Neckbeard – a DevOps/platform/SRE plugin for your agent

Neckbeard is a DevOps platform and SRE plugin that simplifies the process of turning applications into reviewable cloud infrastructure with cost estimates. It supports AWS, GCP, or Azure, GitHub Actions or GitLab CI, serverless containers or Kubernetes, and is driven by a bundled catalog of tested OpenTofu modules. The agent reads the app, explains the planning, rendering, and validation, and ensures the infrastructure is deterministic code without inventing resources.

To use Neckbeard, simply run "/neckbeard:analyze" in the app's repository directory, which installs a small CLI required for the plugin. The CLI, a sha256-verified release binary, or go install when Go is present, is all that's needed for setup. Cost estimates and static validation use tools such as Infracost, OpenTofu, and Checkov, and "/neckbeard:doctor" provides information about missing tools and their capabilities.

Neckbeard is designed for coding agents that excel at app code but struggle with freehand infrastructure. It inverts the roles of the coding agent by providing named findings instead of forcing infrastructure onto the nearest module. The reference app, a smallest tier HTTP service + worker + cron + Postgres, is used for pricing via Infracost, with three verdicts: baseline-while-idle costs, explicit usage assumptions, and low/expected/high scenarios. Unsupported findings are explicitly noted, and the plugin reports findings in three verdicts.

The plugin's product today focuses on onboarding, providing reviewable infrastructure, pipelines, cost estimates, and a bootstrap runbook. The deployment lifecycle (neckbeard release ...), is currently experimental and tracked gate by gate in the launch plan. Static validation should not be treated as deployment proof. Landing zones (org structure, guardrails) are designed, not built, and the repository root contains the plugin's operating manual, commands, and the Go CLI with the module catalog.