Pumble
Pumble is a team communication platform for channels, direct messages, scheduled messages, and workspace collaboration.
Verdict
Common use cases
- Summarize a week of channel discussion
- Find who approved a feature request
- Clean up outdated pinned announcements
- Pull customer feedback from support chat
- Audit compliance-sensitive conversations
Integration
- Vendor
- Pumble
- Category
- communication
- Auth
- API_KEY
- Tools
- 9
- Composio slug
pumble
Tools
- Delete Messagedestructive
Permanently delete one Pumble message identified by channel ID and message ID.
- Edit Message
Replace the text of an existing Pumble message in a channel.
- Get Current User
Return the Pumble user and workspace associated with the connected API key. Use this to identify the connected account or verify its credentials.
- Get Message
Fetch one Pumble message by its message ID and channel ID.
- List Channel Messages
Return one page of recent messages from a channel, continuing backward through history with a cursor.
- List Channels
List channels visible to the connected Pumble user, including channel IDs needed by message tools.
- List Users
List users in the current Pumble workspace with stable IDs and profile details.
- Search Messages
Search Pumble messages and return one bounded page. For the first page, provide at least one of text, author_ids, or channel_ids; for later pages, pass the returned cursor alone.
- Send Message
Immediately post a text message to a Pumble channel identified by its channel ID.
Setup
Setup guide
- 11. Open your Pumble workspace settings and navigate to the API section to generate a new API key with read and write permissions. 2. Copy the key immediately because Pumble shows it only once. 3. In Switchy, click Integrations in the left sidebar, find Pumble, and click Connect. 4. Paste your API key into the prompt and click Authorize. 5. Switchy will verify the key by fetching your user profile; if successful, the integration appears as connected. 6. Open any Project, type '@Pumble list channels' in the chat, and confirm the AI returns your workspace channels. 7. To test message access, pick a channel ID from the list and try '@Pumble list messages in channel <ID>' to see recent conversation history. 8. Remember that this key grants full message access, so store it securely and rotate it if team members leave.
What teammates see: by default, memories from Pumble are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Projects have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Weekly Channel Digest
@Pumble list channels, then fetch the last week of messages from our #product-planning channel and summarize the three most important decisions and any open action items.
Search Past Decisions
@Pumble search all messages for 'pricing model' and show me the three most relevant results with enough context to understand what was decided.
Find Message Author
@Pumble search messages for 'we should delay the launch' and tell me who wrote it and in which channel.
Bulk Edit Announcements
@Pumble search for messages containing 'office hours are 9-5' and edit each one to say 'office hours are now 10-6 starting Monday'.
Audit Compliance Thread
@Pumble list all messages in the #legal-review channel and format them as a timestamped audit log with author names.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@pumble search for messages mentioning 'Q1 roadmap' in the #product channel from the last two weeks
Found 7 messages matching 'Q1 roadmap' in #product. The most recent thread (3 days ago) has Sarah proposing three feature tiers, with replies from Mike and Jordan debating timeline feasibility. An earlier message (9 days ago) links to a Figma file with wireframes. Two other mentions appear in side discussions about budget allocation. This illustrative example shows how Pumble's search returns message snippets with author names and timestamps; actual results depend on your workspace's message history and channel access.
This example demonstrates read-only search across channel history. The MCP requires a valid channel ID (obtained via List Channels first) and respects your API key's workspace permissions—you'll only see messages in channels you can access. Search is paginated; large result sets require multiple calls with cursors.
@pumble post a message to #engineering saying 'Deployment window starts at 2pm ET today—staging env will be unavailable for ~30min'
Message posted to #engineering at 1:47pm ET. The message ID is msg_8x2kp9 in channel ch_eng_001. Pumble will notify all channel members per their notification settings. This illustrative example shows a successful write operation; the actual message appears immediately in your team's Pumble workspace. Note that the MCP does not support rich formatting (bold, links, mentions) in this release—plain text only.
This example demonstrates the Send Message tool, which creates a new message in a specified channel. You must provide a valid channel ID; the MCP does not auto-resolve channel names like '#engineering'. The operation is immediate and irreversible without a follow-up Delete Message call. API rate limits apply (typically 60 requests/minute per workspace).
