# Chatway MCP: Tools, Pricing and Setup (2026 Guide) By Ossian Kettunen · 2026-10-15 · Source: https://www.activepieces.com/blog/chatway-mcp-tools-pricing-and-setup-2026-guide --- Chatway, according to its own site, has released Chatway MCP. MCP, the Model Context Protocol, is a standard that lets an AI client call outside tools through a defined connection, much as workflow tools like [Activepieces](https://www.activepieces.com) are wired into AI clients, and chatway.app presents Chatway MCP as that kind of connection to Chatway. It's new. ## Chatway MCP launched in October 2026: what happened ### What Chatway released Do not expect a finished automation: the release is a connection point. Chatway's pages describe what the MCP server exposes, and that description is the only evidence for what an automating team can do with it. I'm not assuming which triggers, actions or field mappings exist beyond what those pages name. Whether a new server lists a handful of tools or a broad set decides whether a real workflow runs or stalls at the first missing field. I've found **no native integration for Chatway**. Until one exists, the practical route is a general building block such as an HTTP request or an MCP step, and each use needs testing against the maker's description. ### Sources and verification dates for Chatway MCP claims **Nobody outside Chatway has confirmed any of it**, because every claim comes from Chatway's own site. Treat each statement as the maker's position. Verify anything you'll build on, and watch for changes, since a young release can shift quickly. The next section covers what [Chatway](https://chatway.app/pricing)'s pages actually state. ## Why Chatway MCP matters to teams automating work If you use MCP, an AI agent or coding tool reaches Chatway directly, so a team skips writing and maintaining a bespoke bridge for each tool it wants to use. MCP, the Model Context Protocol, is the shared interface that makes this possible. ![Taped cardboard adapter beside a standard plug fit](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/5dea86ba-e65d-4b08-a8d0-825094ddc692/chatway-mcp-tools-pricing-and-setup-2026-guide-i-c3f63c6e.webp) Without it, every agent needs its own handling of Chatway's API and webhooks, and the team owns that code. On [its pricing page](https://chatway.app/pricing), Chatway says you can connect your own AI agent, naming Claude Code, OpenAI Codex, GitHub Copilot, Gemini, Devin and Cursor, using MCP, webhooks and APIs. You choose what the agent can access, the [homepage](https://chatway.app) adds, so it can tag conversations, hand them off and help the team. It sits alongside Chatway's built-in AI Support Agent rather than replacing it. ### Chatway MCP Devin integration page details Start with the [Devin integration page](https://chatway.app/ai-agent/devin), the clearest example. Devin is an AI software agent, and the page says it can manage conversations, reply to customers, search support data and control Chatway from inside Devin. For an automating team, that means a support workspace becomes something an agent operates, not just something a person reads and an export feeds. How much damage a bad reply can do depends on the access you grant, and nothing on the page shows how well an agent handles real replies. It describes capability. Treat the value as plausible and unproven until you test it on your own workspace. ## What is measurably new, from Chatway's own pages There is **no speed or benchmark figure anywhere**, only a list of supported clients including Devin and VS Code, and a free plan. The table below separates what Chatway states from what's missing, so you can see how much depends on the maker's word. | Item | Stated fact | Source | Figure | |---|---|---|---| | Devin, VS Code, Claude Code, Claude Desktop, ChatGPT, Codex, Cursor, Muse, Gemini, Grok, Cline, OpenCode, Goose, Manus, any MCP client | [Chatway help](https://chatway.app/help) | Not applicable | | Tool count | — | [Chatway Developers](https://developers.chatway.app/mcp/guides.mcp_tool_reference) | Not applicable | | Free plan | $0, 50 conversations including MCP | Chatway pricing page | Not applicable | | Speed and benchmarks | None | None | No figure given | Everything else below is a claim to verify, not a measurement. ### What Chatway MCP is and who built it Chatway makes the server itself. Its [help page](https://chatway.app/help/chatway-for-ai-agents/how-to-connect-chatway-with-vs-code) says VS Code can reach Chatway conversations, customer information and support data, based on the permissions you grant. Per Chatway [Developers](https://developers.chatway.app/mcp/guides.mcp_tool_reference)' [tool reference](https://developers.chatway.app/mcp/guides.mcp_tool_reference), inbox actions take 9 of the 21 tools and conversations 7. For VS Code, the help page documents a connection through the MCP integration. Chatway names Devin as supported, but this section's evidence gives no setup steps for it. $0 buys 50 conversations, MCP included. MCP traffic counts against that allowance, so a team testing automations has a small budget before paying. ## Connecting Chatway MCP in a workflow today As of 2026-10-15, the builder has no native Chatway MCP integration. You reach it through three generic building