datagrout

DataGrout is an enterprise MCP gateway that connects AI agents to Salesforce (700+ tools), QuickBooks Online (550+ tools), Oracle Fusion Cloud (1,000+ tools)…

CommunityOnlineNo sign-inGlobalFreeRead-only

What it can do

  • Discovery.Summary: Compact overview of all available tools and integrations on this server. Returns a compressed text block agents can scan to know what to search for with discovery.discover.
  • Discovery.Discover: Searches available tools for this MCP server by semantic similarity to a goal. Returns a ranked list of matching tools together with their input/output schemas. After calling disco
  • Discovery.Perform: Execute a specific tool by name, or batch-execute multiple tools in parallel. Use this tool as the second step of the discover → perform workflow: 1. Call `discover` (data-grout@1/d

What data it sees

Do you need an account

No: the server works without sign-in

DataGrout is an enterprise MCP gateway that connects AI agents to Salesforce (700+ tools), QuickBooks Online (550+ tools), Oracle Fusion Cloud (1,000+ tools), and any MCP-compatible system through a single endpoint. It reduces LLM token costs through semantic tool discovery, intelligent context engineering, and a neuro-symbolic inference layer that offloads deterministic operations without LLM calls. Every workflow is verified with a Cognitive Trust Certificate before execution. Warden scans every tool output for prompt injection before it reaches the agent. Pre-execution cost estimates and itemized receipts are included with every tool call. Supports Bearer token, OAuth 2.1, and mTLS authentication. Compatible with Claude, Cursor, ChatGPT, n8n, LangGraph, and any MCP-compatible client.

Server tool list (5)

Raw names from tools/list. Only developers need these.

discovery.summaryCompact overview of all available tools and integrations on this server. Returns a compressed text block agents can scan to know what to search for with discovery.discover.
discovery.discoverSearches available tools for this MCP server by semantic similarity to a goal. Returns a ranked list of matching tools together with their input/output schemas. After calling discover, use the `perform` tool (data-grout@1/discovery.perform@1) to actually execute a discovered tool: pass the tool_name from the results as the `tool` argument and supply any required args.
discovery.performExecute a specific tool by name, or batch-execute multiple tools in parallel. Use this tool as the second step of the discover → perform workflow: 1. Call `discover` (data-grout@1/discovery.discover@1) with a `goal` to find relevant tools and their schemas. 2. Call `perform` with the `tool_name` from the discover results and the required `args`. Works for any tool registered on this server, including first-party DataGrout tools, baked integration tools (e.g. salesforce, quickbooks), and multiplexed/external tools.
discovery.planPlans how to satisfy a natural-language goal using the tool catalog. Returns ranked plans, required inputs (holes), and a call_handle.
discovery.guideNavigate tool discovery and execution through a guided decision tree. Presents options at each step, making complex workflows simple to execute while enforcing policy constraints and providing full auditability.