763,"category":10,"tags":764},"GitLab 18.10 brings AI-native triage and remediation ","Learn about GitLab Duo Agent Platform capabilities that cut noise, surface real vulnerabilities, and turn findings into proposed fixes.",[760],"Alisa Ho","https://res.cloudinary.com/about-gitlab-com/image/upload/v1773843921/rm35fx4gylrsu9alf2fx.png","2026-03-19","GitLab 18.10 introduces new AI-powered security capabilities focused on improving the quality and speed of vulnerability management. Together, these features can help reduce the time developers spend investigating false positives and bring automated remediation directly into their workflow, so they can fix vulnerabilities without needing to be security experts.\n\nHere is what’s new:\n\n* [**Static Application Security Testing (SAST) false positive detection**](https://docs.gitlab.com/user/application_security/vulnerabilities/false_positive_detection/) **is now generally available.** This flow uses an LLM for agentic reasoning to determine the likelihood that a vulnerability is a false positive or not, so security and development teams can focus on remediating critical vulnerabilities first.  \n* [**Agentic SAST vulnerability resolution**](https://docs.gitlab.com/user/application_security/vulnerabilities/agentic_vulnerability_resolution/) **is now in beta.** Agentic SAST vulnerability resolution automatically creates a merge request with a proposed fix for verified SAST vulnerabilities, which can shorten time to remediation and reduce the need for deep security expertise.  \n* [**Secret false positive detection**](https://docs.gitlab.com/user/application_security/vulnerabilities/secret_false_positive_detection/) **is now in beta.** This flow brings the same AI-powered noise reduction to secret detection, flagging dummy and test secrets to save review effort.\n\nThese flows are available to GitLab Ultimate customers using GitLab Duo Agent Platform. \n\n## Cut triage time with SAST false positive detection\n\nTraditional SAST scanners flag every suspicious code pattern they find, regardless of whether code paths are reachable or frameworks already handle the risk. Without runtime context, they cannot distinguish a real vulnerability from safe code that just looks dangerous.\n\nThis means developers could spend hours investigating findings that turn out to be false positives. Over time, that can erode confidence in the report and slow down the teams responsible for fixing real risks.\n\nAfter each SAST scan, GitLab Duo Agent Platform automatically analyzes new critical and high severity findings and attaches:\n\n* A confidence score indicating how likely the finding is to be a false positive  \n* An AI-generated explanation describing the reasoning  \n* A visual badge that makes “Likely false positive” versus “Likely real” easy to scan in the UI\n\nThese findings appear in the [Vulnerability Report](https://docs.gitlab.com/user/application_security/vulnerability_report/), as shown below. You can filter the report to focus on findings marked as “Not false positive” so teams can spend their time addressing real vulnerabilities instead of sifting through noise.\n\n![Vulnerability report](https://res.cloudinary.com/about-gitlab-com/image/upload/v1773844787/i0eod01p7gawflllkgsr.png)\n\n\nGitLab Duo Agent Platform's assessment is a recommendation. You stay in control of every false positive to determine if it is valid, and you can audit the agent's reasoning at any time to build confidence in the model. \n\n\n## Turn vulnerabilities into automated fixes\n\nKnowing that a vulnerability is real is only half the work.  Remediation still requires understanding the code path, writing a safe patch, and making sure nothing else breaks.\n\nIf the vulnerability is identified as likely not be a false positive by the SAST false positive detection flow, the Agentic SAST vulnerability resolution flow automatically:\n\n1. Reads the vulnerable code and surrounding context from your repository  \n2. Generates high-quality proposed fixes  \n3. Validates fixes through automated testing   \n4. Opens a merge request with a proposed fix that includes:  \n   * Concrete code changes  \n   * A confidence score  \n   * An explanation of what changed and why\n\nIn this demo, you’ll see how GitLab can automatically take a SAST vulnerability all the way from detection to a ready-to-review merge request. Watch how the agent reads the code, generates and validates a fix, and opens an MR with clear, explainable changes so developers can remediate faster without being security experts.