October 2025 Summaries

Harness has introduced support for the Model Context Protocol (MCP) within its Feature Management and Experimentation (FME) platform, which facilitates managing feature flags using AI-powered IDEs like Claude Code and Windsurf. This integration allows developers to use natural language prompts to explore, understand, and manage feature flags directly within their development environment, thereby reducing context switching and simplifying workflows. The tools enable teams to audit, compare, and inspect feature flags across different projects and environments, aligning with best practices for governance and risk mitigation. With a user-friendly interface, non-technical stakeholders can also query and understand feature flags without needing deep technical expertise. The approach aims to streamline feature management, making it more conversational and integrated into the development workflow, ultimately helping teams make faster, data-driven decisions.

In the third quarter of 2025, Harness Continuous Deployment introduced significant enhancements to expand platform support and infrastructure flexibility, including alpha Salesforce deployments, Serverless Framework V4, VM container step groups, and Azure Terraform connectors. The update also included improvements in pipeline resiliency, data retention, and GitOps features, setting the stage for future advancements in performance, observability, and governance. Notable features include automated Salesforce deployments, support for Google Cloud Run without immediate traffic shifts, and enhanced rollback capabilities for deployment failures. Additionally, Harness enhanced its GitOps functionalities with first-class ApplicationSet support, improved service summaries, and scalability optimizations, reinforcing its commitment to empowering teams with secure and scalable continuous delivery tools.

Harness Chaos Engineering's GCP Cloud Monitoring probe transforms existing Google Cloud Platform (GCP) metrics into objective, automated pass/fail validations for chaos experiments, eliminating the need for subjective observation. By supporting flexible authentication methods and PromQL queries, it allows users to validate infrastructure performance against predefined thresholds during controlled failure scenarios, thus creating an audit trail for system behavior under stress and enabling automated resilience testing within CI/CD pipelines. The probe integrates with GCP's Cloud Monitoring to leverage existing metrics, such as CPU usage and memory pressure, to provide clear success criteria for chaos experiments. It offers two authentication options: IAM with workload identity for seamless integration within GCP and GCP service account keys for more granular control, particularly useful when chaos infrastructure runs outside GCP. Users can craft PromQL queries to monitor specific metrics and define pass/fail thresholds, while tuning probe behavior through runtime properties to match experiment characteristics. This approach turns passive monitoring into active validation, providing an audit trail that aids in capacity planning, SLO refinement, and infrastructure budgeting, ultimately enhancing the learning process in chaos engineering by providing objective and repeatable measurements of system resilience.

Harness AI is a comprehensive feature of the Harness Platform, designed to enhance all aspects of the Software Delivery Lifecycle (SDLC) through AI-driven automation and intelligent features. Since 2017, Harness has led DevOps transformation by introducing AI capabilities that autonomously manage various stages of software delivery, from code generation and semantic search to the optimization of continuous integration and deployment pipelines. The platform offers tools for improving test coverage, managing incidents with confidence using feature flags and automated rollbacks, and controlling cloud costs through intelligent resource management. With a focus on enterprise security and privacy, Harness AI ensures user data remains private and is not used for training purposes. Organizations using Harness AI have reported significant improvements in efficiency, including faster build times, reduced test maintenance, and increased productivity in DevOps processes.

Harness, in collaboration with Google Cloud, has developed Harness AI, an AI-driven solution to streamline software delivery workflows by automating tasks such as testing, security scanning, and deployments. Utilizing Google Cloud's Vertex AI platform, Harness AI enables teams to create enterprise-grade CI/CD pipelines rapidly, automatically generate adaptable tests, and deploy with real-time issue detection, enhancing both speed and reliability. This integration ensures that security and compliance are maintained without compromising efficiency, as the Software Delivery Knowledge Graph provides a comprehensive, context-aware understanding of the entire software delivery environment. Harness AI operates within the existing Harness Platform, requiring no additional installations and offering seamless integration with current tools. The partnership also includes deeper integrations with Google Cloud, further enhancing automation and interoperability for engineering teams.

Feature Flags as a Service (FFaaS) provides a sophisticated solution for managing feature flags, offering scalable control, advanced targeting, and real-time analytics that surpasses the capabilities of in-house solutions. By decoupling deployment from release, FFaaS allows teams to accelerate delivery, reduce technical debt, and make informed decisions while maintaining high standards of stability and flexibility. This service enables teams to implement best practices such as continuous delivery, A/B testing, and canary releases, facilitating gradual transitions from monolithic to microservice architectures. A robust FFaaS platform supports complex use cases with tailored SDKs and APIs, ensuring seamless integration across applications. It enhances product decision-making with detailed analytics and provides crucial functionalities like a global kill switch, access control, and audit trails. Companies like Walmart Labs and platforms like Split demonstrate how leveraging FFaaS can speed up innovation and reduce risk.

