Awesome Agentic Devops

Awesome Agentic DevOps

[](LICENSE) [](https://github.com/DevOpsAIguru123/awesome-agentic-devops/actions/workflows/validate.yml) [](CONTRIBUTING.md)

Awesome Agentic DevOps is a curated map for DevOps, Cloud, SRE, and Platform Engineering agents across Claude Code, Gemini ADK, OpenAI Agents SDK, and MCP.

This repository evaluates which agents are safe, useful, and production-adjacent for infrastructure automation. Every link is re-audited by scheduled CI for reachability, archival, and freshness.

Contents

• [Recently added](#recently-added)
• [Why this exists](#why-this-exists)
• [Who it is for](#who-it-is-for)
• [Safety-first disclaimer](#safety-first-disclaimer)
• [Evaluation labels](#evaluation-labels)
• [How entries are scored](#how-entries-are-scored)
• [Categories](#categories)
• [Top picks by use case](#top-picks-by-use-case)
• [Curated catalog](#curated-catalog)
• [How to contribute](#how-to-contribute)

Recently added

| Date | Entry | Category | | --- | --- | --- | | 2026-07-06 | vantage-sh/vantage-mcp-server | FinOps / cloud cost | | 2026-07-05 | hashicorp/vault-mcp-server | IaC / secrets management | | 2026-07-05 | cloudflare/mcp-server-cloudflare | Cloud / edge | | 2026-07-05 | CircleCI-Public/mcp-server-circleci | CI/CD | | 2026-07-05 | mongodb-js/mongodb-mcp-server | Data platform | | 2026-07-05 | backstage/backstage (mcp-actions-backend) | Platform toolkits |

Why this exists

Most agent lists stop at discovery. Infrastructure teams need more than discovery: they need to know whether an agent can touch production, whether it has approval gates, whether it preserves evidence, and whether it can be rebuilt as a safe reference workflow.

This repo is intentionally operator-grade. Each entry is scored by real infrastructure usefulness, operational risk, human approval gates, tracing/evidence, and maturity.

Who it is for

• DevOps and platform engineers evaluating AI automation.
• SREs designing incident-response copilots.
• Cloud engineers comparing agent frameworks.
• Security reviewers assessing infrastructure-agent risk.
• Builders creating portfolio-grade reference agents.
• Interview candidates who want a practical DevOps AI artifact.

Safety-first disclaimer

These agents may touch infrastructure. Prefer read-only or proposal mode first. Require human approval before write actions. Never expose secrets to model context. Use least-privilege credentials, dry runs, Terraform plans before applies, and CI checks before merge.

Evaluation labels

| Label | Meaning | | --- | --- | | 🟢 | production-adjacent OSS | | 🟡 | useful prototype | | 🔵 | MCP/server integration | | 🛡️ | has approval/safety controls | | 📊 | has tracing/evidence/evals | | ⚠️ | write-capable; review before use |

Labels are shorthand for structured fields recorded on every entry in [data/repos.yaml](data/repos.yaml): ⚠️ maps to action_level: write-capable, 🛡️ to human_approval: true, 📊 to tracing or eval evidence, and 🟢/🟡 to maturity.

How entries are scored

Every entry records five verifiable dimensions. They are assessed by reading each project's documentation and exposed tool surface, not its marketing.

| Field | Question it answers | How it is verified | | --- | --- | --- | | action_level | Can it touch production? | Read the project's exposed tool list. Only query tools (get, list, search) means read-only. Suggesting changes for a human to execute means proposal. Any tool that mutates real state (create, deploy, delete, sync) means write-capable and the entry gets ⚠️. | | human_approval | Does a human gate write actions? | Look for gates in the server (write tools disabled by default, dry-run modes, confirmation flags) or in the client (permission prompts). Server-side gates are stronger because they hold no matter which client connects. | | evidence_tracing | Can you prove what it did afterward? | Check for audit logs, OpenTelemetry support, structured run records, or evaluation output. Scored yes, partial, or none. | | maturity | Is it stable enough to depend on? | Observable signals only: vendor support status, API stability (GA versus alpha), and recent activity. The audit script checks freshness and archived status automatically. | | risk_notes | What is the blast radius if it goes wrong? | Combines the above with what the tool connects to. Write-capable identity tooling is treated very differently from a read-only diagram generator. |

Example: reading one entry

- name: PagerDuty/pagerduty-mcp-server
  action_level: write-capable    # tools can create incidents and schedule overrides
  human_approval: true           # write tools ship disabled by default
  evidence_tracing: partial      # logs exist, but no structured trace guarantee
  maturity: production-adjacent  # official vendor server
  risk_notes: "Keep write tools disabled by default and require approval for changes."

