Slack has introduced Slack Code, a new workspace for teams to collaborate with AI coding agents on software development. It brings coding work into dedicated Slack channels, where teams can follow agent conversations, review code changes, check live previews, provide feedback, and approve work within the same workspace.
Slack Code
Slack Code uses project-based code channels, with each channel mapped to a specific project or coding task. Users can tag a coding agent from any Slack conversation to create a dedicated code channel for the work. The agent can use context already shared in the conversation, including screenshots and documents.
The code channels include:
- Project-based workspace: Each channel is dedicated to one project and can be archived after the task is completed. The record remains available as an audit log, while the history stays searchable.
- Automatic channel creation: Tagging a coding agent from a conversation creates a code channel for the task, bringing the agent and relevant context into a dedicated workspace.
- Development views: Separate tabs provide access to the conversation, plan, code diffs, and live previews, allowing team members to follow the work as it progresses.
- Code review and previews: Code diffs show the old and new lines for review, while live previews display the resulting HTML output so teams can check the changes before they are shipped.
- Workflow APIs: APIs allow organizations to customize how agents work, including automatic account provisioning, user authentication, and workflow automation.
Slack Code can be used for tasks such as building a new feature, updating a web page, or fixing a bug. Users can tag agents such as Claude Tag or OpenAI’s ChatGPT, after which the agent works in the code channel. Team members can review the conversation, code diffs, and live preview, provide feedback, and approve the work before it is shipped.
The same workspace can include developers, product managers, designers, and non-technical team members. This allows people involved in a project to participate in the development process and interact with the coding agent from Slack.
Coding with agents in a shared workspace
Slack gives an example of a product manager finding a bug report in a channel and asking a coding agent to help architect a fix. The agent reviews the existing conversation, screenshots, and documents before proposing code. An engineer can then review the diff and preview before giving the agent approval to open a pull request and merge the changes to production.
Team members can pause, redirect, or stop an agent while it is working. Once the task is completed, the code channel can be archived while its history remains available for future reference.
Slack says its code channel APIs will be opened to the broader developer community. The APIs will allow custom agents to join code channels for software development and other workflows, including marketing campaign development and legal document review.
Coding agent support
Slack Code connects with coding agents from its launch partners:
- Anthropic: Claude
- Cognition: Devin
- GitHub: Copilot
- OpenAI: ChatGPT
- Vercel: Vercel’s agents
Anthropic, Cognition, GitHub, OpenAI, and Vercel are already developing in and for Slack Code using its code channel APIs. Access to each respective partner agent is required.
Agents across Slack
Slack is also updating how agents work outside code channels. Agent DMs now clearly identify agents, while threads receive titles so users can return to individual conversations.
The new Agents tab provides a central location for agents and agent sessions. It lets users browse agents and conversations, name sessions, check statuses, flag blocked threads, and stop agents while they are working.
Slack’s Add to Slack feature supports agents built with:
- NanoClaw
- Lovable
- Hyperagent
- Superhuman
- n8n
- Vercel
- OpenAI
- LangChain
- Runlayer
- Skydive
Slack handles the OAuth setup, manifest configuration, and environment configuration when an agent is added. Agents have a clear identity in Slack and follow the same channel norms and administrative controls across code channels, DMs, and project channels.
Security and human oversight
Agents in code channels use Slack’s existing security model, permissions, and administrative controls. Slack says they work within the existing setup without requiring separate infrastructure or identities to be managed.
For high-stakes actions, human approval remains part of the workflow. When code needs to be pushed to production, the agent packages its work for an expert to review and sign off within the Slack channel.
Pricing and availability
Slack Code is available on any Slack plan and requires access to the respective partner agent. Slack also offers over 500 AI apps and Agents through its Marketplace and plans to make the code channel APIs available to the broader developer community.
Note: Pricing and packaging are subject to change, and availability may vary by region and customer agreement.
Speaking on the launch, Rob Seaman, EVP and GM of Slack, Salesforce, said:
AI only creates value when it’s part of how a team actually works, not something people go do alone in another tab. Code channels bring that work into the open, where the rest of the team can see it, shape it, and build on it.