New Relic probes integrated into Harness Chaos Engineering allow for objective validation of system performance against predefined Service Level Objectives (SLOs) during chaos experiments, offering metrics-driven resilience checks rather than subjective testing. By real-time querying New Relic metrics and comparing results against success criteria, teams can programmatically ensure systems maintain acceptable performance levels, even during induced failures, thus enabling automated reliability testing within CI/CD pipelines. This setup transforms chaos engineering into an automated, repeatable validation process by using probes to monitor application performance metrics continuously, helping to identify performance regressions before they reach production. The New Relic probe utilizes NRQL (New Relic Query Language) to target specific metrics, and users can define success criteria based on data type, comparison operator, and threshold values. This approach ensures that systems can handle adverse conditions by integrating probes into CI/CD pipelines, thus bolstering confidence in a system’s resilience and catching potential regressions early.

Implementing feature flags is a crucial strategy for software teams to test and deploy new features safely in production environments. This technique allows teams to toggle features on or off without altering the codebase, ensuring smoother rollouts and reducing the risk of widespread bugs. Best practices include targeting internal users for initial tests, automating test flows for ongoing validation, and using approval gates and canary releases for controlled deployments. Managing toggle debt is essential to maintain a clean codebase, as unused or conflicting toggles can accumulate over time. Platforms like Split offer tools to manage feature flags effectively, enabling teams to conduct A/B testing and gather contextual data to assess feature performance. By integrating feature flags into the continuous delivery process, teams can increase release velocity, reduce mean time to recovery (MTTR), and maintain high levels of developer confidence and productivity.

Canary releases and feature flag rollouts are two prevalent strategies in software deployment that aim to mitigate risks by gradually exposing new code to users. While both methods seek to enhance the safety and frequency of continuous delivery, their approaches differ significantly. Canary releases involve deploying a new version of an application to a subset of production machines, allowing for early feedback and gradual scaling across all machines, but they typically lack granular control over individual features. In contrast, feature flag rollouts deploy the updated version to all machines but control the exposure of new features through flags, which can be adjusted dynamically to target specific user groups. This approach allows for independent feature management, enabling quick rollbacks and adjustments without affecting the entire deployment. Combining both strategies can offer a robust framework for safer and more efficient software delivery, with feature flags providing superior targeting and rollback capabilities, thereby supporting continuous improvement and innovation.

Harness Database DevOps now integrates with CockroachDB, enhancing CI/CD automation for distributed SQL databases by allowing teams to manage schema changes via Git-driven workflows, ensuring consistency and traceability. This integration simplifies multi-environment deployments, reduces human error, and accelerates database delivery, leveraging CockroachDB's distributed SQL capabilities and fault tolerance for seamless scaling. By using Liquibase for schema management, Harness transforms manual database updates into a controlled CI/CD process, enabling faster and safer deployments with clear audit trails and rollback capabilities. The integration supports standard Liquibase operations, detects version conflicts, and ensures secure database connections, making it a robust solution for modernizing database delivery pipelines. This approach empowers developers and DBAs to collaborate effectively, promoting transparency, repeatability, and confidence in database operations across regions.

Harness is hosting the DevSecOps Summit 2025, a virtual event centered on the imperative task of safeguarding AI systems from initial code development to production deployment. As AI rapidly transforms software development, the summit aims to address the emerging challenges of AI security, emphasizing the need for a comprehensive approach that extends beyond traditional security measures. With AI-generated code becoming more prevalent and large language models accessing proprietary data, the summit will bring together industry leaders, security experts, and AI innovators to explore strategies for ensuring robust protection across the entire software development lifecycle. Discussions will cover topics such as securing AI-native applications, integrating AI into development workflows safely, and implementing dynamic security controls to keep pace with evolving threats. Attendees will gain insights into practical frameworks for bridging gaps between development and operations security while exploring real-world examples of AI security challenges and solutions. The event underscores the importance of embracing AI in software development securely and responsibly, offering valuable guidance for security professionals, developers, and leaders in navigating the AI-driven future.

