Agentic Index

AI agent platforms that support Model Context Protocol tools (2026)

Of the 548 agentic AI platforms in the Agentic Index, 321 document Model Context Protocol (MCP) support and extensibility in full. Only 137 also document tool calling into real systems and model flexibility in full. That is 25% of the field. This is a bar, not a leaderboard: a platform either documents all three in its own public materials or it does not clear, and Partial evidence on any one of the three does not count.

The finding underneath the list is more useful than the list. The tool layer opened and the model layer did not. Tool calling is documented in full by 72% of the pool and MCP by 59%, but model flexibility and routing by only 38%. Of the 173 platforms sitting exactly one capability short of the bar, 112 are short on model flexibility alone. MCP is sold as the end of integration lock in, and close to half the platforms that speak it still choose the model for you.

There is a second number worth carrying into a vendor call. 147 of the 396 platforms that document tool calling in full do not document MCP support. More than a third of the platforms that can call a tool cannot show they speak the protocol. Tool calling and MCP get used as synonyms in vendor marketing and they are not the same claim.

The bar, and how the 548 platforms score against it

Capability	What has to be documented	Full	Only blocker
MCP and extensibility	Model Context Protocol (MCP) support as a client, a server or both, plus a public API or SDK the buyer can build against	321 (59%)	22
Integrations and tool calling	documented tool calling into real systems, through prebuilt connectors, an MCP server or both	396 (72%)	39
Model flexibility and routing	bring your own model or bring your own key (BYOK), or documented routing between providers rather than one hardwired model	207 (38%)	112

Full means the vendor publishes evidence meeting the capability in its own public materials, under the Agentic Index verification standard. Only blocker counts platforms that document the other two in full and fail on this one alone.

Clears the bar and scores 12.5 or higher of 14 overall

These 36 platforms document MCP and extensibility, tool calling and model flexibility in full, and also sit at the top of the Agentic Index coverage score across all 14 capabilities. Ordered by total coverage, ties broken alphabetically.

1. 1.Appian

   14.0 / 14 capabilities

   enterprise operations agents, agent builders

   Appian exposes itself as well as consuming tools: the Appian MCP Server lets outside agents reach its data fabric, and its AI agents use external tools hosted on MCP servers beside process models configured as tools. Each agent runs on the model set in its model selection, defaulting to the latest model from the configured provider.

2. 2.FLOWX.AI

   14.0 / 14 capabilities

   enterprise operations agents, multi-agent platforms, agent builders, agent infrastructure platforms

   Enterprise agentic platform with an agent builder, a catalog of 246 agent blueprints and documented orchestration patterns. Exposes SDKs for Angular, React, iOS and Android plus REST APIs, its connectors are independently deployable microservices over external systems, and customers configure their own model providers and keys or use the FLOWX managed key.

3. 3.Gumloop

   14.0 / 14 capabilities

   agent builders

   Visual canvas whose MCP story runs both directions: custom MCP servers plug into agents through a dedicated Connectors tab, and webhook triggers with unique URLs let any external form, alert or product event start a run with the data it sends.

4. 4.Mastra

   14.0 / 14 capabilities

   agent infrastructure platforms

   TypeScript SDK installed from npm whose agents and workflows are exposed to outside callers through a Mastra server, a client, custom routes and A2A, while agents take MCP connections and custom tools defined with a typed schema. The model router reaches thousands of models by provider and model string, with the customer's own keys accepted.

5. 5.Microsoft

   14.0 / 14 capabilities

   enterprise operations agents

   Copilot Studio agents can be extended with Model Context Protocol so the tools and resources a server offers become available, beside connectors, flows and APIs, and the Microsoft 365 Agents SDK attaches agents built in C#, JavaScript or Python. The maker picks the primary model per agent, an external model included.

6. 6.ServiceNow

   14.0 / 14 capabilities

   enterprise operations agents, multi-agent platforms

   Action Fabric runs MCP in both directions plus A2A: an outside agent, Claude, ChatGPT and Gemini included, can execute governed ServiceNow actions headlessly, the MCP Server Console controls what is exposed, and ServiceNow agents reach other systems as an MCP client. An administrator picks the model provider in AI Admin Hub.

