You can now expose long running agents via MCP and use it in Claude. You want to run a research agent or kick off a CI job and listen to the result. You also can now ensure with elicitations that a human is accepting an action or filling in data, or delegate to a URL for payments
Weโre adding support for MCP Tasks and Elicitations in Claude.
Tasks let Claude give plugins long-running jobs such as deep research and report back when theyโre done.
I have recently started to using Claude to write proofs for my software. Itโs so much more tractable. What our research team did goes much further, but it shows the beginning of making proofs just part of normal software
New on our Frontier Red Team blog: using Claude to formally verify software at large scale.
Formal verification can prove that code is free of certain bugs, but historically it has only been feasible for small codebases (or required a lot of expert time).
With AI models, however, this is beginning to change. In collaboration with @theoremlabs, we used Claude to prove security properties in over one hundred real-world open-source libraries over the span of a few weeks.
Read our research in full here: https://t.co/zUyx8CjBEk
Agent Skills started as an experiment in Claude Code. Many across the industry have come to adopt it, so we at Anthropic found it more fitting to donate it to the @AgenticAIFdn, next to MCP.
Skills has become the best way to get an agent working on something new. like MCP, weโre glad so many have found the standard to be useful. and like MCP, we think itโs time for skills to be stewarded by @AgenticAIFdn, where it can continue to be open and community-driven ๐
I am so proud that we are taking the most impactful standard for Agents in the last years and donate it to the Agentic AI Foundation. I am looking forward to Anthropic steering the Agent Skills standard together with Earendil (PI), OpenCode, Microsoft, Google and OpenAI
We're donating Agent Skills to the Agentic AI Foundation (@AgenticAIFdn), the same foundation we donated MCP to.
Since December 2025, 70+ agent platforms have integrated the standard, including ChatGPT, Cursor, GitHub Copilot, and VS Code.
Here's what that means for you:
Great point by @dsp_ : when an authโd MCP server includes skills, it can send down skills specific to the logged in user. Maybe even personalized to that user!
I hadnโt even considered that benefit of skills-over-MCP.
If all the govs shutting down servers for a PHP update, all the hosters & enterprises terrified by the vuln, gave $100k, or just $50k a year to the PHP Foundation, things could look very differently. Go, make your employer sponsor *today* https://t.co/z1mUcXMqy9 #php #phpc @Th@ThePHPF