ThinkStack MCP. Control chatbot knowledge and conversation flow.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
ThinkStack MCP Server lets you programmatically manage AI chatbots and their knowledge bases through your agent. Use it to list, add, or remove documentation sources; check conversation history; and send queries to any bot in real time.
It gives your AI client full control over the source material and operational state of your bots.
What your AI agents can do
Add source
Crawls a provided URL or document path and adds it as new knowledge material for the bot.
Check thinkstack status
Runs a simple check to confirm that the ThinkStack API is online and ready for use.
Delete source
Permanently removes an existing knowledge source from the bot's documentation base.
Send a prompt to a specific chatbot and receive an AI-generated response based on its configured knowledge base.
Add new URLs or documents to the bot's knowledge base, or remove old sources entirely.
Retrieve a list of all available chatbots and get detailed configuration information for any single bot.
Pull up details about past chat sessions, including full message threads and user context.
See all the external API actions a chatbot has configured (e.g., booking calendars or checking inventory).
Ask AI about this MCP
Supported MCP Clients
OAuth 2.0 Compatible Gemini Waiting for input…
ThinkStack MCP Server: 10 Tools for Bot Management
These ten tools let you programmatically list, modify, and interact with every component of your chatbot infrastructure, from sources to conversation history.
Make your AI actually useful.
Add this MCP to Claude, Cursor, or Windsurf and your AI stops guessing. It gets real tools to look things up, take action, and handle the stuff you keep doing by hand.
Start using ThinkStack on Vinkius019dd173add source
Crawls a provided URL or document path and adds it as new knowledge material for the bot.
019dd173check thinkstack status
Runs a simple check to confirm that the ThinkStack API is online and ready for use.
019dd173delete source
Permanently removes an existing knowledge source from the bot's documentation base.
019dd173get bot
Retrieves detailed metadata about a specific, named chatbot instance.
019dd173get conversation
Pulls the full message history and user details for a single chat session.
019dd173list actions
Lists all external API actions that the chatbot can trigger (like sending an email or creating a record).
019dd173list bots
Returns a list of every single chatbot configured under your ThinkStack account.
019dd173list conversations
Lists metadata for all past chat sessions, helping you find the session ID needed for `get_conversation`.
019dd173list sources
Shows a list of every knowledge source (URLs/documents) currently attached to your bots.
019dd173send query
Sends a message directly to the chatbot and gets an immediate, context-aware response.
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 ThinkStack, then connect any of our 4,800+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,800+ 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
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by ThinkStack. 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 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Checking if your bot's knowledge base is actually up-to-date shouldn't require a manual audit.
Today, if you suspect a chatbot answer is wrong, you have to manually check the admin dashboard. You click through tabs—Documentation Sources, Chat History, Bot Settings—just to confirm *why* it gave that response. It's a mess of clicks and cross-referencing.
With this MCP server, you run `list_sources` directly in your agent. In seconds, you see every single URL and document the bot relies on. This immediate visibility lets you prove whether the answer is flawed because the source material is stale or simply wrong.
Using send_query: Get a real-time response inside your workflow.
Before, if you needed an AI to perform a function—like retrieving details for customer X—you'd have to copy the data out of the chat UI, paste it into another system, and wait for two different systems to communicate. It was slow, fragile, and involved multiple human hands.
Now, your agent runs `send_query` directly. The bot handles the knowledge retrieval and response generation in a single, atomic step within your code. You get the answer instantly, without leaving the workflow.
What you can do with this MCP connector
ThinkStack MCP Server - Manage Bots & Knowledge
You're not just talking to a chatbot; you're running an operation. This server lets your agent treat AI chatbots like services you can fully manage from the backend. You gain direct, programmatic control over everything—from which knowledge sources are feeding their answers to reviewing every single conversation thread.
