Your infrastructure can explain itself, heal safely, and prove it recovered.<br> 让基础设施自己解释故障、安全完成修复，并证明它真的恢复了。
How to install flawless?
william-lu-stack/flawless
$git clone https://github.com/william-lu-stack/flawlessInstalls into the current project.
›Prefer a prompt? Paste this to your agent
Install flawless by running `git clone https://github.com/william-lu-stack/flawless`, then use it for the current task and follow its documentation at https://github.com/william-lu-stack/flawless.
README.md
| 1 | # Flawless |
| 2 | |
| 3 | [](#quick-start) |
| 4 | [](#development) |
| 5 | [](#kubernetes-deployment) |
| 6 | [](#docker) |
| 7 | [](#langfuse) |
| 8 | [](#license) |
| 9 | |
| 10 | **Your infrastructure can explain itself, heal safely, and prove it recovered.**<br> |
| 11 | **让基础设施自己解释故障、安全完成修复，并证明它真的恢复了。** |
| 12 | |
| 13 | **Flawless** is an AI-native SRE control plane for Kubernetes and cloud infrastructure. It connects alerts, evidence, topology, human approval, controlled remediation, and recovery verification in one auditable AgenticOps loop. |
| 14 | |
| 15 | 它不是另一个只会给建议的运维聊天框。Flawless 将“发现问题、收集证据、生成预演、人工授权、执行变更、恢复验证、经验沉淀”连接成一个可审计闭环。 |
| 16 | |
| 17 | Maintained by the **Flawless Contributors** community. |
| 18 | |
| 19 | Current release: **3.2.2**. |
| 20 | |
| 21 | Release 3.2 adds persistent remediation lineage: every failed strategy, action, |
| 22 | verification result, and replacement plan stays linked across operator-approved |
| 23 | follow-up jobs. The effectiveness ledger is persisted on the runtime volume so |
| 24 | model comparisons and recovery records survive Pod restarts. |
| 25 | |
| 26 | ## The AgenticOps Loop |
| 27 | |
| 28 | `discover → diagnose → preview → approve → execute → verify → learn` |
| 29 | |
| 30 | - **Evidence first**: connect alerts, events, logs, metrics, topology, runbooks, and recent changes. |
| 31 | - **Guarded action**: keep RBAC, policy, dry-run, human approval, and audit outside the model boundary. |
| 32 | - **Verified recovery**: test the original symptom after execution instead of treating a successful command as success. |
| 33 | |
| 34 | ## Why This Exists |
| 35 | |
| 36 | Modern cloud systems fail in ways that are hard to reason about from a single log line: |
| 37 | |
| 38 | - a Pod restart can hide a PVC, image, scheduling, network, quota, or rollout issue; |
| 39 | - a small workload change can affect services, data pipelines, middleware, and downstream users; |
| 40 | - repeated human firefighting leaves valuable operational knowledge outside the platform; |
| 41 | - model output is useful only when it is constrained by evidence, policy, permissions, and rollback. |
| 42 | |
| 43 | Flawless is built as an SRE control plane. It uses a model as a planner and explainer, but the platform keeps the execution boundary: RBAC, action catalog, dry-run, approval, audit, and recovery verification. |
| 44 | |
| 45 | ## Core Features |
| 46 | |
| 47 | - **SRE Chat**: ChatGPT-style operations console with cluster, namespace, workload, and risk context. |
| 48 | - **Inspection Queue**: scheduled or manual scans across Rancher/Kubernetes scopes with severity ranking. |
| 49 | - **Controlled Remediation**: evidence collection, change preview, human approval, execution, post-change verification, and evidence-driven replanning that remembers failed strategies across follow-up jobs. |
| 50 | - **Topology Impact**: 2D/3D topology, CMDB-style dependencies, eBPF/data-flow adapters, blast-radius analysis. |
| 51 | - **Release Governance**: SLO, error budget, canary/risk gate, emergency fix path, and release audit chain. |
| 52 | - **Skills Library**: portable operation skills that encode expert knowledge and can be reused by other agents. |
| 53 | - **Knowledge Base**: upload text, Markdown, PDF, Word, Excel, logs, YAML, and runbooks for operations RAG. |
| 54 | - **Model Lab**: configure multiple OpenAI-compatible or OAuth-protected model gateways and compare outcomes. |
| 55 | - **Measurable Outcomes**: persistent remediation lineage, changed-resource history, recovery evidence, and model effectiveness comparisons. |
| 56 | - **Observability**: Prometheus metrics, Loki logs, Tempo traces, Grafana links, and optional Langfuse traces. |
| 57 | - **Extensible Infrastructure**: adapters for Kubernetes, Rancher, databases, virtual machines, storage, and middleware. |
| 58 | |
| 59 | ## Architecture |
| 60 | |
| 61 | ```text |
| 62 | Frontend Console |
| 63 | ├─ SRE Chat / Inspection / Topology / Release |
Related
6 picksType
- mcp-server-circleciA Model Context Protocol (MCP) server implementation for CircleCI, enabling natural language interactions with CircleCI functionality through MCP-enabled clientsJul 2026
- mcp-for-argocdArgo CD MCP ServerJul 2026
- mcp-server-kubernetesMCP server for interacting with Kubernetes clusters via kubectlJul 2026
- hookdeck-cliCLI for Hookdeck: forward webhooks to localhost (ngrok alternative), manage and query Event Gateway resources (sources, connections, destinations, events), run the MCP server for AI agents. Free for dev.Jul 2026
- heroku-mcp-serverHeroku Platform MCP ServerJul 2026
- mcp-server-azure-devopsAzure DevOps reference server for the Model Context Protocol (MCP)Jul 2026
