Loving Toad by @willmcgugan - my default choice for driving models now. Streaming updates is great user feedback when the Agent/LLM is generating long tool calls (like writing files). Seen here with Kimi K2 Thinking (via HF), using a Model Evaluation skill.
Group chats in ChatGPT are now rolling out globally. After a successful pilot with early testers, group chats will now be available to all logged-in users on ChatGPT Free, Go, Plus and Pro plans.
I WANNA SHIP BUT I DONT WANNA DO FRONTEND
we end up looking over a LOT of PRs given how many we receive for opencode i can boil down a lot of mistakes into not understanding the idea of state and derived state the rule of thumb is you want to default to having as few pieces of state as possible and deriving values…
I feel people don’t demonstrate what a good client can be (including ourselves). Progressive discovery of tools can easily be done. Using a programmatic interfaces to parallelize and combine tool calls effectively are just the tip of the iceberg of how to build a good client.
Explicitly support UUID7 for JSON RPC request IDs so that you do not need to track an external monotonically incrementing message ID. Most of the SDKs already implicitly support it, and it is within the JSON-RPC spec. x.com/evalstate/stat…
We're growing the uv team! uv is one of the fastest-growing developer tools ever, built and maintained by a small, talented team :) Write Rust, for Python. Full-time, open source, remote. We're very well-funded and in it for the long haul. Link in next tweet 👇 or DM me here.
this MCP server was supposed to be providing a secure sandbox the secure sandbox: 😭
Today we describe how we leverage AlphaEvolve, a @GoogleDeepMind system for iteratively evolving code, to morph snippets of code towards better proof elements in complexity theory that can be automatically verified by a computer program. Read more at: goo.gle/3IR6vEN
the new gpt-oss chat template (harmony) is fascinating... - JSON gets converted and rendered to a typescript style-schema. - inbuilt tools get their own namespace. - tool names are used to correlate request/response (rather than call ids).
We❤️MCP at Hugging Face. Here we share our experiences building and running our unique server - and explaining what transport options you have as well as some production challenges.
pssst...no official release yet but things are coming along: github.com/modelcontextpr…
Whoa! Native support for MCP on Windows.
fast-agent now has Open Telemetry support. if you are developing MCP servers this is really good as you can observe the behaviour of different server/tool mixes, descriptions and prompts against different models. it's also easy to set up (especially with docker).
I want to see more MCP hosts to provide Sampling and really enable rich interactions. We just scratch the surface of what’s possible, and MCP hosts are sleeping on the opportunity.
@code has an official MCP server thanks to Microsoft's Semantic Workbench team! github.com/microsoft/sema…
@TerminalMan_ss shows us how to bring math smarts to Claude Desktop (or any MCP-compatible app), powered by @Wolfram_Alpha
United States Tendances
- 1. FIFA 140K posts
- 2. FINALLY DID IT 409K posts
- 3. The Jupiter 80.5K posts
- 4. The WET 107K posts
- 5. $MAYHEM 2,705 posts
- 6. Infantino 25K posts
- 7. The BONK 241K posts
- 8. Warner Bros 181K posts
- 9. #NXXT_AI_Energy N/A
- 10. #FanCashDropPromotion 2,558 posts
- 11. HBO Max 73.7K posts
- 12. Kevin Hart 3,470 posts
- 13. #NXXT_CleanEra N/A
- 14. Hep B 10.5K posts
- 15. NextNRG Inc 2,910 posts
- 16. #GenerationsShift_NXXT N/A
- 17. Chris Henry Jr 9,792 posts
- 18. Matt Campbell 5,871 posts
- 19. Paramount 31.7K posts
- 20. Robbie Williams 2,717 posts
Something went wrong.
Something went wrong.