Slack Launches New AI Coding Integration to Streamline Collaboration and Development Workflows

Slack Code creates project-specific code channels that archive automatically when tasks finish, leaving a searchable, auditable record of what was done and by whom.
Agents operate within Slack's security framework: they inherit existing permissions, and high-stakes actions such as merging to production require human sign-off under Slack’s enterprise security model.
Slack is expanding its AI-coding ecosystem beyond Claude and Devin to include Codex, GitHub Copilot, and Vercel agents; Claude Code originated as a research preview in December 2025, with Claude Tag beta in June 2026.
Slack Code is available on all Slack plans, designed to include non-engineers such as product managers and designers who can participate without leaving Slack.
Slack has launched Slack Code, a multiplayer AI coding feature that lets teams write and review software directly inside group chats The Next Web. Users tag AI agents like Claude, Devin, or GitHub Copilot to spin up dedicated code channels where the agents can work with full conversation context, show live HTML previews, and display code changes for teammate review The Register. Channels automatically archive when tasks finish, creating a searchable record of who did what The New Stack.
The system keeps humans in control. High-stakes actions like merging code to production still require human sign-off under Slack's security model The Register. The feature is available on all Slack plans and designed to include non-engineers—product managers and designers can participate without switching tools HeadTopics.
When a user tags an AI agent in a Slack message, the agent automatically creates a dedicated code channel for that project The Next Web. The channel pulls in all relevant context from the conversation. Team members see live HTML previews as the agent works, and diff views show exactly what changed before anyone ships code The Register. Once the task completes, the channel archives automatically, leaving behind a full audit trail of decisions and changes The New Stack.
Slack is expanding beyond its initial Claude integration to support a broad ecosystem of coding agents The Next Web. Available partners include Claude Code from Anthropic, Devin from Cognition Labs, OpenAI's Codex, GitHub Copilot, and Vercel Agent HeadTopics. The company positions this multi-agent approach as giving teams flexibility to pick the right tool for each task, all without leaving Slack Gizmodo.
Slack Code operates within strict security guardrails The Register. Agents inherit the same permissions as the human who tagged them. Slack requires explicit human sign-off before any high-stakes action—such as merging code to production—can execute The Register. This design prevents agents from making unauthorized changes while still speeding up routine coding tasks. Team members can pause, redirect, or stop agents at any point if something looks wrong The Register.
Slack Code reflects a larger industry trend: moving AI tooling directly into workplace chat apps Gizmodo. Instead of switching between Slack, code editors, and AI services, engineers, PMs, and designers stay in one place. Tasks that once required toggling between five or six windows now happen in a single channel. The automatic archiving feature also helps organizations maintain compliance records and learn from past projects The New Stack.
Publishers
10
Articles
5
Reach
15