GPTBots
GPTBots is an enterprise AI agent platform for conversations, knowledge management, databases, user data, and usage analytics. This toolkit auth config supports Singapore Agent API Keys only; JP and TH keys are unsupported because requests use the fixed Singapore base URL.
Verdict
Common use cases
- Export bot transcripts for training review
- Audit credit burn by conversation topic
- Debug why a user query failed
- Pull knowledge base configs before migration
- Track conversation volume by time window
Integration
- Vendor
- GPTBots
- Category
- other
- Auth
- API_KEY
- Tools
- 12
- Composio slug
gptbots
Tools
- Create Conversation
Create a GPTBots conversation for a developer-defined user on the Agent connected by this API key. Returns the conversation ID needed for subsequent message and history operations. GPTBots documents a shared default limit of 3 requests per
- Get Agent Information
Return the GPTBots Agent bound to the connected Agent API Key, including its ID and runtime configuration. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free
- Get Total Credit Consumption
Return aggregate Agent credit consumption by category for a reporting range of up to 365 days. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan, serial
- List Conversation Credits
Return one page of per-conversation Agent credit consumption for a time range of up to 30 days, optionally filtered by conversation ID. GPTBots code 40001 means the time range is invalid and must be corrected, not retried. GPTBots documents
- List Conversation Messages
Return one page of current v2 message details for a GPTBots conversation. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan, serialize calls and back of
- List Conversations
Return one page of Agent conversations filtered by source, recent-chat time range, and optional user. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan,
- List Database Tables
Return database tables configured for the Agent ID supplied by the caller. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan, serialize calls and back o
- List Knowledge Bases
Return all knowledge bases attached to the GPTBots Agent connected by this API key. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan, serialize calls a
- List Knowledge Documents
Return one page of documents in a GPTBots knowledge base. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free Plan, serialize calls and back off on HTTP 429 or
- List Metadata Fields
Return global and per-knowledge-base document metadata field definitions for the connected GPTBots Agent. GPTBots documents a shared default limit of 3 requests per minute for Free Plan accounts; higher or custom tiers may differ. On Free P
- Query User Attributes
Return configured GPTBots Agent attributes for either developer user IDs or anonymous IDs. Exactly one non-empty selector is required. Missing identities are successful results with no property values. GPTBots documents a shared default lim
- Send Message
Send the latest user text to a GPTBots conversation and wait for the Agent's complete blocking response. This tool does not use streaming, webhooks, multimodal files, or human handoff. It consumes credits and creates a permanent message. Do
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to the MCP Integrations panel. 2. Click 'Add Integration' and select GPTBots from the list. 3. Log into your GPTBots account, navigate to the Agent you want to connect, and copy its API key from the developer settings page. 4. Paste the API key into Switchy's configuration dialog and click 'Connect'. 5. Switchy will verify the key by fetching your Agent's information; if successful, the MCP appears in your integration list. 6. Inside any Project, type '@GPTBots get agent information' to confirm the connection returns your Agent ID and config. 7. To test conversation tools, first create a conversation with '@GPTBots create conversation for user_id test_user', note the returned conversation ID, then retrieve its messages with '@GPTBots list conversation messages conversation_id <ID>'. 8. If you hit rate limits, GPTBots will return code 40001; upgrade your plan or space out requests.
What teammates see: by default, memories from GPTBots 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
Audit This Week's Spend
@GPTBots get total credit consumption for the last 7 days. Show me which categories used the most credits and flag any spikes.
Export Recent Conversations
@GPTBots list conversations from the last 48 hours. Include user IDs and recent-chat timestamps so I can prioritize which to review.
Retrieve Full Transcript
@GPTBots list conversation messages for conversation_id <paste_ID_here>. Show the full exchange so I can trace where the bot went off track.
Check Agent Configuration
@GPTBots get agent information. Tell me the Agent ID, model version, and which knowledge bases are currently attached.
List Knowledge Base Sources
@GPTBots list knowledge bases. Show me the name and status of each so I can confirm the latest product docs are live.
