Slack Automation via Rube MCP

Automate Slack workspace operations including messaging, search, channel management, and reaction workflows through Composio's Slack toolkit.

Prerequisites

• Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
• Active Slack connection via RUBE_MANAGE_CONNECTIONS with toolkit slack
• Always call RUBE_SEARCH_TOOLS first to get current tool schemas

Setup

Get Rube MCP: Add https://rube.app/mcp as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.

1. Verify Rube MCP is available by confirming RUBE_SEARCH_TOOLS responds
2. Call RUBE_MANAGE_CONNECTIONS with toolkit slack
3. If connection is not ACTIVE, follow the returned auth link to complete Slack OAuth
4. Confirm connection status shows ACTIVE before running any workflows

Core Workflows

1. Send Messages to Channels

When to use: User wants to post a message to a Slack channel or DM

Tool sequence:

1. SLACK_FIND_CHANNELS - Resolve channel name to channel ID [Prerequisite]
2. SLACK_LIST_ALL_CHANNELS - Fallback if FIND_CHANNELS returns empty/ambiguous results [Fallback]
3. SLACK_FIND_USERS - Resolve user for DMs or @mentions [Optional]
4. SLACK_OPEN_DM - Open/reuse a DM channel if messaging a user directly [Optional]
5. SLACK_SEND_MESSAGE - Post the message with resolved channel ID [Required]
6. SLACK_UPDATES_A_SLACK_MESSAGE - Edit the posted message if corrections needed [Optional]

Key parameters:

• channel: Channel ID or name (without '#' prefix)
• markdown_text: Preferred field for formatted messages (supports headers, bold, italic, code blocks)
• text: Raw text fallback (deprecated in favor of markdown_text)
• thread_ts: Timestamp of parent message to reply in a thread
• blocks: Block Kit layout blocks (deprecated, use markdown_text)

Pitfalls:

• SLACK_FIND_CHANNELS requires query param; missing it errors with "Invalid request data provided"
• SLACK_SEND_MESSAGE requires valid channel plus one of markdown_text/text/blocks/attachments
• Invalid block payloads return error=invalid_blocks (max 50 blocks)
• Replies become top-level posts if thread_ts is omitted
• Persist response.data.channel and response.data.message.ts from SEND_MESSAGE for edit/thread operations

2. Search Messages and Conversations

When to use: User wants to find specific messages across the workspace

Tool sequence:

1. SLACK_FIND_CHANNELS - Resolve channel for scoped search with in:#channel [Optional]
2. SLACK_FIND_USERS - Resolve user for author filter with from:@user [Optional]
3. SLACK_SEARCH_MESSAGES - Run keyword search across accessible conversations [Required]
4. SLACK_FETCH_MESSAGE_THREAD_FROM_A_CONVERSATION - Expand threads for relevant hits [Required]

Key parameters:

• query: Search string supporting modifiers (in:#channel, from:@user, before:YYYY-MM-DD, after:YYYY-MM-DD, has:link, has:file)
• count: Results per page (max 100), or total with auto_paginate=true
• sort: 'score' (relevance) or 'timestamp' (chronological)
• sort_dir: 'asc' or 'desc'

Pitfalls:

• Validation fails if query is missing/empty
• ok=true can still mean no hits (response.data.messages.total=0)
• Matches are under response.data.messages.matches (sometimes also response.data_preview.messages.matches)
• match.text may be empty/truncated; key info can appear in matches[].attachments[]
• Thread expansion via FETCH_MESSAGE_THREAD can truncate when response.data.has_more=true; paginate via response_metadata.next_cursor

3. Manage Channels and Users

When to use: User wants to list channels, users, or workspace info

Tool sequence:

1. SLACK_FETCH_TEAM_INFO - Validate connectivity and get workspace identity [Required]
2. SLACK_LIST_ALL_CHANNELS - Enumerate public channels [Required]
3. SLACK_LIST_CONVERSATIONS - Include private channels and DMs [Optional]
4. SLACK_LIST_ALL_USERS - List workspace members [Required]
5. SLACK_RETRIEVE_CONVERSATION_INFORMATION - Get detailed channel metadata [Optional]
6. SLACK_LIST_USER_GROUPS_FOR_TEAM_WITH_OPTIONS - List user groups [Optional]

Key parameters:

• cursor: Pagination cursor from response_metadata.next_cursor
• limit: Results per page (default varies; set explicitly for large workspaces)
• types: Channel types filter ('public_channel', 'private_channel', 'im', 'mpim')

Pitfalls:

…