Dan Miller retweeted
let me introduce you to @tembo
1
2
9
481
i might have the most over engineered solution to tracking macros...
2
4
96
the best part is that if it doesnt work, all of these repos for these pieces are included in the same tembo project. i literally just swap it to astra and say fix it. then it works next time
1
1
26
bring your own roles -- give it whatever permission you feel comfortable with. tembo can now swap between accounts on demand
Tembo now natively supports @AWS. Give your cloud agents access to your AWS organization! No static credentials required, define your own permissions and swap accounts on demand.
1
3
143
previously we had tried using an AWS MCP for this, but honestly it's much easier to just use the aws cli. tembo sets up the AWS profile by assuming your predefined iam role, and then just calls whatever cli command you want
1
24
Dan Miller retweeted
i can play minecraft in 60fps while waiting for my cloud agents to finish
6
2
28
2,882
I finally fixed the AWS integration for tembo. This is really nice because now the agent can dynamically assume roles in different accounts on demand. Combine this with a scheduled automation, and now we can send an agent to automatically analyze cost explorer, find orphaned resources, etc across the org. It only took a few minutes to setup, but automations like these are life savers
4
111
building sandbox infra to replace local dev means expecting the user to break their sandbox. they can and should be able to exhaust mem, corrupt the disk, or do whatever they want. it's your job to make it recoverable
3
58
it was only matter of time before local dev was beat by the cloud. hard to find a reason to do anything locally anymore
holy shit.. cloud coding agents are real frontend? iterate in real time. the vm is right there need nested virtualization? a database? your entire stack? it's linux. run whatever you want it's basically local development, except it's all in the cloud no one at tembo uses a local dev environment anymore
1
5
133
I’m on call this week and last night a cpu alert comes in while I’m watching a movie. I handled the whole thing from my phone 🧵
1
3
63
We use incident.io for alerts, which we have configured for a few slack channels. When a message hits one of those channels, Tembo reacts automatically. Tembo kicks off a custom skill we built with scoped access to AWS, our database, and Tailscale. It diagnoses the alerts, categorizes it, groups related or duplicate messages, and recommends the next steps.
1
1
55
So last night a CPU alert comes in. Tembo kicks off the investigation. It immediately gives me a table of the customers running hot in our org. Turns out it was legit, heavy usage. So I Tembo to bump the ASG limit. It does, opens a PR, my automation pipelines apply it. Alert resolved. Entirely from my phone
1
22
kudos to the team over at masterpoint. really talented team over there so much easier to stay ahead of the terralith than to dig yourself out of one!
The faster a team ships, the faster their Terraform setup falls apart. And Cursor is one of the fastest-shipping companies I've seen. So fast, in fact, they had to move off Terraform Cloud entirely. Years of rapid growth and copy-paste scaling had left the company in an untenable situation: - Their entire AWS infrastructure was managed as a single Terraform workspace. - Creating new workspaces required manual admin configuration. - Diffs were so noisy, nobody was reading them. - Plans took over 10 minutes. Ironically, a company building AI tools to help developers scale and move faster was being held back and slowed down by its own infrastructure. Until they started working with Masterpoint, who decomposed Cursor's Terraform monolith into focused workspaces and migrated them to Spacelift and OpenTofu. As a result: - Plans decreased to 2 minutes. - Weekly infrastructure PRs went from 194 to over 500. - Blast radius dropped from 54 resources/run to 5. - Workspaces grew from 21 to 141, all self-service. Cursor finally has infrastructure that moves as fast as its teams. You can read Masterpoint's full case study here: masterpoint.io/case-studies/…
1
77
this has been a game changer for us. Giving an agent private network access, locked down with a tailscale ACL has been so helpful
Tembo now supports @Tailscale. Give Tembo VMs and MCP servers scoped access to devices on your tailnet. docs.tembo.io/integrations/t…
49
splitting terraform into the smallest possible lifecycle groups might be the single best thing i've done for my stress levels
39
Dan Miller retweeted
Orgs rationing $200 token plans are the same ones paying full salaries for teams to stay perennially blocked. When your developers say they're running out of tokens, it's not a cost problem. That's what winning looks like.
4
2
12
1,626