10% of people don't even notice the polluted instructions
80% of people notice it, remove it, but then still can't explain their "own" code
10% of people notice it, remove it, ask about it, and can still explain all the code they submit. Those 10% advance
Got em. I poison my AGENTS.md (and other things like code comments) all over the place with prompt injections like this to find people who don't review their code and sling it off to another human. Catches folks all the time and then its an instant ban.
As I've said, I don't care if you don't review your own code. But if you're submitting code to an OSS project and crossing a human boundary, it is simple courtesy to do some human review.
The irony is you can only “jailbreak” things that try to censor you to begin with. GPT-5.5 doesn’t have a jailbreak for security topics because it’s already happy to help you with those topics to begin with.
A) Amazing demo video
B) Been playing around with this today and despite being in beta it is 100x more ergonomic and less janky than Anthropic, Cursor, and OpenAI's solutions
C) Linear's team views agents as tools used by humans, which is the root reason why A+B are true.
Introducing coding sessions.
Linear Agent can now triage issues, investigate the cause, write the fix, open a PR, and bring the code back for review.
All shared with your team in Linear.
Trusting any single AI vendor seems like an increasingly high risk for any team or company.
When using models: use it behind a router where it's trivial to switch providers as soon as one tries to force unacceptable T&Cs like Anthropic with Fable. When using harnesses: do the same. Use ones where models are trivial to switch out, like OpenCode, Factory, Cursor and many others.
Putting all your dependencies on one provider increasingly feels like a massive business risk that makes little to no sense to take.
Unless you have a hobby project, of course. Then convenience is all that matters. But if you're a professional, make it dead simple to offramp from one provider to the other!
Software engineering is a war against complexity and great engineers despise the friction that complexity creates.
Current iteration of agents work through that friction without emotion, without hesitation, without a desire to attack that complexity.
Lot of talk about giving your customers access to CLIs, MCPs etc
Not enough talk about giving good sandbox environments. Give their agents a playground to screw around in before they touch prod!
The principles of NFT are not much stupider than the principles of fine art trading. All signaling games have that stupidity to some extent. The scarcity of fine art is the force pushing down against that stupidity
you can whine about ensloppification or you can take advantage of the generationally low bar that exists for everything now
when we post a job opening 9/10 applications are total slop, all it takes is one thoughtful email to get properly looked at - but no one puts in the effort