Vectorshift
The End-to-End AI Automations Platform for building and deploying AI workflows, pipelines, chatbots, and knowledge bases.
Verdict
Common use cases
- Audit all chatbots before a product demo
- Spin up a test bot for a new pipeline
- Delete staging chatbots after QA sprint
- Retrieve knowledge base metadata for docs
- List pipelines shared across the team
Integration
- Vendor
- Vectorshift
- Category
- developer-tools
- Auth
- API_KEY
- Tools
- 12
- Composio slug
vectorshift
Tools
- Create Chatbot
Tool to create a new chatbot. Chatbots are conversational AI interfaces built on pipelines. Use when you need to create a new chatbot with a specific pipeline configuration.
- Delete Chatbotdestructive
Tool to delete a chatbot by its ID. Permanently removes the chatbot from the account. Use when you need to remove a chatbot that is no longer needed.
- Get Chatbot
Tool to fetch an existing chatbot by its ID or name. Returns chatbot configuration and metadata. Use when you need to retrieve details about a specific chatbot. Either chatbot ID or name must be provided.
- Get Knowledge Base
Tool to fetch an existing knowledge base by its ID or name. Returns knowledge base configuration and metadata. Use when you need to retrieve details about a specific knowledge base.
- Get Pipeline
Tool to fetch an existing pipeline by its ID or name. Returns pipeline configuration and metadata. Use when you need to retrieve a specific pipeline's details, configuration, or metadata.
- List Chatbots
Tool to list all available chatbots in the account. Use when you need to retrieve chatbot IDs or full chatbot details.
- List Knowledge Bases
Tool to list all available knowledge bases in your VectorShift account. Use when you need to retrieve knowledge base information by id or name.
- List Pipelines
Tool to list all available pipelines in the VectorShift account. Use when you need to retrieve the catalog of pipelines. Supports filtering for shared pipelines and verbose output with full pipeline details.
- List Transformations
Tool to list all available transformations in the account. Use when you need to retrieve transformation IDs or complete transformation objects.
- Run Pipeline
Tool to run a VectorShift pipeline with the given inputs. Use when you need to execute a pipeline and get its results or run_id for asynchronous execution. Returns the pipeline execution status, run_id, and outputs if execution completed sy
- Run Pipeline in Bulk
Tool to run a VectorShift pipeline in bulk with multiple sets of inputs. Use when you need to batch process multiple pipeline executions in a single API call. Returns the overall status and an array of outputs with run_id for each execution
- Terminate Pipeline Execution
Tool to terminate a running pipeline execution. Use when you need to stop a pipeline run by its run_id.
Setup
Setup guide
- 11. Open your Switchy workspace and navigate to Settings → Integrations → Developer Tools. 2. Click 'Connect' next to Vectorshift. 3. Log into your Vectorshift account and go to Account Settings → API Keys. 4. Generate a new API key with read and write permissions for pipelines, chatbots, and knowledge bases. 5. Copy the key and paste it into the Switchy connection dialog, then click 'Authorize'. 6. Switchy confirms the connection and shows which tools are now available. 7. Open any Space and type '@Vectorshift list my pipelines' to test — you should see a list of your existing pipelines. 8. If the list appears, the connection is live and you can invoke any of the 12 tools by @mentioning Vectorshift in a message.
What teammates see: by default, memories from Vectorshift are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
List All Chatbots
@Vectorshift list all chatbots in my account with full detailsOpen in a Space →
Create Support Bot
@Vectorshift create a chatbot named 'Support Bot v2' using the pipeline called 'Customer FAQ Handler'Open in a Space →
Fetch Pipeline Config
@Vectorshift get the pipeline named 'Onboarding Assistant' and show me its configurationOpen in a Space →
Check Knowledge Base
@Vectorshift get the knowledge base called 'Product Docs Q1' and summarize what it containsOpen in a Space →
Clean Up Test Bots
@Vectorshift delete the chatbot with ID 'cb_test_20250115' since we shipped the production versionOpen in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Vectorshift list all pipelines in my account and show me which ones are marked as shared
Found 8 pipelines in your VectorShift account. 3 are marked as shared: 'Customer Support Flow' (ID: pip_abc123, last modified 2 days ago), 'Data Extraction Pipeline' (ID: pip_def456, last modified 1 week ago), and 'Content Generation Workflow' (ID: pip_ghi789, last modified 3 weeks ago). The remaining 5 pipelines are private to your account. Each pipeline includes configuration metadata showing input/output nodes and processing steps.
This example demonstrates read-only catalog access. The MCP requires a valid API key with read permissions on pipelines. Response shape depends on your account's pipeline inventory — accounts with no pipelines will return an empty list. Useful for auditing what automation workflows exist before creating new chatbots.
@Vectorshift create a new chatbot called 'Support Bot v2' using the Customer Support Flow pipeline
Created chatbot 'Support Bot v2' (ID: chat_xyz789) connected to pipeline 'Customer Support Flow' (pip_abc123). The chatbot is now active and ready to handle conversations using the pipeline's configured logic. You can access it via the VectorShift dashboard or embed it using the provided widget code. Note: The chatbot inherits all knowledge bases and model settings from the underlying pipeline configuration.
