Comparison
Choose who owns the operating burden.
Slack's official MCP path is real. The decision is whether your team should own runtime, tokens, rollout, and support—or buy a managed workflow layer that keeps those responsibilities together.
The market changed in February 2026
Slack launched an official MCP server and a Real-time Search API on February 17, 2026. That changes the wedge. The decision is no longer “Slack MCP or nothing.” It is “official or self-host is enough” versus “managed rollout and reliability is worth paying for.”
Use official or self-host when
- Your team already owns runtime, deployment, and token lifecycle internally.
- Buyer review and rollout sequencing are already handled inside your org.
- You need the full 21-tool OSS surface and want direct operator control.
- You are comfortable treating Slack MCP as infrastructure you maintain yourself.
Use managed Cloud when
- You want one remote endpoint with hosted credential handling and plan-aware account state.
- You need deployment review, procurement-ready security review, or bounded rollout help.
- You want Claude-first setup with Gemini CLI as a supported second path.
- You care about named workflow outcomes such as summary, action items, decisions, support triage, incident summary, and handoff summary.
Decision matrix
- Official / self-host: better when transport ownership is already solved and you do not need hosted inference.
- Free ($0): validate hosted inference — 2000 requests + 25 AI tool calls a month, no card.
- Pro ($19/mo): better when one operator wants unlimited inference tools, saved profiles, scheduled delivery, and opt-in encrypted credential storage.
- Team ($49/mo) flat: better when 2-10 people share the workflow surface and need synced profiles.
- Safeguard ($199/mo (waitlist)): agent approval gates, searchable workspace memory, and an audit log — the agent-oversight roadmap (waitlist).
- Ops engagement (quoted): better when SLA, custom retention, SOC2 evidence, or multi-tenant isolation become hard requirements — priced per organization, not a self-serve tier.
Commercial and compliance reality
Slack's developer terms matter if the product is commercially distributed. The right near-term move is not to out-"official" Slack. It is to out-operate the official path on workflow continuity, structured briefs (channel summaries, action items, decisions), and rollout support that the official path doesn't ship.
Named workflow outcomes
The commercial layer is not “more CRUD.” It is a productized workflow layer for support triage, incident comms, founder briefings, and product ops, backed by rollout and support boundaries.
Next step
If the team already knows it wants managed rollout, start Free, Pro, or Team on the first-party hosted surface. If the team is still deciding whether official/self-host is enough, use deployment review and the explicit security boundary.