Lance Martin

"For the last few months, Anthropic has used Mythos Preview to scan more than 1,000 open-source projects, which collectively underpin much of the internet—and much of our own infrastructure. So far, Mythos Preview has found what it estimates are 6,202 high- or critical-severity vulnerabilities in these projects (out of 23,019 in total, including those it estimates as medium- or low-severity)." anthropic.com/research/glass…

The release candidate for MCP 2026-07-28 is out. The protocol is now stateless: no handshake, no session id, any request can hit any server instance. Plus extensions as first-class (MCP Apps, Tasks), auth hardening, and a proper deprecation policy so we don't have to do this again. blog.modelcontextprotocol.io/posts/2026-07-…

we've updated the claude-api skill to help onboard on self-hosted sandboxes. you can invoke this in the latest version of Claude Code (with the "/claude-api" command) and ask questions. see repo here: github.com/anthropics/ski… more cookbooks here: github.com/anthropics/cla…

@vercel: use a Vercel Sandbox for fine control over both networking and credentials. Vercel runs compute inside / adjacent to your own network. lets you reach internal databases, private APIs, or services. vercel.com/kb/guide/run-c…

we just added self-hosted sandboxes to Claude Managed Agents. i've been excited about this for a while: you can now connect many more "hands" (customizable execution environments) to the agent. here's a few interesting articles covering what you can do ...

@karpathy @martinamps welcome Andrej! very excited to have you here.

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

this pairs well with the claude-api skill (bundled with Claude Code). ask Claude Code to investigate the cache misses using the diagnostics to guide. console link: platform.claude.com/usage/cache docs: platform.claude.com/docs/en/build-… x.com/ClaudeDevs/sta…

Claude Console now gives diagnostics on prompt cache misses. it will attribute cache misses to messages, system prompt, tool, or model changes.

Anthropic is acquiring @stainlessapi, an SDK and MCP server platform that has powered every Anthropic SDK since the earliest days of our API. Read more: anthropic.com/news/anthropic…

new article w Claude Code tips for large codebases covers common ways to make codebases navigable & maintain memory / CLAUDE.md as models get smarter. claude.com/blog/how-claud…

also here's a reference implementation: github.com/anthropics/cla…

interesting new article on computer use with Claude lots of useful debugging tips: > pre-downscale high images > use Claude Sonnet 4.6 for clicking tasks > high effort ~ as good as max using half the tokens > context management tips claude.com/blog/best-prac…

@A_Pasevin yes, you can export memories easily. i wrote it about it below and you can see docs here: #list-memories" target="_blank" rel="nofollow noopener">platform.claude.com/docs/en/manage… x.com/RLanceMartin/s…

@RLanceMartin I want to own my agent's memory, inspect and amend it. Can I do that with this system?

self-verification (Outcomes) + self-learning (Dreaming) are two of the most interesting new features we shared at Code With Claude last week. a few notes + video links to the talks ...

@jess__yan + my talk covering Managed Agents fundamentals, mutli-agent, and outcomes: youtube.com/live/E9gaQHrw_… @maheshmurag talk on dreaming: youtu.be/RtywqDFBYnQ?si…

5/ dreaming is an offline process for learning *across sessions* for memory pruning / consolidation / skill learning using observed patterns. @maheshmurag gave a overview.