jedisct1

Moonshot AI are also whitelisting user-agents: https://swival.dev/pages/providers.html#generic-openai-compa...

Weird.

Swival.dev has a tool to run Python code, and it can expose the MCP tools as Python functions.

This saves a couple prompts and tokens, but only in very specific cases, only with MCP servers that don't already support code mode, and only with large models.

In practice, I've not found that to make a real difference.

Plus, it's likely that more and more MCP servers are going to use code mode natively anyway.

Typescript is perfect for that kind of application. Plus, the same code runs everywhere, no need to ship native binaries for every OS and CPU.

As someone who had pretty much given up on Rust after being very disappointed with its SIMD support, and had stopped following developments in that area, I was genuinely surprised to see someone announce this today:

https://val.markovic.io/articles/philbin-the-safest-and-fast...

Performance is competitive with Zig, without using the "unsafe" keyword or nightly Rust.

Nice.

This is exactly what Swival was built for: https://swival.dev

Looks similar to Anonymized DNS, which DNS users have been using for more than seven years: https://www.ietf.org/archive/id/draft-denis-dprive-dnscrypt-...

Happy with swival.dev ...

You know, swival.dev is fully opensource, doesn't hide anything, fully supports GLM, has excellent context management to keep token usage low, and doesn't send anything you didn't ask to the cloud.

EdgeDNS was rewritten and is now EtchDNS https://etchdns.dnscrypt.info

No support for Zig?

MLX quants for Apple M5 with 128 GiB RAM: https://huggingface.co/jedisct1/Qwen3.8-Flash-Next-oQ4e-128k

I’ve given up trying to read PR descriptions and comments blindly copied from Claude vomit. It’s insulting, and it gives me a headache.

Whenever I use Claude models, I do something similar, telling it to never write documentation directly, but asking "uvx swival -- --profile qwen" to do it after describing the changes to it.

And I just stopped reading PRs and comments blindly copied from Claude vomit. It's unreadable by a human.

Have you heard about fil-c?

Buy a subscription, get something different every month.

How context management works in Swival https://swival.dev/pages/context-management.html

They got a new user.

https://swival.dev remains my favorite agent for local models.

Related: https://pcre-vera.github.io/pcre-vera/ (WIP)

Some clients do that automatically: https://swival.dev/pages/secrets.html

This is a fantastic option for swival.dev given its very efficient context management compared to e.g. Claude Code.

Swival Audit feature gives really good results even with small, open models: https://swival.dev/pages/audit.html

This is not an issue with Claude, but with Claude Code.

With swival.dev you can use long sessions without ever doing any manual compaction or reset, even with Claude models.

https://swival.dev is an AI agent that was designed for open models since day 1.

"This browser doesn't support WebAssembly JSPI, which Firefox WASM needs to run."

What a waste of tokens.

Not opensource.

I build a Signal client for Swival: https://github.com/swival/nbclaw

And I use it daily for simple tasks.

GLM-5.2 is a great model!

But it already works really well with existing harnesses, I'm not sure why a dedicated one is needed?

I use it with https://swival.dev and everything works perfectly, no tool calling issues and it works fine with long sessions.

Monetization Gateway for Bunny CDN: https://github.com/dip-proto/x402

and for Fastly: https://github.com/dip-proto/x402-fastly