Coze MCP. Manage bot lifecycle and knowledge bases with code.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Coze. This MCP Server gives your AI agent full control over the Coze bot platform. You can manage conversation history, list workspaces, upload documents to knowledge bases, and publish bot drafts.
It's how you connect your agent to the entire lifecycle of a Coze bot, from chat interaction to data ingestion and deployment.
What your AI agents can do
Clear conversation
Removes all messages from a specific chat session.
Create chat
Starts a chat with a Coze bot and returns the initial response.
Delete document
Removes a document from a knowledge base using its unique ID.
Your agent can start a chat with a bot, retrieve the full message history, and clear the conversation record when needed.
The agent can upload raw text or files from URLs, list existing knowledge base datasets, and delete outdated documents.
It allows the agent to list all available Coze workspaces/spaces and check which bots are published within a specific space.
The agent can publish a draft bot or submit external tool outputs required by a bot's plugin logic.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Coze MCP Server: 11 Tools for Bot & Data Management
These tools allow your agent to manage the entire Coze bot lifecycle, from uploading source documents to starting conversations and publishing live bots.
019d842aclear conversation
Removes all messages from a specific chat session.
019d842acreate chat
Starts a chat with a Coze bot and returns the initial response.
019d842adelete document
Removes a document from a knowledge base using its unique ID.
019d842aget conversation history
Retrieves the complete message list for a given conversation.
019d842alist bots
Shows all published bots within a specific Coze Space.
019d842alist datasets
Lists all available knowledge base datasets in a Coze Space.
019d842alist workspaces
Shows all available Coze workspaces (spaces) associated with your account.
019d842apublish bot
Takes a draft bot and makes it live and accessible to users.
019d842asubmit tool outputs
Sends external data or results back to a bot when it requires human or plugin-driven input.
019d842aupload document
Uploads a raw text document directly to a knowledge base dataset.
019d842aupload file url
Uploads a file by providing an external URL to Coze storage.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Coze, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
This MCP Server gives your AI agent full control over the Coze bot platform. You can manage everything from live chat interactions to deploying finished bots and handling knowledge documents. Your agent can start a chat with a bot using create_chat and then pull the complete message list with get_conversation_history.
When the conversation is done, it can wipe the record clean with clear_conversation. To handle the bot's knowledge, your agent can upload raw text using upload_document or give it files via external URL with upload_file_url. It can also check what knowledge base datasets exist with list_datasets and take out outdated docs using delete_document.
To see the bot environment, your agent can list all available Coze workspaces with list_workspaces and check which bots are live in a specific space using list_bots. The agent can take a draft bot and make it live and accessible with publish_bot. When a bot needs external input for its plugin logic, the agent sends the data back using submit_tool_outputs.
Finally, the agent can show the bot's full capabilities by listing all available bots and datasets.
How Coze MCP Works
- 1 Subscribe to the server and get a Personal Access Token (PAT) from Coze.
- 2 Configure your AI client to use the correct Base URL and credentials.
- 3 Your agent then calls the tools (like
list_workspacesorcreate_chat) to interact with the Coze API directly.
The bottom line is, you give your agent the necessary credentials, and it uses the 11 available tools to manage the Coze platform like a native user.
Who Is Coze MCP For?
Knowledge Engineers and RAG Developers need this. If your job involves connecting multiple data sources to an AI model, you need to manage the data lifecycle programmatically. This tool suite lets you build, test, and deploy complex bots without manual clicking or copy-pasting through multiple dashboards.
Build complex, multi-step AI pipelines that require dynamic management of knowledge bases and bot state.
Automate the ingestion of documentation across multiple datasets and track which documents are currently available for AI retrieval.
Connect disparate systems by using the bot APIs to manage data flow and trigger external actions.
What Changes When You Connect
- You get full control over bot interactions. Use
create_chatandget_conversation_historyto manage multi-turn conversations, keeping a perfect record of every user query. - Knowledge base management becomes programmatic. Tools like
upload_documentanddelete_documentlet your agent index new documentation or clean up old, irrelevant files in bulk. - Monitor your entire bot ecosystem with simple calls.
list_workspacesandlist_datasetslet your agent see every available space and know exactly what knowledge bases exist. - Automate the deployment process. Instead of manually publishing, use
publish_botto take a draft bot and make it live instantly. - Handle complex workflows. If a bot needs external data (like a calculation), use
submit_tool_outputsto feed the result back into the conversation without human intervention. - Maintain consistency. Your agent can
clear_conversationto start a new session orlist_botsto check exactly which bots are active in a given space.
Real-World Use Cases
Debugging a poor RAG response
A data scientist finds a bot giving irrelevant answers. They use list_datasets to check all available knowledge bases, then use get_conversation_history to pinpoint the exact turn where the context failed. Finally, they run delete_document to remove the garbage data causing the problem.
Onboarding a new corporate guide
The technical writer needs to make a new policy guide available to the bot. They use upload_document to push the raw text file into the correct dataset, then use list_datasets to confirm the document is indexed, and finally call publish_bot to make the updated bot active.
Tracking a complex service interaction
A support engineer needs to know what was discussed in a specific support chat. They use create_chat to simulate the interaction, then call get_conversation_history to pull the entire transcript, which they can then log into a CRM system.