Harness Chaos Engineering's Locust loadgen fault is a strategic tool designed to simulate heavy traffic on applications, allowing developers to identify bottlenecks, validate capacity, and ensure systems recover effectively from load-induced stress. Using Locust, an open-source load testing tool, it generates realistic traffic patterns to test how applications handle extreme conditions, such as Black Friday-level demand. The process involves configuring parameters like target hosts, chaos duration, user count, and spawn rate within a Kubernetes environment, and employs Grafana for real-time monitoring. By incorporating continuous validation through chaos probes and focusing on metrics such as latency, request counts, and overall system health, this approach helps uncover weak points and improve infrastructure resilience. The Locust loadgen fault thus emphasizes the importance of proactive load testing to prevent real-world failures, offering insights crucial for capacity planning and optimization.

The text outlines a series of on-demand videos from a virtual event titled "{unscripted}" focused on the future of AI-powered software delivery. The event covers various topics including the integration of AI in software delivery, the evolution of platforms for enterprise scale, FinOps and AI for cost-efficient cloud spending, and enhancing developer experience through platform engineering. Prominent speakers from major organizations such as Deluxe, USAA, Black Rock, and Lloyds Banking Group provide insights on modernization, security, DevOps, and platform engineering. The event aims to explore how AI and platform strategies can break bottlenecks in software delivery and improve velocity and experience for developers.

Harness has introduced AI-Powered Database Migration Authoring to automate and streamline database changes within the DevOps process, integrating them seamlessly into CI/CD pipelines. This innovation allows developers to describe schema changes in plain language, with Harness generating compliant, production-ready migrations that include rollback and governance features, addressing the long-standing bottleneck of manual database updates. By treating database changes like application code, Harness offers versioning in Git, policy-as-code validation, governed deployment pipelines, and automatic rollback capabilities. This AI-driven automation aims to enhance release speed, governance, and reliability, transforming the database from a hindrance into a catalyst for innovation.

Harness Database DevOps offers a comprehensive solution for managing database changes by integrating automation, governance, and GitOps into a unified platform, positioning itself as a more advanced option compared to traditional migration tools like Flyway and Liquibase. While Flyway is praised for its simplicity and speed, suitable for smaller teams with a focus on forward-only SQL migrations, and Liquibase is noted for its flexibility and rollback capabilities using multiple DSLs, both still require significant manual setup and scripting. Harness, however, is designed with a CI/CD-first approach, embedding governance, rollback automation, and GitOps directly into database delivery pipelines, making it ideal for enterprises needing to scale database delivery with consistency and compliance. Additionally, Harness offers enhanced visibility into deployed changes and employs AI capabilities for generating schema migrations, providing a modern, enterprise-grade solution that reduces manual overhead and accelerates delivery without compromising on best practices.

The rise of AI coding assistants and tools like vibe-coding is leading to an increase in code generation by both professional and "citizen developers," prompting organizations to balance innovation with stability and quality in software delivery. At a recent EngineeringX meetup, platform leaders expressed concern over the pressure to quickly deploy applications created by non-experts, such as senior business members, without adequate risk assessment. The solution lies in developing an Internal Developer Platform that offers self-service boilerplates, automated governance, intelligent verification, and context-aware automation to ensure secure and efficient deployment processes. This approach allows organizations to accommodate rapid development while maintaining enterprise-grade security and compliance standards, ultimately creating a system where the safest way to deploy software is the only way. The emphasis is on using modern platforms to integrate delivery, security, resilience, and governance features seamlessly, avoiding reliance on outdated pipelines that cannot handle the current scale and speed of software development.

Infrastructure as Code Management (IaCM) plays a vital role in enhancing Infrastructure as Code (IaC) practices by providing tools and frameworks that manage the entire lifecycle of infrastructure code, addressing challenges related to security, compliance, and efficiency. As infrastructure complexity increases, IaCM becomes essential for maintaining organization, reducing risks of unauthorized changes, and ensuring consistent management of secrets. It offers features like drift detection, advanced CI/CD integration, and cost management, which help teams standardize, govern, and scale their operations effectively. Platforms like Harness IaCM integrate deeply with tools such as OpenTofu and Terraform, enabling the creation of reusable infrastructure components and automated pipelines, thus facilitating better collaboration and compliance. IaCM improves audit capabilities by tracking changes and fosters seamless integration with existing developer workflows, ultimately allowing organizations to manage infrastructure intelligently and securely at scale.

