Norin

ok Claude Fable 5 (Mythos) is finally here if you are on a subscription, go use it NOW because it may be removed from the subscription in a few days

Lot of buzz around loops/workflows. Many saying they’ve been doing this but not sharing their code. We’re sharing our code. This was not trivial to construct, it’s taken hundreds of hours of studying coding agent implementations, feedback from developers using it in repos that are 10M+ LoC, and re-architecting 3x from scratch. This is a production loop as in you can and are supposed to use it to gain actual ROI with your agent. We built in public you can find it on Github under Atomic (bastani-inc). Real production code needs management across dependencies, teams, and not infinite tokens to burn. We realized you need a way for the developer to define their ‘loop’ explicitly with good design, no provider lock in, review gates, verbatim compaction (not what you see today in coding agents), HIL, and the ability to observe and steer agents mid run. Why share it? Because we think everyone should benefit from knowledge on how to use these because we can get better with each other faster. Less hype, just the code. Overall, write up on our learnings coming soon.

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.

Lot of buzz around loops/workflows. Many saying they’ve been doing this but not sharing their code. We’re sharing our code. This was not trivial to construct, it’s taken hundreds of hours of studying coding agent implementations, feedback from developers using it in repos that are 10M+ LoC, and re-architecting 3x from scratch. This is a production loop as in you can and are supposed to use it to gain actual ROI with your agent. We built in public you can find it on Github under Atomic (bastani-inc). Real production code needs management across dependencies, teams, and not infinite tokens to burn. We realized you need a way for the developer to define their ‘loop’ explicitly with good design, no provider lock in, review gates, verbatim compaction (not what you see today in coding agents), HIL, and the ability to observe and steer agents mid run. Why share it? Because we think everyone should benefit from knowledge on how to use these because we can get better with each other faster. Less hype, just the code. Overall, write up on our learnings coming soon.

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.

Lot of buzz around loops/workflows. Many saying they’ve been doing this but not sharing their code. We’re sharing our code. This was not trivial to construct, it’s taken hundreds of hours of studying coding agent implementations, feedback from developers using it in repos that are 10M+ LoC, and re-architecting 3x from scratch. This is a production loop as in you can and are supposed to use it to gain actual ROI with your agent. We built in public you can find it on Github under Atomic (bastani-inc). Real production code needs management across dependencies, teams, and not infinite tokens to burn. We realized you need a way for the developer to define their ‘loop’ explicitly with good design, no provider lock in, review gates, verbatim compaction (not what you see today in coding agents), HIL, and the ability to observe and steer agents mid run. Why share it? Because we think everyone should benefit from knowledge on how to use these because we can get better with each other faster. Less hype, just the code. Overall, write up on our learnings coming soon.

Here’s what’s gonna happen: - you replace your code review with feedback loops (sentry, datadog, support tickets, etc) - you stop reading the code - software factory fixes everything - one day something breaks at 3am, agent can’t fix it - nobody’s read the code in 3 months - you have 3 weeks of downtime trying to re-onboard and fix it - you lose significant % of your contracts and users - your company is now dead

Lot of buzz around loops/workflows. Many saying they’ve been doing this but not sharing their code. We’re sharing our code. This was not trivial to construct, it’s taken hundreds of hours of studying coding agent implementations, feedback from developers using it in repos that are 10M+ LoC, and re-architecting 3x from scratch. This is a production loop as in you can and are supposed to use it to gain actual ROI with your agent. We built in public you can find it on Github under Atomic (bastani-inc). Real production code needs management across dependencies, teams, and not infinite tokens to burn. We realized you need a way for the developer to define their ‘loop’ explicitly with good design, no provider lock in, review gates, verbatim compaction (not what you see today in coding agents), HIL, and the ability to observe and steer agents mid run. Why share it? Because we think everyone should benefit from knowledge on how to use these because we can get better with each other faster. Less hype, just the code. Overall, write up on our learnings coming soon.

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.

Lot of buzz around loops/workflows. Many saying they’ve been doing this but not sharing their code. We’re sharing our code. This was not trivial to construct, it’s taken hundreds of hours of studying coding agent implementations, feedback from developers using it in repos that are 10M+ LoC, and re-architecting 3x from scratch. This is a production loop as in you can and are supposed to use it to gain actual ROI with your agent. We built in public you can find it on Github under Atomic (bastani-inc). Real production code needs management across dependencies, teams, and not infinite tokens to burn. We realized you need a way for the developer to define their ‘loop’ explicitly with good design, no provider lock in, review gates, verbatim compaction (not what you see today in coding agents), HIL, and the ability to observe and steer agents mid run. Why share it? Because we think everyone should benefit from knowledge on how to use these because we can get better with each other faster. Less hype, just the code. Overall, write up on our learnings coming soon.

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore. You should be designing loops that prompt your agents.