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@GPTBots list all conversations from the last 7 days and show me which ones had the highest credit consumption
Found 23 conversations in the past week. Top 3 by credit usage: Conversation abc123 (source: web, 145 credits, last active 2 days ago), Conversation def456 (source: api, 112 credits, last active 4 days ago), Conversation ghi789 (source: slack, 89 credits, last active yesterday). The web conversation included multiple file uploads and knowledge-base queries, which drove higher costs. Total consumption across all conversations: 1,247 credits.
Combines List Conversations with List Conversation Credits to surface usage patterns. Requires an Agent API Key with read access. Free Plan accounts hit a 3 req/min rate limit, so batching 23 conversations may take multiple calls. Useful for cost monitoring before month-end billing.
@GPTBots create a new conversation for user emma@example.com and send her an onboarding message explaining our return policy
Created conversation conv_9x8y7z for emma@example.com. Sent message: 'Welcome! Our return policy allows 30-day returns on unused items with original packaging. Refunds process within 5–7 business days. Need help with a specific order? Just ask.' Conversation ID conv_9x8y7z is now active and ready for follow-up messages.
Demonstrates the Create Conversation tool followed by a message send (via the Agent's runtime). This writes state to your GPTBots account and may consume credits immediately. Confirm your Agent's prompt and knowledge base are configured before automating customer-facing messages.
@GPTBots pull the agent configuration and all attached knowledge bases, then draft a one-paragraph summary of what this bot is trained to do
Your Agent (ID: agt_abc123) is configured as a customer-support assistant with three knowledge bases attached: Product Catalog (142 documents), FAQ Archive (89 documents), and Shipping Policies (12 documents). Based on this setup, the bot is trained to answer product questions, troubleshoot common issues, and provide shipping and return guidance. It uses GPT-4 with a temperature of 0.7 and has access to a structured database of order history for personalized responses.
Chains Get Agent Information and List Knowledge Bases to give you a high-level view of your bot's capabilities. Helpful when onboarding teammates or auditing scope before a launch. Does not consume conversation credits, but still counts against the 3 req/min rate limit on Free Plan.
Use-case deep-dives
When you need to start GPTBots chats from your own app
A 5-person SaaS team wants users to click a help button in their dashboard and land in a GPTBots conversation that already knows their account context. The Create Conversation tool lets you spin up a new chat, pass in a user ID, and return the conversation link—all from a Switchy prompt or a backend script. You write a prompt that reads the user's recent activity from your database, creates the conversation, and drops the link into Slack for the support rep to pick up. The boundary: if your users are starting chats from a static widget on your marketing site, you don't need this—GPTBots handles that natively. If you're embedding the bot in an authenticated app and want to pre-seed context, this MCP is the bridge. Wire it up and stop asking users to repeat themselves.
Frequently asked
What does the GPTBots MCP let me do in Switchy?
It connects your GPTBots agents so your team can create conversations, send messages, pull chat history, and track credit consumption directly from Switchy prompts. You can also inspect agent configuration, list knowledge bases, and query database tables. Think of it as programmatic access to your GPTBots deployment without switching tabs or writing API wrappers.
Do I need a paid GPTBots plan to use this MCP?
No, but the free tier limits you to 3 requests per minute across all endpoints, which will feel slow if multiple teammates are querying the same agent. Paid and custom tiers lift that ceiling. You'll need an Agent API key from GPTBots regardless of plan; generate it in your GPTBots dashboard under API settings.
Can the MCP send messages on behalf of different users?
Yes. The Create Conversation tool accepts a developer-defined user identifier, so you can spin up separate conversation threads for each person on your team or each end customer. GPTBots tracks these as distinct sessions, which means credit consumption and history stay isolated per user.
How is this different from embedding a GPTBots widget on my site?
The widget is for end-user chat; this MCP is for backend automation and reporting. Use it when you want to trigger agent responses from Switchy workflows, audit credit spend across conversations, or pull chat transcripts into your own tools. The widget can't inspect knowledge bases or list database tables.
Who on the team should connect the GPTBots API key?
Whoever owns your GPTBots account and can generate Agent API keys. That key grants read and write access to conversations, credit data, and agent configuration, so treat it like a production credential. If you rotate keys, you'll need to reconnect the MCP in Switchy's integration settings.