7. 7.UiPath

   14.0 / 14 capabilities

   enterprise operations agents, multi-agent platforms, agent builders

   Agentic automation orchestrating agents, robots and people end to end through Maestro. Autopilot is documented against GPT, Gemini and Claude running on the customer's own cloud, which is why it clears the axis that stops 112 other platforms.

8. 8.Pega

   13.5 / 14 capabilities

   enterprise operations agents, agent builders

   Since Infinity 26, customers invoke Pega agents and workflows from AI assistants over MCP, and a Connect MCP rule points an agent at an external MCP server so its tools are used during case processing, with A2A for third party agents. Model choice sits with the customer's architects across Azure OpenAI, AWS Bedrock and Google Vertex.

9. 9.StackAI

   13.5 / 14 capabilities

   enterprise operations agents, agent builders

   No code enterprise platform with a documented API Reference covering flow runs, knowledge bases and analytics, a StackAI MCP Server that exposes the platform to MCP clients, and an enterprise CLI, so the interop surface is written down at each layer rather than asserted.

10. 10.Agno

    13.0 / 14 capabilities

    agent infrastructure platforms

    High performance agent runtime, formerly Phidata, Apache 2.0 licensed. The lightest thing on this list and it still clears all three, which is what a runtime rather than a suite ought to look like.

11. 11.AutoGPT

    13.0 / 14 capabilities

    multi-agent platforms, agent builders, agent infrastructure platforms

    An MCP tools block sits beside a Claude Code block, an E2B sandbox, a generic HTTP request block with host scoped authentication, and blocks for media generation and common services. The platform is open source, so anything missing can be written as a new block.

12. 12.Base44

    13.0 / 14 capabilities

    agent builders

    Prebuilt integrations cover AI, email, file storage, payments and messaging, with Airtable, Slack, Stripe and Zapier named, and a community integrations list lets third parties contribute through a create and approve flow.

13. 13.CrewAI

    13.0 / 14 capabilities

    multi-agent platforms

    Open source multi agent orchestration for collaborative crews. Open source frameworks often clear this bar, because portability is the reason people pick them.

14. 14.Databricks Agent Bricks

    13.0 / 14 capabilities

    multi-agent platforms

    Agents are served as REST APIs and created programmatically through the Databricks SDK, the Genie One MCP server exposes Genie to outside MCP clients, and agents take tools from managed, external and custom MCP servers, with model providers connected on the customer's own keys under Unity Catalog permissions.

15. 15.Dataiku

    13.0 / 14 capabilities

    multi-agent platforms, agent builders

    Agents are exposed as A2A servers at a per agent endpoint beside a documented MCP server, they call MCP servers, OpenAPI specs and API endpoints as tools, and the LLM Mesh connects OpenAI, Anthropic, Bedrock, Vertex, Mistral and other providers through administrator defined connections.

16. 16.HappyRobot

    13.0 / 14 capabilities

    enterprise operations agents

    AI workers for operations teams, built in a workflow editor, whose interop is documented at each layer: a REST API across more than 20 resource areas, a TypeScript SDK, a Web SDK for browser voice and an MCP server, beside 200+ prebuilt connectors to systems such as Salesforce, HubSpot, ServiceNow and Zendesk whose actions read records and write outcomes.

17. 17.n8n

    13.0 / 14 capabilities

    agent builders

    Open source workflow automation whose MCP Server Trigger node turns n8n itself into a Model Context Protocol server, making its tools and workflows available to MCP clients over SSE or streamable HTTP, with separate documentation for consuming MCP the other way.

18. 18.Relevance AI

    13.0 / 14 capabilities

    multi-agent platforms, agent builders

    A TypeScript SDK runs agents, tasks and workforces with streaming, an MCP server gives MCP clients access to the platform, and agents call tools from a no code builder and a large integration catalog, choosing their model per agent with a backup model from a different provider.

19. 19.Replit Agent

    13.0 / 14 capabilities

    agent builders

    App builder and runtime with a documented Replit MCP Server connectable by compatible clients, and an agent that connects outward through a directory of pre listed MCP servers or custom ones, so it sits on both sides of the protocol.

20. 20.Sema4.ai

    13.0 / 14 capabilities

    agent builders

    Enterprise agent platform whose Actions framework is open source under its own GitHub organization, with an Action Server that deploys an action with zero configuration, so the extensibility layer is inspectable source rather than a gated portal.

