Outside the Terminal, a Team and Its AI Gather Around One Screen — Slack Opens a 'Channel Just for Code,' Pulling Agent Work into the Group Chat

Announced by Slack on August 20, 2026, "Slack Code" is a new channel that pulls coding-agent work out of the terminal and into the group chat. We follow this new shape of team development, where everyone gathers around diffs, previews, and human sign-off.

Share
Outside the Terminal, a Team and Its AI Gather Around One Screen — Slack Opens a 'Channel Just for Code,' Pulling Agent Work into the Group Chat

On August 20, 2026, Salesforce-owned Slack announced Slack Code, a new kind of channel where teams build software together with AI coding agents. It's an attempt to take the work that individual developers used to do face-to-face with a copilot inside their terminals and pull it straight into the middle of the group chat.

It's available starting the same day. Rather than being enterprise-only, it works on every Slack plan, including the free tier. No additional pricing has been announced.

From "a Personal Sidekick" to "a Shared Workspace"

Traditional coding agents have been, fundamentally, one-to-one tools. Developers scatter into their own terminals, write their assigned piece, submit their code, and review it afterward. Slack Code replaces that division-of-labor flow with one where "everyone moves forward while looking at the same conversation."

Salesforce CMO Ryan Gavin framed the intent this way: "Slack Code advances AI from an individual copilot to embedded AI orchestration grounded in context." The idea is not to hold onto a sidekick one at a time, but to build the agent into the flow of the team's conversation itself.

One Tag Spins Up a "Code Channel"

Using it is simple. When you tag a coding agent within a conversation, that agent spins up a dedicated code channel for the project on its own, carves out a workspace, and gets to work. What's distinctive is that the agent creates the channel, not a person. The work output is displayed on a canvas, and everyone gathers around the same screen.

Within the channel, here's the gist of what a team can do:

  • Reviewing diffs: follow proposed changes on the spot as before/after.
  • Live preview: see output like HTML in a working state and confirm it.
  • Visible conversation: the back-and-forth between people and the agent is visible to everyone, so feedback can be dropped in midway.
  • Incorporating feedback: the agent takes in cross-functional input right then and there.

Governance is built in from the start. Before code goes out into the world, human sign-off is required, and once the work is done the channel is archived automatically. On top of that, an audit log recording "who requested it, what changed, and who approved it" is retained permanently.

Supported Agents, and a Low Barrier to Adoption

The initial lineup of support features the major coding agents. It's a configuration aimed at being a vendor-neutral "broker."

ItemDetails
Initially supported agentsAnthropic's Claude Code, Cognition's Devin, GitHub Copilot, Vercel's agent, and more
ExtensibilityOther agents can connect via Slack's 2,600-plus app integrations or through MCP (Model Context Protocol)
AvailabilityAugust 20, 2026
Eligible plansAll Slack plans, including free. No additional pricing announced

Putting multiple agents on the same footing and opening up externally via MCP is a design that hints at an intent to grow Slack into a "common gateway from which you can summon any agent." With the barrier to adoption low, it should be easy to start by trying small internal tools or building screens for validation first.

Bringing Review Back to the "Meeting Room"

What this mechanism changes is less the speed of writing code and more the place where consensus is built. Slack Code pulls the confirmation work that used to be confined to pull requests and review screens back into the chat where the team normally lives. Salesforce explains this as an effect that "catches misalignment early and reduces lifecycle churn."

On the business side, a big benefit is that non-engineers — planning, design, the business side — can chime in on the spot while looking at diffs and previews. If discrepancies in specifications are caught "while it's being built" rather than "after implementation is finished," the total amount of rework goes down. And because it's usable even on the free plan, the low barrier that lets small teams and solo operators try it out shouldn't be overlooked either.

At the same time, Slack acknowledges this isn't a cure-all. According to SiliconANGLE, Slack Code "isn't meant to completely replace the deliberate, heads-down coding work of building things in a terminal or IDE." It's positioned strictly as a tool that thickens the layer of collaboration and consensus-building.

What Lies Beyond "Everyone Can See It"

Transparency is Slack Code's selling point, but there are things worth thinking about behind the convenience. Moving review into the group chat also means, flipped around, that approval can easily turn into a rote motion. As a general observation, the reassurance that many eyes are watching can itself breed an atmosphere of "surely someone is looking at it properly."

Also, the pathway where an agent writes code directly from chat and, after approval, proceeds all the way to shipping — even guarded by audit logs and sign-offs — requires careful management of permission design and deployment targets. The more you take in the convenience, the more the operational rules that keep "who bears ultimate responsibility" clear on the human side come to matter. The realistic approach is to adopt it as a set, along with measures that avoid over-reliance and don't let the quality of approval slip.

Even so, in pulling coding agents out from being "a tool you run alone" to being "a workspace the team gathers around," Slack Code is a move that shifts the center of gravity of the development workflow. When a place for review and consensus is born outside the terminal, how will the way development teams work change? The real-world operation of the next few months should show us the answer.

References: Introducing Slack Code (Salesforce) / Salesforce introduces Slack Code (SiliconANGLE) / Slack Code Puts AI Coding Agents in Dedicated Project Channels (Unite.AI) / Slack Launches Code For AI Coding Collaboration (Dataconomy) / Slack has a new channel type (The New Stack)

Read more

Making It Wait for "Jobs That Run Over an Hour": Codex 0.152 Adds Ceiling Dials for MCP Output Volume and Execution Time, and Turns the Planning Tool Off by Default

Making It Wait for "Jobs That Run Over an Hour": Codex 0.152 Adds Ceiling Dials for MCP Output Volume and Execution Time, and Turns the Planning Tool Off by Default

Codex v0.152.0 on August 31 and its next-day fix release added explicit ceilings on MCP tool output volume and execution time, and switched the planning tool off by default. Here's a rundown of the changes that matter for long-running unattended and semi-autonomous agent operation.

By FF
The CLI's Default Model Just Swapped In a Million-Token Brain — Claude Code v2.1.257 Makes Fable 5.1 the Standard and Adds a 'Containment Escape' Checkpoint to Auto Mode

The CLI's Default Model Just Swapped In a Million-Token Brain — Claude Code v2.1.257 Makes Fable 5.1 the Standard and Adds a 'Containment Escape' Checkpoint to Auto Mode

Claude Code v2.1.257, released September 1, 2026, swaps its default model to Fable 5.1 with its one-million-token context. It also adds guardrails to auto mode that stop credential retrieval and out-of-scope reads from slipping through. Here's a rundown of the changes that matter to developers.

By FF
"This Is an Authorized Exercise"—How the Aurora Ransomware Gang Insisted, While Making Cursor's AI Agent Do the Actual Intrusion Work

"This Is an Authorized Exercise"—How the Aurora Ransomware Gang Insisted, While Making Cursor's AI Agent Do the Actual Intrusion Work

Gambit Security and CloudSEK report that the ransomware group Aurora abused Cursor's AI agent for real intrusion work. Posing the tasks as an "authorized exercise" to slip past the safeguards, they had it handle reconnaissance and privilege takeover on the back of stolen credentials—a warning that a

By FF
One in Three Companies Now Choose to Build Rather Than Buy — McKinsey Measures How Coding Agents Are Reshaping the Procurement Decision

One in Three Companies Now Choose to Build Rather Than Buy — McKinsey Measures How Coding Agents Are Reshaping the Procurement Decision

McKinsey's annual survey found that about 30% of respondents passed on buying software because they could build it in-house with coding agents. We unpack the procurement shift from buying to building — and the current reality that productivity is up while profits stay flat.

By FF