#sre

A Kubernetes lab that reproduces real production incidents on a live, instrumented microservice stack — for testing root-cause-analysis tools and agents.

The self-improving SRE agent

Model-agnostic DevOps AI agent harness: policy, approvals, audit and rollback enforced outside the LLM. Works with Claude Code, OpenCode, Cursor, Copilot and any MCP client.

Multi-cluster Kubernetes dashboard — the whole UI baked into one Go binary. Resource lists stream binary MessagePack deltas over WebSocket; no polling, no pagination, virtualised end-to-end. Inline scale, restart, rollback, exec, logs, port-forward, YAML edit. Prometheus + metrics-server. Docker, Helm or raw manifests.

Voice on-call agent for AWS: fixes the 3 AM page on your spoken word, verifies it, and ends the night with a pull request. Interrupt the read-back and the fix is withdrawn.

Turn Claude Code into a team that has to prove it. 44 skills, 8 review agents, a validation router and a knowledge layer where nothing is claimed without a check that could have failed.

Claude Code scaffold for building AI platforms securely — agent/LLM security, Kubernetes, SRE, observability, identity, and supply chain, grounded in published framework canon (OWASP, NIST, CIS, SLSA). Data-science lanes included.

Autonomous Claude Code CI agent in a rootless Podman sandbox — GitLab CI & GitHub Actions, two personalities (read-write Agent, read-only Advisor), OpenTelemetry audit to Elastic, Helm chart for AKS/OpenShift.

An autonomous, LLM-powered SRE agent that takes production incidents from alert to fix/pulls metrics and logs, diagnoses the root cause, runs remediation, and writes the incident report.

Demonstrating an attribution gap in LLM inference telemetry: per-request tracing records the symptom of KV-cache contention and not the cause

Read-only AI SRE agent for Kubernetes: investigates, health-checks and reports via MCP and a guarded kubectl

AI AgenticSafe - stop your AI agent from breaking production at 3 a.m. 8 battle-tested doctrines, 3 real post-mortems, zero dependencies. By Yacine Mahboub, Founder of WEVIA.

Local-first incident triage agent with typed tools, weighted evidence, and structured root-cause reports.

面向多租户的 Agent Harness / Runtime 参考实现 —— 凭证代管、权限控制、多租户隔离、可回归 eval

Free 24/7 AI operations checklists, runbooks and escalation templates for autonomous AI agents and servers.

Evidence-based Kubernetes AIOps incident analysis platform with Prometheus, Loki, rule correlation, LLM-assisted RCA, FastAPI, Docker and Helm Visibility: Public

AI-powered AWS incident triage agent that investigates ECS and CloudWatch evidence, identifies failure patterns, and generates structured RCAs with read-only AWS access.

LLM-assisted self-healing agent for Kubernetes on GCP (GKE Autopilot + self-managed RKE2). Claude diagnoses failing workloads from live telemetry; deterministic guardrails decide whether to roll back, restart, raise memory, or escalate. Terraform, kustomize, GitHub Actions.

Write the agent, not the chat integration. Kubernetes-native broker between Slack and streaming A2A AI agents (kagent, Google AX): threaded sessions, alert investigations, restart recovery.

Autonomous AI Cloud Agent 2026: Auto-Fix Code & PRs with Multi-LLM Sandbox

AI-powered DevOps & SRE automation platform

Autonomous local SRE agent — ingests logs, detects anomalies, and routes all remediation actions through a policy webhook before execution. TDD-built, fail-closed, no cloud dependency.

A signed, priced, replayable bus for agent handovers. Signals in, agents decide or investigate behind budgets and closed tool lists, and what survives reaches a person. Two live deployments: production alerts, and one operator's own work signals.

Git-native forensic archive of AI provider service status. Polls OpenAI, Anthropic, and Google Cloud every 5 minutes and commits only when state changes — the git history is the time-series database.

Claude Code mod: guardrails for kubectl, helm, terraform and Kubernetes Secrets — named targets, script checks, plan gate, audit, and Secret values the model never sees

Graph-grounded incident root cause, containment and on-call agent on FalkorDB. Every answer comes with its path. Graph Hacks 2026, Track 01.

Autonomous Incident Remediation Operator for Kubernetes workloads

Agent-drivable, cross-platform harness that tests PowerShell & Python scripts in throwaway VMs over SSH and asserts deterministic pass/fail. Validated live on Hyper-V + WSL. CI included.

A read-only MCP server exposing Slurm scheduler state to agents — allowlist enforced in code, not requested in a prompt, with progressive disclosure.

Evidence-driven Kubernetes root cause analysis with LangGraph-orchestrated, bounded read-only investigation and deterministic diagnosis.

Incident management and KTLO operations platform: .NET 8 DDD API, Azure Functions, Service Bus, SignalR, Python analytics with Azure OpenAI, React console, Bicep, GitHub Actions.