Validating bot dependencies
A system architect wants to ensure a new bot can access all required knowledge. They use list_workspaces to see the environment, list_bots to check the specific bot's status, and list_datasets to confirm all required data sources are present before deployment.
The Tradeoffs
Manual API calls
A developer has to jump between the Coze UI, the API documentation, and their local script to manage document uploads and chat history. This requires copy-pasting IDs and dealing with multiple authentication steps.
→
Use upload_document or upload_file_url to push data directly, and then use get_conversation_history and clear_conversation within a single agent call to manage the full interaction lifecycle.
Treating the bot as a black box
Relying on the bot's built-in dashboard to publish updates or manage datasets. This means the agent has no control over the deployment timing or the data's source.
→
Use publish_bot to control the deployment moment, and manage all data via list_datasets and delete_document to ensure the knowledge base is exactly what the bot needs.
Forgetting external inputs
A bot requires a calculation result from an external service, but the developer doesn't know how to feed that output back into the chat flow. The bot simply fails.
→
When the bot needs external data, use submit_tool_outputs to feed the result back into the conversation, keeping the workflow continuous and automated.
When It Fits, When It Doesn't
Use this MCP Server if your primary job is managing the full lifecycle of bots built on the Coze platform. You need an agent that can programmatically interact with conversation history (get_conversation_history), ingest data (upload_document), and manage the deployment state (publish_bot). Don't use it if you only need to chat with a bot occasionally—use a simple messaging API wrapper instead. If your goal is pure data visualization or running complex database queries that aren't bot-related, you need a specialized database connector, not a bot orchestration tool.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Coze. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 11 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Managing Bot Data and Conversations is a Mess of Manual Clicks.
Today, updating a bot's knowledge base means logging into the Coze UI, navigating to the dataset, manually uploading a document, and then checking to make sure the new data was indexed correctly. Changing a bot's behavior often requires juggling multiple tabs just to get the right IDs for the API.
With the Coze MCP Server, your agent handles the whole process. It runs `upload_document` to ingest content and uses `list_datasets` to confirm the data is ready. You get a single, auditable, automated workflow instead of a dozen manual clicks.
Coze MCP Server: Control Bot Deployment and State
Manually publishing a bot draft means you have to wait for the UI to process the changes, and you have no way to programmatically verify if the bot's dependencies are ready. If you need to update the knowledge base first, you have to run those steps manually before hitting 'Publish'.
The Coze MCP Server lets your agent call `publish_bot` only after it has run `list_bots` and confirmed the environment is ready. You get deterministic, code-controlled deployment timing, period.
Common Questions About Coze MCP
How do I manage conversation history with the `get_conversation_history` tool? +
The get_conversation_history tool retrieves the full message list for a specific chat session. This lets your agent build a complete record of the conversation for logging or further analysis.
Can I use `upload_document` to add data to a dataset? +
Yes, upload_document pushes raw text content directly to a knowledge base dataset. This is the primary way your agent grounds the bot with new, custom information.
What is the difference between `list_datasets` and `list_workspaces`? +
The list_workspaces tool shows the high-level containers (spaces) where your bots live. list_datasets shows the specific knowledge base collections inside one of those spaces.
Does `create_chat` require a specific bot ID? +
Yes, you must provide a bot ID when calling create_chat so the agent knows which bot to talk to and what response to expect.
If I call `clear_conversation`, is the data permanently gone? +
Yes, clear_conversation removes all messages from the session. This is useful for starting fresh, but remember the history is not saved anywhere else.
How do I manage multiple knowledge sources using `list_datasets` and `upload_document`? +
You list datasets first to see what exists. Then, you use upload_document for each source you need to index. This lets you keep documentation separate and manage RAG context precisely.
What is the purpose of `submit_tool_outputs` when I use `create_chat`? +
You submit outputs when the bot calls a tool that needs external results. It passes required data back to the bot, letting it finish its response or take a final action.
If I need to publish a new bot, do I use `publish_bot` or `list_bots`? +
list_bots only shows existing bots in a space. You must use publish_bot to take a draft and make it live for others to chat with.
Which Base URL should I use for my account? +
If you are using the international version, use https://api.coze.com. For the Chinese version, use https://api.coze.cn.
Can I automatically list all published bots in a space? +
Yes! Use the list_bots tool with your Space ID. Your agent will return a list of all bots that have been published and are ready for interaction.
How do I upload a new document to my knowledge base? +
Use the upload_document tool with the target Dataset ID and the raw text content. Your agent will handle the ingestion process into the Coze RAG engine.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
HERE (Location & Maps)
Build with location data via HERE — geocode addresses, calculate routes, track traffic, and get weather.
Upstash Redis
Equip your AI to directly query, manage, and manipulate key-value data structures inside your serverless Upstash Redis databases.
Lyft
AI ride management: request rides, estimate costs, and track trips via agents.
You might also like
Clockify
Manage time tracking and projects via Clockify — track entries, monitor projects, and audit team hours directly from any AI agent.
Attio
Manage your CRM data with Attio — track objects, records, and relationships via AI.
Baidu Analytics / 百度统计
China's leading web analytics platform — audit traffic, track visitors, and query site metrics via AI.