发生了什么
IT Brief New Zealand reports that Slack launched Slack Code, a that creates dedicated channels for software projects and AI coding agents. The report says users can start coding sessions from existing Slack conversations, review code changes, inspect live previews and retain a searchable project record. IT Brief New Zealand also reports that Slack expanded its broader agent-management features and added integrations with several third-party agent platforms. The report’s product details, partner participation and availability are not independently confirmed here.
IT Brief New Zealand reports that Slack launched Slack Code as a dedicated workspace for software projects involving human teams and AI coding agents. According to the report, a user can ask an agent to start a coding session from an existing Slack conversation. The resulting code channel is intended to separate complex development work from ordinary message threads while preserving the surrounding business context. The report says channels are created for individual projects and archived automatically when a task is complete, with a searchable record retained.
The reported workflow includes collaborative planning, inspection of code diffs and viewing output in a live preview tab. IT Brief New Zealand says participants can follow the agent’s work and help direct it without leaving Slack. The report presents this as a response to the difficulty of tracking multiple rounds of AI-generated development work inside standard threads. It does not independently test whether the works as described, whether previews support particular frameworks, or how the archive behaves across different Slack plans.
The report says users can invoke coding agents by tagging services including Claude, ChatGPT, Devin, GitHub Copilot and Vercel agents from an existing conversation. IT Brief New Zealand also reports that Slack added an agents tab for tracking sessions and status, changed direct messages with agents, and supports agents built on third-party platforms such as NanoClaw, Lovable, Hyperagent, Superhuman, n8n, Vercel, ChatGPT, LangChain, Runlayer and Skydive. The source does not establish that every named integration is available to every customer or that each offers the same capabilities.
IT Brief New Zealand reports that Anthropic, Cognition, GitHub, ChatGPT and Vercel are developing for Slack Code through application interfaces. It also attributes statements about the product’s purpose to Slack executives and partner representatives. The report says Slack automated steps including OAuth, manifest setup and environment configuration for agent deployment. Those claims come from the secondary report; no independent primary documentation, implementation details or customer evidence is provided in the source.
为什么这很重要
The reported launch moves AI-assisted coding into a shared collaboration space rather than leaving it primarily in individual developer sessions. That could make agent activity more visible to product managers, designers and other nontechnical colleagues, while giving teams a common place to review work and approve consequential actions. It also places Slack in competition with developer environments and workflow platforms seeking to become the operating layer for AI agents.
The central significance is organizational rather than merely cosmetic. IT Brief New Zealand reports that Slack is trying to make AI coding a team activity visible to people beyond the developer who initiated it. If accurate, a shared code channel could let product managers, designers and other stakeholders see the request, the agent’s intermediate work, the resulting diff and a preview in one place. That may reduce the gap between deciding what software should do and reviewing what an AI system actually produced.
The change also reflects a broader shift in how companies are packaging coding agents. The report describes competition over whether AI-assisted development happens in a collaboration tool, a developer environment or another workflow system. A platform that owns the discussion around a task may have more influence over permissions, review practices, context and records than a tool used only for one-to-one prompting. That makes the surrounding workflow important even if the underlying coding agents remain supplied by separate companies.
Shared visibility could improve accountability, but it does not by itself establish that generated code is correct or safe. A readable diff and a working preview can reveal some defects while missing security vulnerabilities, faulty edge cases, licensing problems or behavior that appears acceptable in a narrow test. The source reports that production pushes still require human sign-off, which is a meaningful control, but it does not explain who must approve them, what evidence reviewers receive or whether approvals can be delegated or rushed.
The reported partner participation could also affect interoperability and market structure. Slack Code is described as supporting multiple external agents rather than requiring one model provider. That could give organizations more choice, but it may also create uneven data handling, permissions and audit behavior across integrations. The source does not disclose pricing, data-retention terms, geographic availability, model routing, training policies or contractual protections for proprietary source code. Those omissions limit what can be concluded about the product’s practical value and risk.
互动机制：它实际上是如何运作的
以交互方式探索这一发展背后的基础技术。
crm_get_transaction(id='4092').An agent must create a draft calendar event for Tuesday at 2 p.m. Which evidence would establish the requested result?
接下来看什么
Key unknowns include the ’s availability, pricing, supported plans, technical limits and the degree of access each connected agent receives. The report does not provide independent testing of code quality, preview reliability, security controls or productivity effects. Teams should watch whether human approval remains meaningful in practice, how audit records are retained, and whether shared channels reduce coordination costs without increasing review burden or exposing sensitive code.
The first issue to verify is rollout scope. IT Brief New Zealand calls Slack Code a launch but does not say whether it is generally available, limited to a preview, restricted to certain enterprise plans or available in particular regions. Follow-up reporting should establish supported Slack editions, integration requirements, deployment timelines and any usage or retention limits. These details will determine whether the announcement represents an immediately usable product or an early platform direction.
Security and governance deserve close scrutiny because code channels combine business conversation, source code and agents that may be able to act on connected systems. The report says agents inherit Slack’s existing permissions and administrative controls, and that production pushes require human approval. It does not specify how credentials are isolated, whether agents can access private repositories, how prompts and outputs are logged, or how administrators revoke access. Independent security testing and clear documentation would be needed to evaluate whether inherited permissions are sufficient for agentic coding workflows.
The practical test will be whether teams can review agent work efficiently. Useful evidence would include error rates, rollback frequency, time spent reviewing diffs, preview failures, effects on incident rates and results across projects of different sizes. The source supplies no such measurements and includes no independent customer evaluation. Claims that the product makes development more collaborative therefore remain a stated product rationale, not an established outcome.
Finally, observers should watch how Slack’s agent directory and partner interfaces develop. Broader support could make it easier to switch between coding systems, but it could also increase fragmentation if agents expose different controls, records or approval flows. The important question is whether a shared channel gives humans meaningful oversight of AI-generated changes or simply makes a growing volume of automated work easier to observe. The report does not yet answer that question.