This write operation creates a persistent resource in your VectorShift account. Requires API key with chatbot creation permissions. The example assumes the referenced pipeline exists — if the pipeline name is invalid, the operation fails. Side-effect: This chatbot will appear in your account's billing and usage metrics immediately.
@Vectorshift review all my knowledge bases and chatbots, then suggest which knowledge base would improve my 'Support Bot v2' responses
Your account has 4 knowledge bases: 'Product Documentation' (127 documents), 'FAQ Archive' (89 entries), 'Technical Specs' (43 PDFs), and 'Customer Tickets' (312 resolved cases). 'Support Bot v2' currently uses the Customer Support Flow pipeline, which doesn't explicitly reference a knowledge base. Based on typical support queries, connecting 'FAQ Archive' would likely improve response accuracy for common questions, while 'Customer Tickets' would help with troubleshooting patterns. Consider testing both to see which reduces escalation rates.
This synthesis example shows the AI reasoning over multiple VectorShift resources. The MCP fetches raw lists; the AI interprets relevance based on naming and metadata. Recommendations depend on the AI's inference, not VectorShift's analytics — validate suggestions against your actual chatbot performance metrics before reconfiguring pipelines.
Use-case deep-dives
When Vectorshift fits a multi-product support desk
A 6-person support team runs three product lines, each with its own chatbot pointing to a different knowledge base. They rotate chatbot configs weekly as feature docs update. Vectorshift's MCP wins here because the List/Get/Create/Delete chatbot tools let an AI agent audit which bots are live, compare their pipeline IDs to the latest docs, and spin up replacements without opening the Vectorshift UI. The 12-tool scope is narrow enough that the agent won't get lost, and API_KEY auth means one shared credential for the whole team. This breaks down if you're managing more than 10 chatbots—at that scale, you need a real IaC tool, not an MCP. If your support desk has three chatbots and they change monthly, Vectorshift's MCP saves you 20 minutes per rotation.
When you need a read-only pipeline inventory
A 3-person AI ops team at a healthcare startup needs to document every pipeline and knowledge base in their Vectorshift account for a quarterly compliance audit. They don't build new pipelines during the audit window—they just need to export metadata and confirm nothing's orphaned. Vectorshift's MCP is the right call because the List Pipelines, List Knowledge Bases, and Get Pipeline tools give an AI agent read access to the full catalog without risking accidental edits. The agent can generate a markdown report of pipeline names, IDs, and knowledge base linkages in one pass. This scenario doesn't work if you need to diff pipeline configs over time—the MCP doesn't expose version history. If your audit is a one-time snapshot and you trust the agent not to call Delete, this MCP closes the loop in under an hour.
When Vectorshift automates new-hire chatbot provisioning
A 10-person engineering team onboards two developers per quarter, and each new hire gets a personal onboarding chatbot linked to a knowledge base of internal docs. The hiring manager wants the chatbot live on day one without manual setup. Vectorshift's MCP fits because the Create Chatbot and Get Knowledge Base tools let an AI agent check if the onboarding knowledge base exists, then provision a new chatbot with the right pipeline in a single workflow. API_KEY auth means the agent can run unattended from a Slack command or calendar trigger. This falls apart if your onboarding docs live outside Vectorshift—the MCP can't ingest files or sync external sources. If your knowledge bases are already in Vectorshift and you hire predictably, this MCP turns a 15-minute manual task into a zero-touch automation.
Frequently asked
What does the Vectorshift MCP let me do in Switchy?
It connects your Vectorshift account so Switchy can manage your chatbots, pipelines, and knowledge bases. You can create new chatbots tied to specific pipelines, list existing resources, fetch configurations, and delete chatbots you no longer need. The MCP exposes 12 tools covering the core Vectorshift workflow—building conversational AI without leaving Switchy's interface.
Do I need a Vectorshift API key to connect this MCP?
Yes. The MCP uses API key authentication, so you'll need to generate one from your Vectorshift account settings. Paste it into Switchy's connection flow. Anyone on your team with a valid Vectorshift API key can connect their own instance—no admin approval required on the Vectorshift side, though your workspace owner controls who adds MCPs in Switchy.
Can the MCP train or update my Vectorshift knowledge bases?
No. The MCP can list and retrieve knowledge base metadata, but it doesn't upload documents or retrain embeddings. If you need to add data to a knowledge base, do that directly in Vectorshift's UI or via their full API. The MCP is read-only for knowledge bases—it's designed for orchestration and inspection, not content management.
Why use this MCP instead of calling Vectorshift's API directly?
The MCP wraps Vectorshift's API in a format Switchy's AI can reason about and chain with other tools. Instead of writing code to list pipelines, filter chatbots, and parse JSON responses, you describe what you want in natural language. Switchy handles authentication, error retry, and combining Vectorshift actions with tools from Slack, GitHub, or your database in a single workflow.
Who on my team should connect the Vectorshift MCP?
Whoever builds or manages your Vectorshift chatbots and pipelines. Typically a product manager, AI engineer, or ops lead. The API key they use determines which Vectorshift resources Switchy can see—so connect the account that owns the pipelines you want to automate. Each team member can connect their own key if you run separate Vectorshift workspaces.