It gives your client total command of the bot's operational state and its core data. Here’s what you get access to:
Configuring and Checking Bots
To see what bots you've got running, just run list_bots; it spits out a list of every single chatbot configured under your account. Need the deep details on one specific bot? You use get_bot to pull detailed metadata about that named instance. Before you do any heavy lifting, you should check if the whole thing's up by calling check_thinkstack_status.
That simple call confirms the ThinkStack API is online and ready for work.
Controlling Knowledge Sources
ThinkStack lets you dictate exactly what material these bots use. You can see every single piece of documentation attached right now using list_sources, which shows a list of all the URLs or documents currently feeding your bots. If you find some old, dusty info that needs to go, you run delete_source and permanently yank it out of the bot's reference base.
Want to add new material? You use add_source; just give it a URL or a document path, and it crawls that stuff, adding it as fresh knowledge for your bots.
Querying and Interacting Live
When you want an immediate answer, you send a message directly using send_query. This sends the prompt to the chatbot and gets a context-aware response back right away. For knowing what tools are available, run list_actions; this shows every external API function—like booking calendars or checking inventory—that your chatbot can actually trigger when prompted.
Auditing Conversations
You gotta keep track of who's talking to whom and what they talked about. To see a list of past chats, you run list_conversations. This gives you the metadata for all those sessions, specifically providing the session ID you need for the next step. Once you have that ID, you use get_conversation to pull up the full message history and all the user details for that single chat session.
It’s your complete audit trail.
Summary of Control Flow
This server lets you list every bot (list_bots), get detailed info on one (get_bot), and then send it a prompt to get an answer (send_query). Meanwhile, if the knowledge is wrong, you can use list_sources and tell it to update itself with new content using add_source, or clean up old stuff with delete_source.
You've got full visibility over the system status via check_thinkstack_status, and you never lose a conversation because you can list all past chats (list_conversations) and retrieve the full thread details whenever you need it.
019dd173-e3ca-724b-b097-37c53a2c6a43 
How ThinkStack MCP Works
- 1 1. Subscribe to this server and retrieve your ThinkStack API Key.
- 2 2. Configure your agent client (Claude, Cursor, etc.) with the key.
3. Use tools likelist_botsoradd_sourcedirectly in conversation to manage the AI infrastructure.
The bottom line is that you treat the entire chatbot system as a database layer, managing sources and state before querying.
Who Is ThinkStack MCP For?
This server is for developers who need deep control over their AI services. It’s for the product manager whose job it is to prove whether the bot's knowledge base is accurate enough, or the support engineer who needs to audit exactly why a customer received an incorrect answer.
Using list_actions and add_source, they connect external services (like CRM APIs) to the bot's operational scope, making the bot do more than just chat.
They use list_bots and get_conversation to track usage patterns, spot common failure points in responses, and decide where to improve documentation sources.
They rely on check_thinkstack_status before any deployment or high-volume query run to ensure the whole stack is connected and stable.
What Changes When You Connect
- Know exactly what your bot knows. Use
list_sourcesto see every URL or document currently indexed before you trust its answer. - Audit chat performance with
get_conversation. Instead of relying on memory, you can pull the full message history and user metadata for any session. - Stop guessing if a feature exists. Run
list_actionsto view all external API endpoints your bot is wired to—whether it's booking or checking inventory. - Manage the source material programmatically. If documentation changes, you can use
delete_sourceandadd_sourcevia your agent workflow instead of logging into a dashboard. - Track everything. Use
list_botsto see all active agents in one place, ensuring you're querying the right bot for the job.
Real-World Use Cases
Debugging an Off-Topic Answer
The support team gets a vague answer from 'Support Bot.' Instead of just retrying the query, they first run list_sources to check if the bot is pointing to outdated documentation. They realize a key policy document was removed using delete_source, solving the knowledge gap.
Onboarding New Documentation
The product team just wrote a new pricing guide. Instead of manually uploading it, they use their agent to run add_source on the document's URL. The system crawls and indexes the material automatically, making it available for send_query within minutes.
