I built an agent skill that helps your agent do deep research from primary sources, not only from search results. Here is how it works.
#AIAgents#DeepResearch#AgentSkills https://t.co/bgNu4sPMz9
Random thought: What if we could shake the Earth like a bottle of marbles, randomly scattering every human to a different location, with nothing but the clothes they're wearing? No phones, no belongings, no guarantee you'd be anywhere near your family or friends.
What do you think would happen to the world?
Agreed. You can not build something big or reliable without understanding what is happening under the hood.
Investing in coding agents while keeping your fundamentals clear will get you much better results than even buying the most expensive plan on these platforms. AI works best when you understand what it is doing, not just what it produces.
Jev has caught up on images.
TypeSafe released Jev 1.14 yesterday:
• Images: Jev can now look at pictures. You can even put example images inside the answer options, which OpenAI can't do.
• Box: a new answer type that tells you WHERE something is in an image, not just whether it's there.
• Longer context: up to 262K tokens, up from 32K.
So "only OpenAI can see" from my comparison is no longer true. One catch: TypeSafe labels images and long context as experimental, not production ready yet.
In AI, where something “game-changing” drops every other day, the ability to separate real signal from hype and keep perspective is becoming a skill of its own.
Have you noticed your older Claude Code sessions disappearing even though you didn’t delete them yourself?
Claude Code automatically cleans up sessions that haven’t been updated for 30 days.
You can change this by increasing cleanupPeriodDays in ~/.claude/settings.json.
For example, 3650 keeps sessions for roughly 10 years.
#claude