\n\n\u003Ciframe src=\"https://player.vimeo.com/video/1174573325?badge=0&autopause=0&player_id=0&app_id=58479\" frameborder=\"0\" allow=\"autoplay; fullscreen; picture-in-picture; clipboard-write; encrypted-media; web-share\" referrerpolicy=\"strict-origin-when-cross-origin\" style=\"position:absolute;top:0;left:0;width:100%;height:100%;\" title=\"GitLab 18.10 AI SAST False Positive Auto Remediation\">\u003C/iframe>\u003Cscript src=\"https://player.vimeo.com/api/player.js\">\u003C/script>\n\nAs with any AI-generated suggestion, you should review the proposed merge request carefully before merging.\n\n## Surface real secrets\n\nSecret detection is only useful if teams trust the results. When reports are full of test credentials, placeholder values, and example tokens, developers may waste time reviewing noise instead of fixing real exposures. That can slow remediation and decrease confidence in the scan.\n\nSecret false positive detection helps teams focus on the secrets that matter so they can reduce risk faster. When it runs on the default branch, it will automatically:\n\n1. Analyze each finding to spot likely test credentials, example values, and dummy secrets  \n2. Assign a confidence score for whether the finding is a real risk or a likely false positive  \n3. Generate an explanation for why the secret is being treated as real or noise  \n4. Add a badge in the Vulnerability Report so developers can see the status at a glance\n\nDevelopers can also trigger this analysis manually from the Vulnerability Report by selecting **“Check for false positive”** on any secret detection finding, helping them clear out findings that do not pose risk and focus on real secrets sooner.\n\n## Try AI-powered security today\n\nGitLab 18.10 introduces capabilities that cover the full vulnerability workflow, from cutting false positive noise in SAST and secret detection to automatically generating merge requests with proposed fixes.\n\nTo see how AI-powered security can help cut review time and turn findings into ready-to-merge fixes, [start a free trial of GitLab Duo Agent Platform today](https://about.gitlab.com/gitlab-duo-agent-platform/?utm_medium=blog&utm_source=blog&utm_campaign=eg_global_x_x_security_en_).",[751,10,750],{"featured":13,"template":14,"slug":766},"gitlab-18-10-brings-ai-native-triage-and-remediation",{"promotions":768},[769,783,794,805],{"id":770,"categories":771,"header":773,"text":774,"button":775,"image":780},"ai-modernization",[772],"ai-ml","Is AI achieving its promise at scale?","Quiz will take 5 minutes or less",{"text":776,"config":777},"Get your AI maturity score",{"href":778,"dataGaName":779,"dataGaLocation":244},"/assessments/ai-modernization-assessment/","modernization assessment",{"config":781},{"src":782},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/qix0m7kwnd8x2fh1zq49.png",{"id":784,"categories":785,"header":786,"text":774,"button":787,"image":791},"devops-modernization",[751,37],"Are you just managing tools or shipping innovation?",{"text":788,"config":789},"Get your DevOps maturity score",{"href":790,"dataGaName":779,"dataGaLocation":244},"/assessments/devops-modernization-assessment/",{"config":792},{"src":793},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138785/eg818fmakweyuznttgid.png",{"id":795,"categories":796,"header":797,"text":774,"button":798,"image":802},"security-modernization",[10],"Are you trading speed for security?",{"text":799,"config":800},"Get your security maturity score",{"href":801,"dataGaName":779,"dataGaLocation":244},"/assessments/security-modernization-assessment/",{"config":803},{"src":804},"https://res.cloudinary.com/about-gitlab-com/image/upload/v1772138786/p4pbqd9nnjejg5ds6mdk.png",{"id":806,"paths":807,"header":810,"text":811,"button":812,"image":817},"github-azure-migration",[808,809],"migration-from-azure-devops-to-gitlab","integrating-azure-devops-scm-and-gitlab","Is your team ready for GitHub's Azure move?","GitHub is already rebuilding around Azure. Find out what it means for you.",{"text":813,"config":814},"See how GitLab compares to GitHub",{"href":815,"dataGaName":816,"dataGaLocation":244},"/compare/gitlab-vs-github/github-azure-migration/","github azure migration",{"config":818},{"src":793},{"header":820,"blurb":821,"button":822,"secondaryButton":827},"Start building faster today","See what your team can do with the intelligent orchestration platform for DevSecOps.\n",{"text":823,"config":824},"Get your free trial",{"href":825,"dataGaName":51,"dataGaLocation":826},"https://gitlab.com/-/trial_registrations/new?glm_content=default-saas-trial&glm_source=about.gitlab.com/","feature",{"text":506,"config":828},{"href":55,"dataGaName":56,"dataGaLocation":826},1777302606214]