21. 21.Sim

    13.0 / 14 capabilities

    multi-agent platforms, agent builders

    Open source Apache 2.0 agent workspace with a public API, SDKs and a Function block for custom code. More than a thousand native service integrations and fifteen plus model providers documented.

22. 22.Tray.ai

    13.0 / 14 capabilities

    agent infrastructure platforms

    Agent Gateway turns Tray workflows into composite MCP tools and connector operations into individual ones, behind an allowlist, RBAC, rate limits and OAuth2 or token authentication, and Tray publishes OpenAPI specifications, a connector development kit and a headless MCP server for coding assistants. Agents run on Tray's native model or the customer's own OpenAI or Bedrock keys.

23. 23.Warp

    13.0 / 14 capabilities

    multi-agent platforms (secondary lane membership, primary category Coding agent)

    A coding environment whose agents can be reached from outside and reach out in turn: a REST API that starts agent runs with bearer token authentication, a TypeScript SDK and a CLI, integrations with Slack, Linear and GitHub so a mention starts an agent, and a model picker across Claude, GPT, Gemini, GLM and Kimi models with an Auto option and different models on different tasks.

24. 24.Writer

    13.0 / 14 capabilities

    enterprise operations agents, agent infrastructure platforms

    Enterprise platform built on its own Palmyra model family with a dedicated developer estate at dev.writer.com and distribution through Amazon Bedrock. Its connectors write into systems Writer does not own with the verbs documented, and its own FAQ answers the model choice question directly.

25. 25.Akka

    12.5 / 14 capabilities

    multi-agent platforms, agent builders, agent infrastructure platforms

    The Akka SDK exposes services over HTTP, gRPC and MCP endpoints with A2A built into agents, agents call function tools and MCP tools, and nine model providers are built in with any other added as a custom provider, chosen per agent.

26. 26.Boomi

    12.5 / 14 capabilities

    enterprise operations agents, multi-agent platforms, agent builders

    Integration platform with an agent control tower over enterprise systems. MCP support sits next to a vendor agnostic registry that governs third party agents, and the model layer stays open across Bedrock, Snowflake Cortex and OpenAI.

27. 27.Fabrix.ai

    12.5 / 14 capabilities

    enterprise operations agents, agent builders

    Runs MCP both ways, with a REST reference served as Swagger and a command line guide at docs.fabrix.ai, and agents act on roughly thirty named IT products across observability, ITSM, network, security and cloud. Model choice is the first tab of its capability set: cloud providers, or open source models deployed inside the customer's infrastructure.

28. 28.Harvey

    12.5 / 14 capabilities

    enterprise operations agents

    Enterprise legal platform, the one vertical product in this tier, whose developer surface is a documented API at developers.harvey.ai with bearer token authentication and an EU hosted host. It is sold as an additional purchase through the account manager, which is the caveat to read beside the score.

29. 29.IBM watsonx Orchestrate

    12.5 / 14 capabilities

    enterprise operations agents

    The Agent Development Kit carries a Python SDK, a command line and a published OpenAPI specification for the server API, and tools reach external services through their own connections, an OpenAPI plugin and MCP based workflows. Models from several providers are imported and assigned per agent, with policies for load balancing, fallback and retries.

30. 30.Kestra

    12.5 / 14 capabilities

    multi-agent platforms, agent infrastructure platforms

    API first, with every UI action in its REST API and official SDKs for Java, Python, JavaScript and TypeScript, and Go, and its MCP servers expose flows as named tools to outside agents. Its AI Agent task takes any Kestra task or flow, web search, MCP clients and other agents as tools, on the model the customer sets per task.

31. 31.Latenode

    12.5 / 14 capabilities

    agent builders, agent infrastructure platforms

    Low code platform whose extensibility rests on full JavaScript with access to more than a million NPM packages inside workflows, HTTP and webhook support, and a white label embedded iPaaS for building on top of it, which is a code first shape of interop beside the MCP native entries.

32. 32.OutSystems

    12.5 / 14 capabilities

    multi-agent platforms, agent builders

    Enterprise low code incumbent that shipped Model Context Protocol support in Agent Workbench at general availability in September 2025 and followed with Agent Experience, a platform layer exposing A2A and MCP surfaces, in June 2026. The incumbent that documented the protocols rather than announcing them.

