What if your infra could fix itself before you even notice?
Chang Log 3 June 2026
This week, we’re highlighting how engineers are actually using Stakpak in production, from running /init to uncover risks, to step by step tutorials that turn hours of debugging into minutes.
/init
/init scans your infrastructure (cloud accounts, clusters, and configs), detects reachable systems and integrations, analyzes risk areas like permissions, stateful services, drift, and exposure, generates recommended schedules and guardrails, and creates and Apps.md
And then it asks if you want to set up Autopilot for continuous monitoring.
Just open stakpak and run /init and stop babysitting your servers
Stakpak Browser Extension
Stakpak Browser Extension lets Stakpak use your browser. It can navigate, click, and type directly in your active session via simple CLI commands, with full auditability, so it can debug dashboards and use the AWS Management Console easily.
Knowledge Store
Knowledge Store is Stakpak’s new persistent operational memory layer for infrastructure.
It allows both users and stakpak to save, search, read, update, and remove operational knowledge across sessions using stakpak ak.
Instead of starting from zero every time, Stakpak can retain important context like infrastructure discoveries, troubleshooting notes, deployment procedures, architecture decisions, and project specific operational knowledge.
Stakpak vs AWS DevOps Agent
We also published a blog comparing Stakpak to the AWS DevOps Agent. While the AWS DevOps Agent focuses on investigating incidents and suggesting next steps, it still relies on engineers to take action.
Stakpak goes further by actually executing fixes when it’s safe, as well as provisioning and modifying infrastructure across your stack.
How are engineers using Stakpak?
Devs are using Stakpak for real infrastructure work that usually takes hours, and now we have turned it into step by step tutorials:
Bug Fixes
Improved shell command approval safety
Fixed subagents issue with local providers
Better model selection and fallback handling
More reliable model downgrading logic
Improved sandbox and plugin resolution
Tried Stakpak and have feedback, good or bad? Just reply to this email. It goes to a real human, we promise!
Try it yourself: Stakpak.dev
Written with ❤️ (and a lot of coffee) by the Stakpak team ☕💻