The Veles MCP Server is currently in beta. Tools, schemas, and behavior may change as we iterate. We’d love your feedback — reach out at founders@getveles.com.
What is MCP?
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Think of it like a USB-C port for AI — a single, standardized way to plug models into the systems they need to do real work. Veles exposes an MCP server that gives AI assistants direct access to your proposals, quotes, products, pricing rules, and approval workflows. This means you can use natural language to build quotes, check deal status, look up pricing, and manage approvals — all without leaving your AI assistant.Why it matters
Instead of switching between tabs, copying data, and manually entering line items, your AI assistant can talk directly to Veles. A sales rep using Claude can say “Create a new proposal for Acme Corp with our Growth plan at 100 seats” and the assistant will handle it end-to-end through the MCP server.Getting started
Connect in Claude
The fastest way to get started is through Claude’s built-in connector:Open Claude settings
In Claude, navigate to Settings → Connectors (or click the connectors icon in the chat input).
Authorize
Sign in with your Veles account to grant Claude access. You’ll see a permissions screen showing which tools are available.
Connect in other MCP clients
For AI tools that support remote MCP servers (Cursor, Windsurf, Claude Code, OpenAI Agents, etc.), point your client at:Tool permissions
Once connected, you’ll see two categories of tools:- Read-only tools — Always allowed by default. These fetch data from Veles without making any changes.
- Write tools — Require explicit approval (or custom permission settings) before the AI assistant can execute them.
Tools
The Veles MCP Server exposes 22 tools organized across four areas: proposals, quotes, products and pricing, and approvals.Proposals
Proposals (also called opportunities or deals) are the top-level container for a deal in Veles.| Tool | Type | Description |
|---|---|---|
list_proposals | Read | List proposals in your Veles workspace. Filter by status, owner, or search term. |
get_proposal | Read | Get a single proposal with all of its quotes and metadata. |
create_proposal | Write | Create a new proposal (opportunity or deal). |
update_proposal | Write | Update a proposal’s metadata — name, status, quote type, contract, or primary option. |
list_proposal_versions | Read | List version snapshots for a proposal. Useful for tracking changes over time. |
get_proposal_version | Read | Get a specific version snapshot of a proposal, including all option pricing details. |
list_salesforce_opportunities | Read | List opportunities from Salesforce CRM. Helpful when creating proposals from existing CRM data. |
Quotes
Quotes (also called options or scenarios) live inside a proposal and represent a specific pricing configuration.| Tool | Type | Description |
|---|---|---|
list_quotes | Read | List quotes across proposals. |
get_quote | Read | Get a quote with its phases, line items, pricing, and configuration details. |
create_quote | Write | Add a new quote (option) to an existing proposal. |
update_quote | Write | Update a quote’s configuration, pricing, and line items. |
clone_quote | Write | Duplicate an existing quote as a starting point for a new option. Great for creating A/B pricing scenarios. |
Products & pricing
Products (SKUs) and pricing rules define what you sell and the guardrails around how you sell it.| Tool | Type | Description |
|---|---|---|
list_products | Read | List products visible within a proposal’s context. |
get_product | Read | Get a product (SKU) with its plans, charges, and pricing tiers. |
list_pricing_rules | Read | Get active pricing rules, discount guardrails, and validation logic. |
Approvals
The approval workflow lets you route proposals through your organization’s review process.| Tool | Type | Description |
|---|---|---|
list_pending_approvals | Read | List proposals awaiting your approval. |
get_approval_status | Read | Check where a proposal sits in the approval chain. |
submit_for_approval | Write | Submit a proposal’s quotes into the approval workflow. |
review_approval | Write | Approve, reject, or pass items to the next step in the approval chain. |
comment_on_approval | Write | Leave a comment on a proposal’s approval discussion. |
reassign_approval | Write | Reassign the current approver (ball-in-court) on a pending approval package. |
retract_approval | Write | Retract a pending approval package, cancelling the approval request. |
Example prompts
Once connected, try these prompts with your AI assistant to see the MCP server in action:- “List all my open proposals.”
- “Show me the details on the Acme Corp proposal.”
- “What products are available for this deal?”
- “Create a new quote on the Acme proposal with our Enterprise plan.”
- “Clone Quote A and increase the seat count to 200.”
- “What pricing rules apply to this deal?”
- “Submit the Acme proposal for approval.”
- “What approvals are waiting on me?”
FAQ
Which AI assistants are supported?
Which AI assistants are supported?
Any MCP-compatible client works with the Veles MCP Server. This includes Claude (via the built-in connector), Cursor, Windsurf, Claude Code, and any tool that supports remote MCP servers over Streamable HTTP.
Is my data secure?
Is my data secure?
The MCP server uses OAuth authentication and respects the same role-based access controls as the Veles web app. Your AI assistant can only access data that your Veles user account has permission to see. No data is stored by the AI assistant between sessions.
Can the AI assistant make changes to my deals?
Can the AI assistant make changes to my deals?
Yes, but only if you allow it. Write tools (creating proposals, updating quotes, submitting approvals) require explicit permission in your MCP client’s settings. By default, most clients will prompt you for confirmation before executing write actions.
What happens if a tool call fails?
What happens if a tool call fails?
The MCP server returns structured error messages that your AI assistant can interpret and explain. Common issues include missing required fields, permission errors, or invalid references. The assistant will typically ask you to clarify or correct the input.
Is this a replacement for the Veles UI?
Is this a replacement for the Veles UI?
No. The MCP server is a complementary interface. It’s great for quick lookups, bulk operations, and workflows where you’re already working in an AI assistant. Complex visual tasks like drag-and-drop quote building are still best done in the Veles app.