Harness Code has been redesigned to enhance developer workflows by offering a faster, more intuitive experience that minimizes context switching and optimizes for large monorepos. The new interface includes smarter pull request reviews with inline comments and keyboard shortcuts, a file browser optimized for large repositories, and seamless navigation between branches and commits. This unified design system aligns with the rest of the Harness platform to facilitate a consistent user experience, reducing inefficiencies that can hinder developer velocity. The transition to the new system will be seamless, with no migration required, and is set to become the default by January 2026. Harness aims to further improve the platform with faster repository load times and enhanced AI support for pull request reviews, positioning Harness Code as a modern alternative to other platforms like GitHub or GitLab. The company continues to focus on developer-first features, making Harness Code central to efficient software delivery pipelines.

AI is revolutionizing software development by accelerating code creation, yet it introduces a complex "AI Velocity Paradox" where the speed gains in coding do not necessarily translate into better outcomes downstream. Reports from Harness and DORA reveal that while AI tools enhance developers' productivity and speed up the deployment of code to production, they also lead to increased instability and software vulnerabilities due to insufficient feedback loops and quality assurance processes. The METR study highlights the shift in developers' roles from coding to tasks like specifying requirements, validating AI outputs, and integrating them into existing architectures, which can ultimately take longer and lead to quality concerns. The key to resolving these issues lies in mature DevOps practices, which emphasize robust feedback loops and high-quality internal platforms to manage the influx of AI-generated code effectively. The reports suggest that organizations with moderate automation in their continuous delivery processes achieve greater velocity gains, emphasizing the need for systemic health and investment in advanced DevOps to harness AI's potential fully.

As enterprises grow, managing complex infrastructure through Infrastructure as Code (IaC) alone proves insufficient, necessitating the adoption of Infrastructure as Code Management (IaCM) to transform chaos into consistent, secure operations. IaCM elevates IaC into a strategic framework offering centralized control, compliance, and collaboration, which is crucial for large-scale environments with numerous teams, resources, and regulatory demands. Platforms like Harness IaCM enable organizations to handle IaC sprawl, enforce policies, manage secrets securely, and automate complex workflows, ensuring consistent practices across hybrid and multi-cloud environments. The implementation of IaCM involves assessing current states, choosing appropriate tools, standardizing artifacts, enforcing policies, and fostering a GitOps culture, ultimately allowing enterprises to achieve automated, compliant, and efficient infrastructure management. This approach not only mitigates risks associated with inconsistencies and security gaps but also empowers teams to innovate rapidly and operate confidently.

Harness engineers identified and resolved a critical memory leak issue in their CI/CD delegate service, caused by an improper use of Go's context.Context within worker loops. This leak was revealed through a tight correlation between CPU and memory usage, indicating that garbage collection struggled to manage memory efficiently. The problem arose because reassigning context variables in a loop created a chain of contexts that the garbage collector could not free, leading to exponential memory retention across thousands of goroutines. By reducing the test environment to a single worker, the engineers pinpointed that each task's context was improperly linked to the previous one, preventing cleanup. The solution involved altering the variable scoping to prevent these chains from forming, highlighting the importance of careful variable management within loops. The resolution of this issue also paved the way for further optimizations, as a profiler later uncovered an additional goroutine leak to be addressed in the future.

Cloud waste in many organizations is primarily an organizational issue rather than a technical one, and the adoption of FinOps practices can significantly enhance the synergy between finance and engineering teams by linking cloud spending to business outcomes rather than blame. By fostering a collaborative environment where both finance and engineering speak a common language focused on business value, companies can achieve over 30% savings and improve alignment as they scale. FinOps serves as a strategic framework that transforms cloud expenses into investments, promoting transparency, shared accountability, and trust. Harness exemplifies this approach through its Cloud Cost Management (CCM) platform, which integrates real-time visibility and context-aware cost management into existing workflows, thereby reducing friction and establishing cloud costs as a lever for business growth rather than a burden. The key to a successful FinOps culture lies in translating the conversation from mere cost savings to maximizing business value, encouraging a mindset shift from "You're not overspending, you're under-saving," which in turn drives long-term efficiency and value creation.

Hacktoberfest is an annual event organized by DigitalOcean, encouraging developers worldwide to contribute to open-source projects, with a focus this year on chaos engineering through the LitmusChaos platform. Harness and the LitmusChaos community are actively participating by welcoming new contributors, offering mentorship, and recognizing valuable contributions. Participants can engage by addressing curated issues, attending office hours, and receiving guidance from maintainers to support the development of Litmus 4.0, a cloud-native chaos engineering platform used to test system resilience. This initiative allows individuals to explore open-source contributions, grow the chaos engineering ecosystem, and contribute to building a more resilient cloud-native future.