@pumble list all messages in #support from the past 24 hours, then summarize the top three customer issues by frequency
Retrieved 42 messages from #support (past 24 hours). After analyzing message content, the three most frequent issues are: (1) Login failures on mobile app (mentioned in 14 messages, affecting iOS users on v2.3.1), (2) Missing email notifications for new replies (8 messages, appears workspace-wide), (3) Slow file uploads over 10MB (6 messages, primarily EU users). This illustrative example shows the AI synthesizing raw message data into actionable insights; actual results depend on your support channel's activity and message volume.
This example pairs the List Channel Messages tool with AI reasoning to extract patterns from conversation threads. The MCP returns raw message text and metadata; the AI performs the summarization. Pagination limits mean you'll retrieve messages in chunks (default 50 per page)—for high-traffic channels, you may need multiple calls to cover 24 hours. The synthesis quality depends on message clarity and thread structure.
Use-case deep-dives
When Pumble wins for customer-support context capture
A 6-person support team uses Pumble for internal triage and needs to pull conversation threads into Linear tickets when escalating to engineering. The Search Messages and Get Message tools let an AI assistant find the relevant Pumble discussion by customer name or ticket ID, then copy the thread into the Linear description field in one step. This works cleanly when your support volume stays under 200 tickets a week and the handoff pattern is consistent—search by customer name, grab the last 10 messages, paste. If your team juggles multiple Pumble workspaces or needs to search across 50-plus channels daily, the single-workspace scope and page-by-page message retrieval start to slow you down. For small teams with a tight support-to-eng loop, Pumble MCP turns handoff documentation from a copy-paste chore into a two-click lookup.
When this MCP fits distributed standup summaries
A 4-person product team runs async standups in a dedicated Pumble channel and wants an AI to compile the last 24 hours of updates into a single summary doc every morning. List Channel Messages pulls the recent thread, and the AI condenses blockers, wins, and next steps into a shared Notion page. This scenario is ideal when your standup channel sees 15 to 30 messages a day and the team sticks to one workspace. The MCP's cursor-based pagination handles the volume without rate-limit drama, and the Edit Message tool lets the AI post a summary back into Pumble if you prefer in-app digests. If your standup spans three channels or mixes Pumble with Slack, the single-platform scope becomes a friction point. For small remote teams anchored in Pumble, this MCP automates the daily standup recap without adding another tool to the stack.
When Pumble search beats a static wiki for new hires
A 10-person startup keeps onboarding FAQs and process notes scattered across Pumble channels instead of maintaining a formal wiki. New hires ask the AI assistant questions like 'how do we handle PTO requests' and the assistant uses Search Messages to pull the canonical answer from the #team-ops channel, then surfaces the message link and author. This works when your onboarding corpus lives in under 10 channels and the team updates answers in-thread rather than editing old docs. The List Users tool helps the AI tag the right person if the answer is outdated or ambiguous. If your company has 50-plus employees or onboarding spans multiple workspaces, the search scope gets too narrow and you'll want a dedicated knowledge base. For early-stage teams that live in Pumble, this MCP turns your chat history into a queryable onboarding layer without the overhead of Notion or Confluence.
Frequently asked
What does the Pumble MCP let me do in Switchy?
The Pumble MCP connects your Pumble workspace to Switchy so AI agents can read, search, edit, and delete messages across your channels. It also lists users and channels, making it possible to automate message workflows or pull conversation context into other tools without switching apps.
Do I need admin access to connect Pumble via API key?
You need a Pumble API key tied to a user account with permission to view the channels and messages you want Switchy to access. Pumble's API key inherits the permissions of the user who generated it, so if that user can't see a private channel, neither can the MCP.
Can the Pumble MCP send new messages or only read existing ones?
The current tool set focuses on reading, searching, editing, and deleting messages—there's no dedicated send-message tool listed. If you need to post new messages programmatically, you'll want to use Pumble's API directly or request that feature from the MCP maintainer.
Why use this MCP instead of Pumble's web app or API?
The MCP wraps Pumble's API into a format AI agents understand natively, so you can chain message searches with other Switchy integrations—pull a Pumble thread, summarise it, then file the summary in Notion—without writing glue code. The web app is still faster for ad-hoc browsing.
Who on my team should connect the Pumble integration?
Whoever generates the API key determines which channels and messages Switchy can see. If your team uses private channels heavily, connect it under a user account that belongs to those channels. One connection covers the whole Switchy workspace; you don't need separate keys per teammate.