sre

一篇探讨使用 AI 代理进行软件系统监控的实践者博客文章，以 Anthropic 内部的 "superfan" Claude 监控实例为引，论证了基于代理的变更点监控正成为运维领域的重要方向。

Google发布了《Site Reliability Engineering》第二版，距2016年首版十年之际，全面更新了SLO、可观测性、自动化、容量管理等六大实践领域，并深入探讨了AI与SRE的双向关系——AI既是运维对象也是运维工具。

DevOps Advisor is an AI agent that diagnoses delivery failures and tailors CI/CD, infrastructure, and reliability guidance to a team's needs, using DORA metrics and SRE practices with persistent stack memory across sessions.

KloudMate 2.0 在 Product Hunt 上线，将三年前的 AWS 无服务器监控工具升级为全栈 AI 可观测性与 Agentic SRE-Ops 平台，可自动识别、诊断并解决超过 70% 的事件。

A GitHub repository featuring 151 real DevOps/SRE interview records from 85 companies. Questions were recorded verbatim by candidates and cover Kubernetes, Docker, Terraform, AWS, CI/CD, and more, making it convenient for job seekers to prepare by company or topic.

This article discusses the phenomenon of software systems failing due to resource saturation as they scale, drawing an analogy with the limits of biological systems, and introduces the impact of physical resource (CPU, memory, disk, network) saturation and virtual limits on system reliability.

DrDroid's Alert Grouping feature helps reduce alert noise, improving monitoring efficiency.

Discusses how cascading recovery and feedback loops can cause metastable failures in distributed systems, preventing proper recovery.

Uber is dropping PagerDuty after 12+ years due to reliability issues in alerting, highlighting the importance of reliability for infrastructure tools.

Antimetal is an autonomous system using AI agents to manage production software by building a world model of the environment, preventing issues, triaging alerts, and shipping fixes.

SaZabi is building an AI observability system that uses logs as the source of truth to automate debugging and issue resolution, aiming to bridge the gap between automated code and manual debugging.

The author shares a two-year experience deploying AI agents for investigating production incidents across team boundaries, highlighting that while the technical implementation was straightforward, the organizational politics posed the real challenge.

Asuka Zheng argues that the 'running out of training data' panic is misplaced; the real scarcity is a lack of imagination in collecting diverse, long-horizon data, illustrated by her SRE replacement project and broader research trends.

At Gleam Gathering 2026, Rob Durst shared how he rewrote a YAML-to-Terraform configuration tool as a compiler using Gleam, and how he came to appreciate the power of type-driven design and the decoder pattern.

Cognition announces Devin Auto-Triage, an AI agent designed for on-call engineers that monitors incidents and provides context and automated responses via Slack.

A tech commentator advises junior developers to pivot into DevOps/SRE, arguing that systems knowledge is increasingly critical because AI can write code but cannot manage production systems.

A GitHub repository collecting real DevOps, SRE, and Cloud engineering interview questions from 2025-2026, organized by company with 151 write-ups across 85 companies.

The 1Password engineering team uses OpenAI Codex to drastically shorten the cycle from idea to production, and by cultivating AppSec skills and building internal AI SRE agents, they accelerate feature releases and customer feedback loops.