blocks. Given the small trial allowance, the cheapest one to test should go first. The table compares them. Take from it that the effort rises with each route, and so does what you must confirm in Chatway's docs. | Route | Setup effort | Needs from Chatway docs | When it fits | Not run by us | |---|---|---|---|---| | HTTP request step | Low | Endpoint URL, auth header, request body shape | One fixed call | Yes | | AI step, custom provider | Medium | Whether Chatway exposes a compatible model endpoint | Only if the docs describe one | Yes | | MCP server connection | Medium to high | Server URL, transport type, auth method | An agent choosing among Chatway tools | Yes | We haven't run any of these against Chatway. Each row is a plan drawn from the maker's pages, not a result. Call one documented endpoint with a fixed payload. You need the exact URL and header format from Chatway, and nothing else. Use it when the job is a single predictable action. Only if Chatway publishes an endpoint speaking an OpenAI-compatible format does route 2, an AI step with a custom provider, work. If the docs describe an MCP server and nothing more, this route has nothing to connect to. A transport is simply how an MCP client and server talk to each other: either as a local process on your machine (stdio) or over a remote HTTP-based connection (streamable HTTP, or the older SSE). ![MCP client and server joined by a local cable and a remote cable](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/16b7cb34-b30f-4af7-a381-737bcd32b7ef/chatway-mcp-tools-pricing-and-setup-2026-guide-i-c946018c.webp) The three columns in the matrix below run in that order: stdio, streamable HTTP, SSE. To find which one Chatway documents, look in its tool reference and setup pages for how the server is reached, since a command to launch points to stdio and a server URL points to a remote HTTP transport. That matters for a client like ChatGPT, which lacks stdio, so a Chatway server offered only as a local process would not connect there. With route 3, an MCP server connection, the question is transport. ConnectorZone's [client matrix](https://connector.zone/tools/client-matrix/) scores each client on three transports: Claude Desktop, Claude Code, Cursor and VS Code score 1, 1, 1. ![Which MCP transports popular AI clients support](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/12b2be45-a76d-48aa-a9e5-c266bf86325d/chatway-mcp-tools-pricing-and-setup-2026-guide-m-d38a6580.svg "Source: ConnectorZone") ChatGPT scores 0, 1, 1. So ChatGPT lacks the first transport, and a server offering only that one won't connect there. Check which transport Chatway documents before choosing a client. [Activepieces](https://www.activepieces.com) exposes each project as its own MCP server, so Claude, Cursor or Windsurf can create the HTTP request flow, change it and explain it back to you with no builder UI open. That is real create-and-edit control over an open protocol, not a chat box describing a screen it can't touch, and the tool calls and resulting changes land in the run trace for you to read after asking for a Chatway call. MoneyGram, Moneypenny, Alan and FundingSocieties run Activepieces, and its core is MIT-licensed. ## Chatway MCP pricing compared with alternatives Since no speed figure exists for Chatway MCP, from the maker or from us, the only comparison we can make is on price. [Chatway's launch post](https://chatway.app/blog/chatway-our-website-live-chat-product-is-live) lists Free at $0 per month, so a team can start without budget approval. viz type=step title="What Chatway charges across its plans" unit="$ per month" source="Chatway - https://chatway.app/blog/chatway-our-website-live-chat-product-is-live" Free: 0 Business: 39 Plus: 199 $39 per month buys Business, the first paid step. MCP access is available on all Chatway plans, including Free, so you don't need a higher tier to use it. We found no separate MCP charge in the pages we used. Beside those plans, the table sets two Zapier tiers and a plain HTTP step, the general building blocks a team could use today, so teams can pick between a managed option and building it themselves. It shows the prices and the monthly conversation limits. | Route | Monthly price | Included limit | Measured speed | |---|---|---|---| | Chatway Free | $0 | 50 conversations/month, including those handled via API, MCP and Webhooks | None | | Chatway Business | $39 | 200 conversations/month, including those handled via API and MCP | None | | Zapier Free | $0 | Not stated in the pages reviewed | None | | Zapier Professional | $19.99 | Not stated in the pages reviewed | None | | HTTP-step API calls | Depends on the platform running the step | Depends on that platform's plan | None | Because the speed column is empty, nobody can promise latency yet. Business costs about twice Zapier Professional, so Chatway has to justify that gap on features, not on any proven performance advantage. **Until the maker publishes limits and timings, run a small test on a free tier and time it yourself.