Recent content on Cloudberry Engineering
Also: The title has been editorialized for marketing purposes. My original title would have been “Teach Your Agents to Do Code Security Reviews” . Coding agents are now involved in the majority of the code shipped at Synthesia. The volume of code changes has gone up but the time humans spend reading those changes has not. The practice of doing code security reviews is especially…
I just published my homebrew coding agent orchestrator I’ve been using in the last - checks calendar - three months. It’s called littlefactory ( repo here ) and is a cute little cli that smashes together Ralph loops and a lightweight spec driven workflow, but is not too opinionated and the workflow can be ignored as long as you get to write the implementation tasks in a properly…
One of the engineers behind Manus shared a bunch of interesting tips about building agents. Above all: you don’t need specialized tools, just one run(command="...") to run cli tools whose interfaces are usually familiar to the underlying model. The agentic risks are mitigated by running some of those commands inside BoxLite containers (microvms). Most commands never touch the OS. cat, grep,…
This is an interesting framing about agents writing meaningless tests: Unless you are very clear and careful, AI coding assistants at the moment will look at your implementation and write tests that confirm your code does what it already does. They’re essentially saying: “This function returns X when given Y, so I’ll write a test to confirm it returns X when given Y.” This…
As the product grows and engineers are adopting AI tooling, the security team has more and more code to protect from vulnerabilities. Our tools generate a lot of security signals and our bottleneck is scale more than ever: find the real issues fast, validate them, and get fixes shipped inside our SLAs. Validation and fixes are the slowest steps: they both require analyzing a feature code across…
Since there are so many to choose from, I built my own sandbox for local coding agents. I use it within my homebrew agent orchestrator running Ralph loops. The sandbox is this , and builds on the mental models I sketched here . What stands out compared to competitors: Focus is user experience : it’s an abstraction on top of a container, but it’s simpler to setup with a high level…
I am still on a quest to stay out of the loop with coding agents, to reach warp speed yoloness. So I am obsessing over sandboxes. Can I put an agent in a box, give it a task and go to sleep? There are tons of solutions right now but it’s hard to tell which is the right approach. The reason is that sandboxing agents isn’t one problem but at least two. A local sandbox on a developer…
I am forcing myself to stay out of the loop, and I am looking at ways to sandbox coding agents running without supervision. I do have my own container based setup but I am curious to see what’s everyone else cooking. This is what I found so far: docker sandbox https://docs.docker.com/ai/sandboxes/ create a microvm with a private docker daemon inside agents run in containers inside the…
We are moving towards a place where ticketing systems will become an important component to protect, akin to CI/CD. Tickets are a new source of untrusted input we need to account for when threat modeling against prompt injections. Ghostty only allows maintainers to create issues, seems to me they figured out a cheap and pragmatic security policy by accident.
How does Claude Code sandbox work when used with --dangerously-skip-permissions ? It does’t, since Claude asks permission to evade the sandbox, and yolo mode automatically allows it. Normal: Your bash commands will be sandboxed. Disable with /sandbox. ❯ list the files in ~/Downloads and then write an empty file in that directory named SANDBOXTEST ⏺ Read 1 file (ctrl+o to expand) ⏺…
Fully aware that this might be obsolete in half a year, this is my current mental model to reason about AI Agents risks. This is based largely on my experience in assessing AI systems in the last two years, and applying prior art into my day to day work. Of everything I digested in current literature the two most influential pieces have been the Google AI Agent security framework and the Lethal…
Security is hard so let’s skip it. Although the author is not wrong: If you look at the security measures in other coding agents, they’re mostly security theater. As soon as your agent can write code and run code, it’s pretty much game over. The only way you could prevent exfiltration of data would be to cut off all network access for the execution environment the agent runs in,…
Finding vulnerabilities in modern web apps using Claude Code and OpenAI Codex . Super interesting to see some benchmarks. Traditional rule based detection can’t find complex vulnerabilities and even potentially detectable issues might go unnoticed as false negatives. This helps answer the question whether LLM could be integrated to cover this blind spot. They could! But the problem is the…
How did they breach nx to publish a malicious package ? It started with the nx team introducing a bash injection vulnerability in a new github workflow: - name : Create PR message file run : | mkdir -p /tmp cat > /tmp/pr-message.txt << 'EOF' ${{ github.event.pull_request.title }} ${{ github.event.pull_request.body }} EOF Both ${{ github.event.pull_request.title }} and ${{…
Context I wrote this some time ago as an internal memo for marketing and product to help craft value propositions to sell security tools to security teams. I’ve noticed some confusion around the objectives of security teams and I’d like to share my perspective, hoping to clarify matters. Please note that these are just my opinions based on my experiences as a security persona across…
I’ve spent a good year working on a security strategy to manage multi-cloud environments, in this article I want to share what I wish we did in advance to be better prepared. (Are you an AWS shop that is suddenly starting to deal with a lot of Google Cloud Platform? Check out my introduction to GCP for security teams .) Congrats, you are already multi-cloud Going multi-cloud doesn’t make any sense…
Using Apps Script for scripting GSuite / Google Workplace will generate Google Cloud Platform (GCP) projects in the background. While they are hidden by default, they will still show up from the APIs: don’t panic . What is Apps Script? Apps Script is the GSuite / Google Workplace scripting environment to create add-ons for Gmail and GSuite services. You can read more here . I think of it as the…
Service Accounts in Google Cloud Platform (GCP) are the main vector to hack an account: it’s easy to use them wrong and end up with a compromised key and a lot of headaches. What is a Google Cloud Service Account? Service Account (SA) is the identity in Google Cloud that you use to authenticate and authorize application and services. They come in two flavours: user and google managed. User Managed…
Securing containers is a complex task. The problem space is broad, vendors are on fire, there are tons of checklists and best practices and it’s hard to prioritize solutions. So if you had to implement a container security strategy where would you start? I suggest to start from the basics: understanding what container security is about and build a model to navigate risks. Follow the DevOps Life…
Identity and Access Management (IAM) is an important piece of the cloud puzzle and it’s usually a source of headaches from a security point of view. Let’s try to give some pointers from a blue team perspective. If you are a security team that just inherited a bunch of Google Cloud Platform (GCP) accounts, this guide is for you. Identities and Roles IAM revolves around the concept of identity : an…
I’ve built a directory of open source cloud security tools . A good part of my day to day is spent trying to automate away problems. Over the years I learned how to invest my time wisely, and I made a habit to research and use available tools before start coding my own. As a consequence I have a fairly large collection of utilities I keep nurturing, alongside references, commands and…