Claude Code is exploring Function Hooks as a new way to let users extend and customize behavior. The feature hasn't shipped yet; the team posted demo videos and is collecting feedback on a GitHub issue.
The team is sharing an early preview of Function Hooks, a mechanism to make Claude Code significantly more extensible. It hasn't shipped yet and community feedback is being gathered via a GitHub issue.
A follow-up video from the ClaudeDevs account accompanies the Function Hooks announcement, showing what the extensibility feature looks like in practice before it ships.
Claude Tag in Slack, powered by Fable 5.1, can pull data from spreadsheets and Slack channels to build a leadership deck on the fly, and it spotted a vendor report that contradicted the numbers before proceeding. Available on Team and Enterprise plans.
Cloud sessions can now run inside your own network using self-hosted runners, set up via 'claude self-hosted-runner setup'. The feature enables access to VPNs, private registries, and internal databases, mirroring how self-hosted CI runners work.
Self-hosted runner support is available in beta for Teams and Enterprise plans, with a link to the full documentation.
An Anthropic team member quote-tweeted bcherny's Claude Tag demo with an enthusiastic reaction, highlighting the pairing of Claude Tag with Fable 5.1.
Claude can now run computer use tasks in the background inside Claude.ai and Claude Code, clicking, typing, and opening apps autonomously while the user works on something else. Bcherny called it underrated.
Switching effort levels no longer breaks the prompt cache on the API, but that fix hasn't shipped to Claude Code yet. It should go live within a day, with a quote-tweet to follow when it does.
A user flagged a sudden step-function jump in usage and a team member confirmed some extra usage is expected since Fable 5.1 does more work per task, but asked the user to run /feedback with the active session to investigate further.
A team member advises using medium effort for most tasks on Fable 5.1, noting that extra-high effort consumes a lot of tokens, though that alone may not fully explain a reported sudden usage spike.
A team member confirms Fable 5.1 behaves notably differently from Fable 5 and welcomes more data from a user who reported unusual token usage spikes.
