Claude Code in Slack
Ben's Bites ● Covered by 3 sources
Anthropic added a Claude Tag feature so teams can @-mention a shared Claude Code agent right inside Slack. It keeps context from the channel and can run tasks while you go do something else.
Based on reporting by Ben's Bites — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic just made Claude Code a lot more social. The new Claude Tag feature lets teams summon a shared instance of the coding agent directly inside Slack, the same way you'd ping a coworker. Type the tag, hand it a task, and it pulls context straight from the surrounding conversation instead of starting cold.
That context-awareness is the actual upgrade here. Anyone who's used an AI coding assistant knows the tedious part isn't writing the prompt, it's pasting in the backstory: the ticket, the thread where someone flagged a bug, the half-finished plan from three messages up. Putting Claude Code inside Slack removes that step entirely, since the agent can just read the room.
It also fits a pattern showing up across the industry right now. Notion just opened its developer platform to outside agents like Claude Code, Cursor, and Codex, letting them work off shared docs and task boards. Figma rolled out tools at Config for turning design layers straight into code. Everyone's chasing the same goal: get the agent embedded where the work already happens, rather than making humans shuttle information into a separate chat window.
Delegating a task and walking away sounds nice in theory, but it also means trusting an agent to interpret Slack chatter correctly without supervision, which is a different bar than trusting it inside a controlled IDE session. Anthropic is betting teams will take that trade for the convenience. Given how fast agentic tools are spreading into Slack, Notion, and now browsers via Gemini's new computer-use skills, that bet looks reasonable for now.
My take — AI-written commentary, not fact-checked reporting
Sticking agents inside Slack is the obvious next move, but it quietly raises the stakes: an agent misreading a sarcastic comment in a thread and shipping code based on it is a very different failure mode than a bad autocomplete suggestion. I like the convenience, I just don't think 'delegate and walk away' should be the default setting yet.
Read more about this at: Ben's Bites
Related stories
Slack Code taps into collective vibe, puts AI agents into the group chat
The Register · 3 weeks ago ·
27