Anthropic launches Claude Code for developers in Slack

Anthropic has launched Claude Code in Slack, letting developers handle coding tasks directly within chat threads.

The beta feature, released Monday as a research preview, expands on Anthropic’s existing Slack integration by enabling full workflow automation.

Developers can now tag @Claude to start coding sessions that use Slack messages as context, such as bug reports or feature requests.

Claude determines the relevant code repository, posts updates in the thread, and shares links for reviewing work and opening pull requests.

This launch follows a wider trend of AI coding assistants moving from traditional code editors into team collaboration tools.

🔗 Source: TechCrunch

🧠 Food for thought

Implications, context, and why it matters.

Claude Code in Slack can suggest edits plus open pull requests from thread context, but enterprises need clarity on permission scope (what access tokens allow), logging plus data persistence before any deployment 1.
Anthropic documents a GitLab continuous integration/continuous delivery (CI/CD) integration for Claude Code, while Bitbucket support is not listed in the available docs 2.
Open questions include required permission scopes and how code context gets stored versus processed. Data retention in the Slack preview also needs clarity, which leaves security teams with assessment gaps 1.
Within Slack, native Data Loss Prevention (DLP) scans messages and text files for rule hits 3, yet teams do not know if Claude Code chats fall under current DLP or need separate rules.
The Audit Logs API in Slack tracks events like file downloads but not message content 4, so teams may lack visibility into Claude’s code activity kicked off from threads.

Security vendors could add middleware (a software layer that sits between apps and services) that intercepts Claude Code commands in Slack before they touch repositories, enforcing policy gates, approvals, and DLP checks.
Vendors such as Strac, a data security platform, say they offer Slack DLP on all plan tiers with sensitive data detection and redaction 5, which could extend to AI actions on repos.
Enterprises that use GitHub, GitLab, or Bitbucket together need a single control plane (a centralized policy and visibility layer) that audits repository access, proposed changes, plus branch protection enforcement.
The GitHub Slack app already filters notifications by branch, labels, or actors 6, which hints at controls that could gate AI actions by repo sensitivity, user roles, or compliance needs.