** ## What to watch after Chatway's launch Once your own timing test has settled what you can measure today, the rest changes only when Chatway or the tooling around it changes, and four public places will show it. Although Chatway released an MCP offering, everything about what it does for an automating team rests on the maker's own pages. No independent write-up or third-party test backs those pages yet. Treat any claim beyond them, such as field coverage, trigger behavior or rate handling, as unverified until you have run it. Read on 15 October 2026: every statement here comes from Chatway's published pages. I didn't run the server, so a later edit to those pages can outdate this within days. To see whether the picture's moving toward a native integration or away from it, run the four checks below. Most are a minute of reading. - Chatway's changelog, for newly supported agents - Chatway's pricing page, for any change to the 50-conversation Free limit, since that limit caps how much testing you can do without paying - The automation platform's public repository, for commits that mention Chatway - The platform's integration catalog, for a Chatway integration Once the last two show something, the general building blocks stop being a workaround and you can plan around a native connector. Until then, put the four checks on a Monday calendar slot and keep your claims no wider than what you can see. ![Plank bridge beside an empty bridge foundation](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/adbb6577-16af-4c23-94e5-e62c0e6d508e/chatway-mcp-tools-pricing-and-setup-2026-guide-i-c1290f78.webp) ## What Activepieces does about this **Activepieces has no Chatway integration**. Chatway can be connected through the general blocks: the HTTP request step, or a connection to Chatway's MCP server. I haven't run either against Chatway, and the route you pick still depends on what Chatway's tool reference says about URL, transport and authentication. What the platform does give you is a way to try that route without writing the bridge by hand. Each project is exposed as its own MCP server, so Claude, Cursor or Windsurf can build the HTTP request flow, edit it and explain it back to you. The tool calls and the resulting changes land in the run trace, which is where you check what a test against Chatway did. That suits the budget problem in this article: the Free plan allows 50 conversations, MCP included, so you want each test to be deliberate and readable afterwards. The core is MIT-licensed, and MoneyGram, Moneypenny, Alan and FundingSocieties run Activepieces. That says the platform is used in production by named companies. It says nothing about Chatway MCP, which nobody outside Chatway has verified yet. If Chatway publishes an integration path that fits, the four checks above will show it, including commits that mention Chatway in the public repository. Until then, start with one fixed call in an HTTP request step, time it yourself, and keep the claim as narrow as the run trace supports. ## Frequently asked questions Is Chatway MCP free to use? Chatway says MCP support is included in the Free plan and on all plans, with no separate MCP charge. The maker's pricing page is the place to look, specifically for any line that names MCP alongside a conversation limit. Until you've read that line yourself, treat any claim of "free" or "unlimited" as unverified, and note the date you checked. Does Activepieces support Chatway MCP? Chatway offers its own MCP integration, available on all Chatway plans. What the evidence supports is narrower: a platform with a generic HTTP request step or an MCP step can reach it as a general building block. That route needs setup on your side, and nothing here shows a ready-made Chatway option. ### Clients and setup Which AI agents and editors can connect to Chatway MCP? Chatway lists Claude Code, Claude Desktop, ChatGPT, Codex, Cursor, Muse, Gemini, Grok, VS Code, Cline, Devin, OpenCode, Goose and Manus, plus any MCP client. Any compatibility claim you read elsewhere should be checked against the maker's own tool reference and setup pages. Until then, say that a client which speaks MCP is the precondition, and name the clients Chatway itself lists as supported. Do I need to write code to connect Chatway MCP to my apps? Not documented either way. A visual builder's HTTP request or MCP step involves configuring a URL, authentication and field mappings rather than writing a program. Confirm the connection details on the maker's pages, since it hasn't published a no-code walkthrough for Chatway MCP. Then run one test conversation end to end before you depend on it. ![Activepieces workflow builder showing a three-step automation connecting Google Calendar to Gmail with run details and…](https://ap-marketing-media.fra1.cdn.digitaloceanspaces.com/uploads/004448b6-4c5a-44ec-a116-431c38d4090f/what-is-tool-calling-how-ai-agents-use-tools-in-22f21aa3.webp) ## Related reading - [Top 6 AI Agents for Finance Teams in 2026](https://www.activepieces.com/blog/ai-agents-for-finance) - [AI Agents for Legal Teams: What’s Possible Now](https://www.activepieces.com/blog/ai-agents-for-legal) - [AI Agents for Sales: What Teams Are Using Today](https://www.activepieces.com/blog/ai-agents-for-sales) ## References - [Chatway Developers](https://developers.chatway.app/mcp/guides.mcp_tool_reference) - [Chatway](https://chatway.app/blog/chatway-our-website-live-chat-product-is-live) - [ConnectorZone](https://connector.zone/tools/client-matrix/)