Each one includes a starter kit and walks you through a real use case, end to end:
We're starting with these three, with more on the way. Bookmark this so you don't miss the next one.
If you've been running the CLI package directly, it's worth switching to sanity for all commands. That one change unlocks a lot, including new global --project-id and --dataset flags so you can run commands outside a project directory without a config file, automatic CI detection so prompts don't hang in automated workflows, and new SDK projects created via sanity init now default to ESM.
Check the full changelog for a full list of improvements.
Content Agent started in your Dashboard. As of last month, it's in Slack too and available via an API, so you can build it into your own tools and workflows.
To get started, install it using this install link. Once it's in your workspace, you can @ mention Sanity in any channel or thread to run content ops, surface insights, and edit your Content Lake without switching context.
At Sanity, we’ve been using it for content audits, web page editing, reviewing content releases, and writing changelogs. See more details in the launch post. Regardless of where you use the agent, every edit is staged as a draft first. Nothing goes live until you review and publish it.
The engineering team has also shipped a couple of improvements worth knowing about. You can now:
"To be completely honest, I only use Sanity through LLMs. It's the only way I interact with the product, and it's how I've migrated all of my projects over."
If that resonates, you’d be happy to hear that Sanity is now available as a connector in Claude and Lovable, and via the Sanity plugin in Cursor. The MCP has been around for a while, but these integrations make it even easier to get going. Check it out:
npx sanity@latest mcp configure/add-plugin sanityThis Reddit thread is also a gentle reminder that we’re on Reddit, where you can have conversations like above with Knut and others.
"Prompting agents is what users do. [...] Writing for agents is what builders do. You're writing system prompts that define who the agent is, what it knows, how it behaves. [...] Prompting is improvisation. Writing for agents is building their architecture."
It's a useful distinction if you're building anything with agents right now. The post goes deeper into what good system prompt writing actually looks like, with lessons from building Content Agent.
On March 19, we hosted our first Sanity Pioneers event in San Francisco. Thank you to everyone who showed up ready to listen, learn, and build.
The Pioneers program brings together the people running content operations at scale to share what's working, what's not, and where the industry is headed.
(Pictured above alongside me): Meaza Dawit Abate (Engineering Manager at Airtable) and Andy Fitzgerald (Content Operations Consultant at Elemeno Health) discussed how they've built real AI workflows on structured content, from granular localization pipelines to LLM-supported taxonomy generation.
You can watch the event recap here. The full fireside chat recording will be available soon.
Follow us on socials and subscribe to Luma to stay up to date on the next event (which will take place in Europe, that's all I'll say for now!) and how you can get involved.
That’s it for this month - see you in Discord 👋
Evelina
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。