Coordinate Terraform and GitOps work with OpenClaw
A developer used multiple OpenClaw agents to handle infrastructure, coding, and GitOps tasks while directing the work in Telegram.
- What they did
- They spun up a set of OpenClaw workers, giving each one a separate Google Workspace address, Git account, and EC2 machine. Those agents were allowed to take AWS roles inside a development account, then used a Redis pub/sub channel they built themselves to divide and sync Terraform repo work, programming tasks, reviews, merges, and a CD/GitOps pipeline. The human stayed mostly in Telegram, assigning bigger pieces of work to the agent group.
- How it went
- They say the system got substantial work done overnight and was far enough along by morning to resemble a small crew of "10x engineers." After the model switch, the agents stopped completing complicated multi-step jobs consistently. They also said Anthropic Max had become too expensive for this setup.
- Tips
- This approach works best when agents have distinct identities, clear permissions, and a way to coordinate state. Be careful about changing models or costs midstream, because both reliability and economics can change quickly.
Try it yourself with OpenClaw
You are a set of OpenClaw agents working as one engineering team. Give each agent its own [email address], [Git account], and [compute instance]. Let them use [AWS role] access only in the [development account], and have them coordinate through [Redis pub/sub]. Use this team to complete [Terraform work], [code changes], [reviews], [merges], and [CD/GitOps tasks]. Before any action that would send something out, merge code, deploy, delete resources, submit a change, or otherwise make an irreversible update, pause and ask me to approve it first. After each task, tell me what was completed, what is still pending, and any risks or blockers.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Sort failed CI runs and repair flaky tests
A QA engineer used OpenClaw to read failed builds, label the cause, alert teammates, and reduce test flakiness.
Track down broken builds and draft the fix
OpenClaw was used inside OpenAI to diagnose failing builds, link them to incidents, and sometimes prepare and merge repairs.
Route Linear tickets into the right repo and PR
They used OpenClaw and Linear as a control layer to send tasks to Cyrus, which picked the repo, chose a model, and opened pull requests.
Build and deploy a simple idle clicker game
A developer had an agent make a clicker game from a chat, then push it to GitHub and deploy it to Vercel.
Deploy safely and scan exposed ports from Telegram
A security professional used an OpenClaw agent to handle deploys and port checks while keeping production changes gated by approval.
Have coding agents test your onboarding flow
Run coding agents through your product’s onboarding and get a scorecard, fix list, and session traces.