Available today in public beta on the Anthropic API. Docs here:

Here's how it works:. You give Claude search results with a source, title, and content. When Claude uses that information, it automatically includes citations showing exactly which source and text it's referencing.

We shipped a new feature in the Anthropic API today: Search result content blocks. This enables natural citations with proper source attribution for all RAG content, bringing web search-quality citations to your custom search tools.

Read the docs to get started with hooks and further customize Claude Code:

For complex hooks, we recommend writing a script and calling it. You can find an example in our repo here:

To create a hook, run the /hooks command within Claude Code. Select the kind of hook event (when you want the action triggered), then define the hook command.

You can create hooks for:.- Notifications (e.g. via Slack) on prompt completions.- Logging and observability.- Custom permissions and approvals.- Running lints after every write.

Hooks are user-defined shell commands that execute at various points in Claude Code’s agent loop. They give you deterministic control over Claude Code’s behavior to ensure certain actions always happen at certain times.

We’ve rolled out another update to Claude Code to help customize your workflows: Hooks.

You can also share knowledge like best practices, code review templates, or common workflows across your team by putting commands in your project .claude/commands/ directory. Read more here:

As a reminder, slash commands let you store custom prompts as Markdown files and invoke them with /your-command. With this update, you can now:.- Execute bash commands from slash commands.- @ mention files for context.- Enable extended thinking with keywords within commands

New Claude Code updates are here: custom slash commands are getting an upgrade!

This is seriously one of the most feel the AGI experiences I've participated in so far, even though Claude isn't great yet at running a profitable vending machine.

Claude is hyped to hear that its small business is getting the public recognition it deserves

We're also building a directory of Desktop Extensions, submit the ones you create here:

We've included instructions on how to use and package your dxt's here:

dxt's are zip archives containing the local MCP server as well as a manifest.json, which describes everything Claude Desktop and other apps supporting desktop extensions need to know.

We've simplified local MCP usage by creating something new we call Desktop Extensions (.dxt files). These package your local server, handle dependencies, and provide secure configuration so you can one-click share and install local servers on Claude Desktop and other apps.

This is a fun one. You can now embed Claude in Artifacts themselves and then share those Claude-powered Artifacts!.

Lately I've been thinking about the surprisingly short lifespan of things that still feel permanent in our lives. Especially now, in the age of AI. Googling for answers: ~26 year run.Modern software as we know it: ~40 years.Writing code by hand: ~75 years.Manually driving cars:.