33. 33.Pipefy

    12.5 / 14 capabilities

    enterprise operations agents, agent builders

    Agents reach outside systems through MCP servers a team connects and through native connectors and custom API calls, and the platform is callable through a documented GraphQL API that creates, updates and deletes agents with their full behavior configuration. An administrator sets the LLM provider for the organization or an individual assistant, custom providers included.

34. 34.Salesforce

    12.5 / 14 capabilities

    enterprise operations agents

    CRM platform whose Agentforce layer runs autonomous agents across sales, service and marketing. Connects through MCP connectors and the Atlas Reasoning Engine is documented across OpenAI, Anthropic and Google Gemini rather than one house model.

35. 35.SimplAI

    12.5 / 14 capabilities

    multi-agent platforms, agent builders, agent infrastructure platforms

    Datafuse Technology Inc builds SimplAI, an enterprise agentic operating system running multi agent systems across cloud, on premises and air gapped environments. A model and intelligence layer routes across multiple LLMs, and the step library spans twenty five plus categories.

36. 36.Torq

    12.5 / 14 capabilities

    enterprise operations agents (secondary lane membership, primary category Security / SOC agent)

    Security hyperautomation platform whose agents act through a catalog of integrations and AI Tools across the security and IT stack. Customers bring their own OpenAI, Azure OpenAI, Vertex AI, Anthropic or Bedrock subscription and pick the model per agent and per task, and the Torq API and webhooks let outside systems start workflows. Qualifies here on secondary lane membership as an enterprise operations platform, which a buyer scanning this list should know.

The remaining 101 platforms that clear the bar

Every one of these documents MCP and extensibility, tool calling and model flexibility in full. They score below 12.5 of 14 on total coverage, which says something about breadth across the whole taxonomy, not about how open they are. Several of the most portable things in the index sit low here precisely because they do one job.

Common questions

Which AI agent platforms support Model Context Protocol tools?

321 of 548 agentic AI platforms in the Agentic Index document Model Context Protocol support and extensibility in full. 137 of them also document tool calling into real systems and model flexibility in full, which is the bar used on this page. That is 25.0% of the field. The list is ordered by total documented coverage across the Agentic Index 14 point capability taxonomy and is graded from public evidence only.

What counts as real Model Context Protocol support in an AI agent platform?

Three things, and a platform has to document all three in public materials to clear the bar used here. Model Context Protocol (MCP) support as a client, a server or both, alongside a public API or SDK. Documented tool calling into real systems through prebuilt connectors, an MCP server or both. And model flexibility: bring your own model or bring your own key (BYOK), or documented routing between providers rather than one hardwired model. Partial evidence on any one of the three does not clear.

Do most AI agent platforms that call tools support MCP?

No. 396 of the 548 platforms document tool calling in full, and 147 of those do not document MCP support. More than a third of the platforms that can call a tool cannot show they speak the protocol. MCP adoption is real but it trails tool calling by a wide margin.

Why do so few AI agent platforms clear the interoperability bar?

Model flexibility, not the protocol. Across the 548 platform pool, tool calling is documented in full by 72% and MCP by 59%, but model flexibility and routing by only 38%. Of the 173 platforms sitting exactly one capability short of the bar, 112 are short on model flexibility alone. MCP is sold as the end of integration lock in, and close to half the platforms that speak it still choose the model for you.

Is this ranking paid or sponsored?

No. No vendor pays for placement, no vendor has reviewed this page, and every grade comes from the vendor's own public materials under the Agentic Index verification standard. 956 vendors are graded against the same 14 capabilities. Data last verified October 3, 2026.

Method: membership is the same 548 platform pool used by the best agentic AI platforms in 2026, drawn from 956 researched vendors. The editorial pages over this pool are one method with different bars, not several opinions. Every grade comes from the vendor's own public materials under the Agentic Index verification standard. No vendor pays for placement and no vendor has reviewed this page. Data last verified October 3, 2026. How this evidence is graded

Related: agent observability platforms, enterprise security and compliance platforms, autonomous AI workforce platforms, platforms that let you bring your own model or route between providers, production ops agents ranked for automated incident resolution, how every vendor scores on MCP and extensibility, compare platforms side by side.