Read as: safe to connect for incident context today, but flip on its write tools only after you have decided who approves an agent-created incident.

What is automated and what is judgment

Link reachability, archived status, and freshness are checked automatically by [scripts/auditgithubrepos.py](scripts/auditgithubrepos.py) and enforced in CI. The safety scores themselves are curator judgment from reviewing each project's docs and tool surface at the time of entry. Verify against your own environment before connecting anything to real infrastructure — [templates/agent-scorecard.md](templates/agent-scorecard.md) is the full per-project checklist used for deep review.

Categories

1. Official cloud MCP servers and agent toolkits
2. Official DevOps and source-control MCP servers
3. Official CI/CD and GitOps MCP servers
4. Official security, code-quality, and agent-security resources
5. Official IaC MCP servers
6. Official SRE and observability MCP servers
7. Official agent skills and agent frameworks
8. Official platform agent toolkits
9. Official MCP SDKs, references, registries, and governance platforms
10. Official diagramming and architecture MCP tools
11. Official data platform MCP servers
12. Official FinOps and cloud-cost MCP servers
13. Community discovery and skill references

Top picks by use case

| Use case | Start with | Why | | --- | --- | --- | | AWS agentic cloud automation | aws/agent-toolkit-for-aws | Official AWS-supported MCP, skills, plugins, IAM-aware controls, CloudWatch metrics, and CloudTrail auditability. | | Azure cloud automation | microsoft/mcp and microsoft/azure-skills | Official Microsoft MCP and skills/plugin sources for Azure resource workflows. | | Google Cloud automation | google/mcp, googleapis/gcloud-mcp, and google/skills | Official Google MCP and skills sources for GCP, Cloud Run, GKE, observability, and storage workflows. | | Source-control DevOps | github/github-mcp-server, GitLab MCP server, and atlassian/atlassian-mcp-server | Official MCP tool surfaces for repos, issues, PRs, Jira, Bitbucket, and related delivery workflows. | | Terraform and IaC | hashicorp/terraform-mcp-server and Pulumi MCP Server | Official IaC MCP sources for Terraform Registry/HCP Terraform and Pulumi Cloud automation. | | SRE incident response | grafana/mcp-grafana, datadog-labs/mcp-server, and PagerDuty/pagerduty-mcp-server | Official observability and incident-management MCPs for metrics, logs, traces, alerts, incidents, and on-call context. | | FinOps and cloud cost | vantage-sh/vantage-mcp-server | Official Vantage MCP server for cloud spend analysis, budgets, anomalies, reports, and provider-resource cost context across Vantage-connected providers. | | Security and code quality | SonarSource/sonarqube-mcp-server, okta/okta-mcp-server, Snyk Studio MCP docs, and Wiz WIN MCP Server docs | Official MCPs and security resources for code quality, application security, identity-aware workflows, agent security, and cloud-security posture. | | CI/CD and GitOps | jenkinsci/mcp-server-plugin and argoproj-labs/mcp-for-argocd | Official Jenkins and Argo Project resources for pipeline, build, deployment, and GitOps workflows. | | MCP development and governance | modelcontextprotocol/python-sdk, modelcontextprotocol/typescript-sdk, modelcontextprotocol/registry, and Docker MCP Catalog and Toolkit | Official SDKs, registry, and Docker governance surfaces for building, packaging, and controlling DevOps MCP servers. | | Agent frameworks and templates | google/adk-python and GoogleCloudPlatform/agent-starter-pack | Official Google agent framework and production templates with CI/CD, evaluation, and observability. |

Curated catalog

The source of truth is [data/repos.yaml](data/repos.yaml). The list below is a readable index of the current official and community research-backed entries.

