Early alphaWorks today — APIs and config can still change between releases.

Terraplane

Terraform belongs in the pull request — and the runner belongs inside your network.

Comment-driven plan and apply, without a sprawling platform. An orchestrator that talks to your SCM, agents that run Terraform where your credentials already live.

Terraplane architectureAn SCM provider sends webhooks to an orchestrator. Separate agents in prod, staging, and dev networks each poll the orchestrator over HTTP.SCM providerOrchestratorwebhooksjobs · locksPR feedbacknetwork · prodagent-prodplan · applynetwork · stagingagent-stagingplan · applynetwork · devagent-devplan · applyEach agent polls out from its own network

Why it’s different

You shouldn’t need a full Terraform platform to get PR automation that reaches private infrastructure. And you shouldn’t have to run plan and apply on the same box that faces the internet.

Orchestrator

Webhooks, job state, locks, PR feedback. Lives where your SCM can reach it — and nowhere near your private control planes.

Agents

Workers that poll the orchestrator over HTTP, claim jobs, and run Terraform where credentials and network access already exist.

Two roles, a YAML file, and PR comments. Environments and stacks route work to the right agent — multi-account without stuffing every secret into one box.

Small surface area

No workspaces UI to adopt. No registry to stand up. Commentterraplane plan, get a plan. Apply when you’re ready. Unlock when you need to.

  1. PR comment hits the orchestrator
  2. Stack resolves to a named agent
  3. Agent plans or applies in-place
  4. Results post back onto the PR

Ready to kick the tires?

Docs are thin but honest. The code is MIT. APIs and config can still change — that’s the deal with early alpha.