Feature Description

Allow agents to compose and send Slack Block Kit messages for richer, more interactive responses.

Use Case

When an AI agent responds in Slack, it currently sends plain text (with markdown). For many use cases — CRM summaries, daily briefings, database query results, action confirmations — Block Kit would provide a much better UX:

• Sections with fields for structured data (e.g., contact cards, org summaries)
• Dividers and headers for visual hierarchy in long responses
• Buttons for quick actions (e.g., "Archive this email", "Create a task", "Approve")
• Select menus for choosing from options (e.g., "Which project?" dropdowns)
• Context blocks for metadata (timestamps, sources, status indicators)
Current State

• PR #2124 adds Block Kit for exec approval buttons (scoped use case)
• Issue #4391 adds streaming support
• General agent responses are plain text via chat.postMessage
Proposed Approach

Allow agents to output structured Block Kit JSON (perhaps via a special tool or message format) that gets sent as blocks in chat.postMessage. This could work alongside plain text as a fallback.

Possible approaches:

1. Tool-based: A slack_blocks tool that accepts Block Kit JSON
2. Markup-based: Agent outputs a special syntax that gets converted to blocks
3. Hybrid: Agent outputs markdown that gets auto-converted to blocks for supported platforms
   Context

Using OpenClaw as a team assistant connected to Slack, Notion, Gmail, and a Supabase CRM. Richer Slack output would significantly improve the experience for non-technical team members who interact with the agent daily.

Environment

• OpenClaw 2026.2.6-3
• Slack socket mode
• Multiple channels with per-channel sessions