Official Cloud MCP Servers and Agent Toolkits

| Repo | Labels | Operator note | | --- | --- | --- | | aws/agent-toolkit-for-aws | 🟢 🔵 🛡️ 📊 ⚠️ | Official AWS-supported toolkit that bundles MCP server configuration, skills, plugins, and DevSecOps agent workflows. | | awslabs/mcp | 🟢 🔵 🛡️ ⚠️ | Official AWS Labs MCP server collection; useful legacy/source reference while AWS transitions capabilities into Agent Toolkit. | | microsoft/mcp | 🟢 🔵 🛡️ ⚠️ | Official Microsoft MCP catalog, including Azure cloud and infrastructure MCP references. | | google/mcp | 🟢 🔵 🛡️ ⚠️ | Official Google MCP repository listing managed and open-source MCP servers for Google and Google Cloud. | | googleapis/gcloud-mcp | 🟢 🔵 🛡️ ⚠️ | Official Google API repository for gcloud, observability, storage, and backup/disaster-recovery MCP servers. | | GoogleCloudPlatform/cloud-run-mcp | 🟢 🔵 🛡️ ⚠️ | Official Google Cloud Platform MCP server for deploying apps to Cloud Run. | | cloudflare/mcp-server-cloudflare | 🟢 🔵 🛡️ ⚠️ | Official Cloudflare MCP servers for account, Workers, and edge configuration workflows. | | vercel/vercel-mcp-overview | 🟢 🔵 🛡️ ⚠️ | Official public overview of Vercel's hosted MCP server for project and deployment context. |

Official DevOps MCP Servers

| Repo | Labels | Operator note | | --- | --- | --- | | microsoft/azure-devops-mcp | 🟢 🔵 🛡️ ⚠️ | Official Azure DevOps MCP server with remote-first onboarding and local server option. | | github/github-mcp-server | 🟢 🔵 🛡️ ⚠️ | Official GitHub MCP server for repository, issue, pull request, code, and workflow automation. | | GitLab MCP server | 🟢 🔵 🛡️ ⚠️ | Official GitLab MCP server is documented as a GitLab-hosted endpoint rather than a standalone GitHub repo. | | atlassian/atlassian-mcp-server | 🟢 🔵 🛡️ ⚠️ | Official Atlassian Rovo MCP server for Jira, Confluence, Jira Service Management, Bitbucket, and Compass. | | docker/mcp-registry | 🟢 🔵 🛡️ 📊 | Official Docker MCP registry and catalog source for verified containerized MCP servers. | | kubernetes-sigs/mcp-lifecycle-operator | 🟡 🔵 🛡️ ⚠️ | Official Kubernetes SIG operator for declaratively deploying and rolling out MCP servers, not a general kubectl MCP server. |

Official Security and Code-Quality MCP Servers

| Repo | Labels | Operator note | | --- | --- | --- | | SonarSource/sonarqube-mcp-server | 🟢 🔵 🛡️ | Official SonarQube MCP server for code quality and security insights in AI agents. | | okta/okta-mcp-server | 🟢 🔵 🛡️ ⚠️ | Official Okta self-hosted MCP server for connecting agents to Okta identity workflows. | | Snyk Studio MCP docs | 🟢 🔵 🛡️ 📊 | Official Snyk documentation for Snyk Studio agentic security workflows and Snyk MCP Server usage. | | snyk/agent-scan | 🟢 🛡️ 📊 | Official Snyk security scanner for AI agents, MCP servers, and agent skills. | | Wiz WIN MCP Server docs | 🟢 🔵 🛡️ 📊 | Official Wiz documentation for the WIN MCP server, adding CNAPP and cloud-security coverage. |

Official CI/CD and GitOps MCP Servers

| Repo | Labels | Operator note | | --- | --- | --- | | jenkinsci/mcp-server-plugin | 🟢 🔵 🛡️ ⚠️ | Official Jenkins plugin that enables Jenkins to act as an MCP server for LLM-powered clients. | | argoproj-labs/mcp-for-argocd | 🟡 🔵 🛡️ ⚠️ | Argo Project Labs MCP server implementation for Argo CD, filling the GitOps/CD gap in the catalog. | | CircleCI-Public/mcp-server-circleci | 🟢 🔵 🛡️ 📊 | Official CircleCI MCP server for build failure logs, pipeline status, flaky test detection, and usage analysis. |

Official MCP SDKs, Reference Implementations, Registries, and Governance Platforms

| Repo | Labels | Operator note | | --- | --- | --- | | Docker MCP Catalog and Toolkit docs | 🟢 🔵 🛡️ ⚠️ | Official Docker documentation for MCP Toolkit, MCP Catalog, profiles, CLI, and MCP Gateway. | | Docker MCP Gateway docs | 🟢 🔵 🛡️ ⚠️ | Official Docker MCP Gateway documentation for secure, centralized orchestration of containerized MCP tools. | | modelcontextprotocol/servers | 🟢 🔵 🛡️ | Official MCP reference server implementations and ecosystem pointers. | | [modelcontextprotocol/py

…

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: DevOpsAIguru123
• Source: DevOpsAIguru123/awesome-agentic-devops
• License: MIT

Install and usage instructions live in the source repository linked above.