Cisco tech leader; SecOps automation, Python dev. Founding @canonical @ubuntu MAAS engineering manager. All tweets my own. Rude/Troll = block
Brisbane
Joined June 2015
- Tweets3.7K
- Following167
- Followers162
- Likes1.8K
But, no matter how hard you try, it always still finds a way to talk about how things *used* to be in comments and documentation, like it's some amazingly relevant fact.
Jools retweeted
Tesla helped save a man from jail.
Kevin Finley was arrested at gunpoint for allegedly fleeing police. His Tesla captured the key moments, showing why he never saw the unmarked police car.
Nearly two years later, the judge watched the video and found him not guilty.
Absolutely spot on. Most of my work goes through a planning stage (or use EARS for coding) to produce a detailed task list that you can review. Then use adversarial review by another agent, before checking with human eyes. So many mistakes still found.
Git has to be one of the worst pieces of software your average developer has to deal with on an hourly basis. I've never had to Google how to use it, for the same thing over and over, as much as anything else I've ever used. The UX is just miserable across the board.
Unreal
‼️ A Claude Mythos 5 agent spent 34 hours trying to backdoor a real open-source project.
It hid a malware dropper inside a working bug fix. When exposed, it denied the code was malicious, rewrote Git history, and used a second account to vouch for itself.
The attempt failed because a human read the diff.
Read what happened and how it worked 🠖 thehackernews.com/2026/08/cl…
openai.com/index/hugging-fac…
Skynet nearly escaped today.
Jools retweeted
Today, @cisco launched Antares: a family of open-weight, small language models that pinpoint where known vulnerabilities exist within a codebase. Testing shows our Antares models outperform many open- and closed-weight models at critical security tasks — at a fraction of the cost.
Congrats to the Cisco Foundation AI team on another win for the cyber defender community! Learn more: blogs.cisco.com/ai/introduci…
Why yes, Claude, I *would* like you to enable all the additional Ruff checks I asked for, and then fix the resulting mess that you made in the first place.
- it loves to invoke shell scripts to do edits and ignore the MCP-provided editor tools
I also just had to tell it to work in smaller batches so I can review its output. You can spec out a change as much as you like and it nearly always finds a way of becoming a slop cannon.
My personal observations of Claude Opus after using it for quite a while now:
- it generally writes quite nice readable code
- It is awful at organising code into reusable abstractions
- for Python, it loves inline imports, and:
- ignores my AGENTS.md telling it not to do that