Reviewing Failed Transactions
A user reports a billing issue. The agent first uses list_conversations to find the ID of the chat that happened last week. Then, running get_conversation pulls up the full context, showing exactly where the bot failed and what actions it was supposed to take.
Testing Bot Functionality
A developer needs to test a new feature without affecting live data. They use list_bots to select their sandbox agent, then manually run send_query with specific inputs to verify the expected response flow.
The Tradeoffs
Assuming Knowledge Availability
The user asks, 'What's our new return policy?' and gets an answer, but later realizes it’s wrong because they never checked if the source documentation was updated.
→
Always check list_sources first. This confirms that the bot is operating with current data before running any critical send_query. If sources are missing, don't query.
Ignoring Conversation Context
The user tries to ask a follow-up question but forgets the specific session ID, and the bot gives a generic or incorrect answer.
→
Run list_conversations first. Get the correct conversation ID, then use get_conversation before running your next send_query. This locks in the context.
Trying to Query an Unknown Bot
The agent is coded to talk to 'BillingBot,' but the bot was renamed last week. The query fails with a generic connection error.
→
Before running any query, always run list_bots to verify the exact name and status of the chatbot you intend to use.
When It Fits, When It Doesn't
Use this server if your core problem is controlling how the AI answers, not just getting an answer. You need programmatic oversight—the ability to manage inputs (sources) and track outputs (conversations). If all you ever do is type a question into a chat box, you don't need this complex toolset; a simple API wrapper will suffice.
However, if your workflow requires steps like: 1. Checking the bot status (check_thinkstack_status), 2. Identifying the correct documentation set (list_sources), 3. Providing that doc list to the agent, and finally, 4. Running the query (send_query), then this server is mandatory. It gives you the necessary visibility into the entire knowledge pipeline.
Common Questions About ThinkStack MCP
How do I list all my chatbots with ThinkStack MCP Server? +
Run the list_bots tool. This returns a manifest of every bot name and basic metadata under your account, letting you know exactly which bot to query next.
What is the best way to update the knowledge base using ThinkStack MCP Server? +
You should use add_source. Just provide the URL or document path, and the tool handles the crawling and indexing process. It’s much cleaner than manual uploads.
Can I view old chat records using ThinkStack MCP Server? +
Yes, you can first use list_conversations to get a list of IDs. Then, pass one of those IDs into the get_conversation tool to pull up the full message history.
Do I need to check status before querying with ThinkStack MCP Server? +
It's best practice. Always run check_thinkstack_status first. This prevents your agent from failing mid-workflow due to a temporary API outage.
How do I authenticate my AI client with ThinkStack MCP Server? +
You must retrieve an API Key from your ThinkStack dashboard. This key provides the necessary credentials for your AI agent to connect and execute tools on your behalf.
What does the `list_actions` tool do in ThinkStack MCP Server? +
The list_actions tool shows all configured REST API actions available for your chatbots. You use this list to see which external functions, like sending emails or updating records, your AI agent can trigger.
How do I audit the knowledge sources using ThinkStack MCP Server? +
Use the list_sources tool to retrieve a comprehensive list of all indexed knowledge bases. This allows you to verify which URLs or documents are currently contributing data to your chatbots.
What affects the response time when I use `send_query` with ThinkStack MCP Server? +
The complexity and volume of retrieved context dictate the speed of the AI response. Keeping queries focused and ensuring sources are indexed helps maintain fast, reliable query performance.
How do I query my chatbot via AI agent? +
Use the send_query tool with the bot ID and your message. The chatbot responds based on its trained knowledge base.
Can I manage knowledge sources programmatically? +
Yes. Use add_source to add new URLs, list_sources to browse, and delete_source to remove outdated sources from any chatbot.
How do I review chat conversations? +
Use list_conversations to see all chats for a bot, then get_conversation to read the full message history of any specific session.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.