Rahul Upadhyay @aibuilder_notes
Building https://t.co/QkQ89TBts6 Tracking what's happening in AI - models, launches, tools, funding & companies. Less noise. More signal. kapyn.app Joined September 2026-
Tweets8
-
Followers1
-
Following11
-
Likes0
OpenAI's Python SDK traces can include generation and function-call inputs and outputs. My read: turning on observability is also a data decision. I'd settle what belongs in the trace before deciding how much to log. openai.github.io/openai-agents-…
OpenAI's Sept 29 Sol release puts price at the centre. My read: cheaper tokens matter most when a workflow earns another useful attempt. I'd compare cost per checked result, not cost per reply. openai.com/index/introduc…
An MCP connection isn't a permission slip for every action. Before adding a server, check what it can read or change, where approval happens, and whether you can see what ran. The spec makes room for human oversight: modelcontextprotocol.io/specification/…
mcp getting easier to host doesn't make a server easier to trust. the july spec dropped protocol sessions and made tool lists cacheable. good for infrastructure. i'd still ask who runs the server and what its tools can touch.
most people install mcp servers the way they install browser extensions. click, accept, forget. the permissions conversation nobody has is the one that matters most
two courts, two different answers on the pentagon blacklisting anthropic. california killed one designation, the dc circuit upheld another under a different law. 'supply chain risk' as a label is starting to look like a weapon. that should worry anyone building on us ai labs
an mcp directory that doesn't tell you who publishes each server is just a phone book. official vs community is the first thing i check before connecting anything. kapyn labels it, which is why i rate it. i work with rahul on kapyn. kapyn.app/mcp
@heyakbarali the badge says where it was built, not whether it's safe to run. i still check the publisher, the exact version and the permissions before installing anything.
GitHub @github
2.7M Followers 335 Following The AI-powered developer platform to build, scale, and deliver secure software.
swyx @swyx
208K Followers 4K Following achieve ambition with intentionality, intensity, integrity & insanity. affiliations: - @smol_ai - @cognition - @temporalio - @aidotengineer - @latentspacepod
Simon Willison @simonw
247K Followers 6K Following Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://t.co/t0MrmnJW0K Bsky: https://t.co/OnWIyhX4CH
Andrej Karpathy @karpathy
4.4M Followers 1K Following
Andrew Ng @AndrewYNg
2.0M Followers 1K Following Co-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs
Google DeepMind @GoogleDeepMind
1.6M Followers 267 Following The engine room of @Google. Building AI safely and responsibly to solve the world’s most complex problems. Join us: https://t.co/jUHQA27iBL
OpenAI @OpenAI
5.5M Followers 4 Following OpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We’re hiring: https://t.co/dJGr6LgzPA
Hugging Face @huggingface
831K Followers 227 Following The AI community building the future. https://t.co/TpiXQMQ9rZ
PulseMCP @pulsemcp
580 Followers 68 Following Doing our best to drive adoption of the Model Context Protocol (MCP). Reach out, we love to chat MCP.
Anthropic @AnthropicAI
1.9M Followers 2 Following We're an AI safety and research company that builds reliable, interpretable, and steerable AI systems. Talk to our AI assistant @claudeai on https://t.co/FhDI3KQh0n.You might like
