Comparison Claims ledger.
Every claim about another vendor on the three comparison pages: 335 claims, each with the vendor’s own words and the page they are on. Every page was checked on 2026-11-15. A “not published” row means the page that was read does not say. Every grid cell on the comparison pages is in this ledger.
The vendor’s own words.
Next to every claim.
The quotes are the exact words saved from each page. A script copies them from the research file, so none is retyped. Markdown marks are removed from quotes that come from a Markdown page.
StackJack and vendor-native MCP servers
Sources, each checked 2026-11-15
- 1Syncro. https://www.syncrosecure.com/platform/mcp-server/
- 2Syncro. https://syncrosecure.com/blog/building-syncro-mcp-server/
- 3SuperOps. https://support.superops.com/en/articles/16782631-superops-mcp-for-msps
- 4HaloPSA. https://www.usehalo.com/guides/2597
- 5HaloPSA. https://www.usehalo.com/product-updates/whats-new-in-halopsa-q3-2025
- 6HaloPSA. https://www.usehalo.com/guides/2875
- 7Thread. https://docs.getthread.com/super-magic/thread-mcp-server-developer-guide.md
- 8Thread. https://docs.getthread.com/super-magic/control-who-can-connect-ai-apps-to-thread-mcp.md
- 9Thread. https://docs.getthread.com/super-magic/issue-mcp-automation-keys-for-team-members.md
- 10NinjaOne. https://www.ninjaone.com/press/katana-ai-and-mcp/
- 11NinjaOne. https://www.ninjaone.com/katana-ai/
- 12N-able. https://developer.n-able.com/gql/docs/mcp
- 13N-able. https://www.n-able.com/press/press-releases/n-able-makes-uem-ai-native-with-mcp-server-launch
- 14N-able. https://documentation.n-able.com/N-central/Release_Notes/GA/Content/N-central_2026.2_Release_Notes.htm
- 15Huntress. https://www.huntress.com/blog/huntress-mcp-server
- 16Huntress. https://support.huntress.io/hc/en-us/articles/50846554117395-Connecting-to-the-Huntress-MCP-Server
- 17Huntress. https://feedback.huntress.com/changelog/huntress-mcp-server-is-now-available
- 18MSP360. https://www.msp360.com/resources/blog/mcp-server-for-backup-management/
- 19MSP360. https://help.mspbackups.com/ai-integration-mcp/connect-claude
- 20MSP360. https://www.msp360.com/msp-software/mcp-server/
- 21MSP360. https://help.mspbackups.com/ai-integration-mcp/manage-mcp
- 22Hudu. https://www.hudu.com/blog/release-update-2.42.0
- 23Hudu. https://www.hudu.com/product/features/mcp
- 24Liongard. https://docs.liongard.com/docs/model-context-protocol-mcp-server
- 25Liongard. https://www.liongard.com/media-press/liongard-platform-expansion-mcp-asset-intelligence/
- 26Liongard. https://docs.liongard.com/docs/liongard-mcp-faqs
- 27Liongard. https://innovation.liongard.com/
- 28Liongard. https://docs.liongard.com/docs/liongards-app-activity-logging
- 29Pax8. https://devx.pax8.com/docs/mcp-server
Grid cells
| Vendor | Row | The cell says | Mark | The vendor’s words | Source |
|---|---|---|---|---|---|
| Syncro | Connects your AI to | Syncro | value | connects your live Syncro data to the AI assistants you're already using | 1 |
| Syncro | Status on its own page | Generally available since 2026-08-11 | value | The Syncro MCP Server is generally available August 11, 2026, included in every Syncro plan. | 1 |
| Syncro | Price of the MCP server | Included in every Syncro plan, with no added fee | check | it's free to connect in every Syncro plan. No added fees or costs. | 1 |
| Syncro | AI clients it names | Claude, ChatGPT, Microsoft Copilot, Cursor | value | including Claude, ChatGPT, Microsoft Copilot, and Cursor | 1 |
| Syncro | Sign-in and access | OAuth 2.1. Credentials are scoped, time-limited and can be revoked | value | Access runs on OAuth 2.1 with scoped, time-limited credentials you can revoke at any time. | 1 |
| Syncro | Write actions | Under the approval rules you set in your AI tool | check | Anything that creates or changes a record follows the approval conventions you set in your AI tool | 1 |
| Syncro | Audit log | Not published as a log. Its blog says each action is attributed to a person | not published | activity logging so every action can be attributed to the person who took it | 2, 1 |
| SuperOps | Connects your AI to | SuperOps | value | SuperOps MCP connects your AI assistant directly to your SuperOps MSP workspace. | 3 |
| SuperOps | Status on its own page | No stage or date on the article | not published | Updated over 2 weeks ago | 3 |
| SuperOps | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 3 |
| SuperOps | AI clients it names | Claude, ChatGPT, a custom LLM | value | An MCP-compatible AI client (Claude, ChatGPT, or a custom LLM) | 3 |
| SuperOps | Sign-in and access | A client ID and a secret key. Each module is read only, read and write, or off | value | Secret Key: the credential your AI client uses alongside the Client ID to authenticate with SuperOps over MCP.; Read only: the AI can retrieve data | 3 |
| SuperOps | Write actions | For Tickets and Assets. Every other module is read-only | check | Read & Write is available for Tickets and Assets only. All other modules are read-only. | 3 |
| SuperOps | Audit log | Not published | not published | No audit wording on the page read. | 3 |
| HaloPSA | Connects your AI to | HaloPSA | value | allows AI assistants such as OpenAI to connect and dynamically invoke tools in the Halo API | 4 |
| HaloPSA | Status on its own page | No stage on its MCP guide. Its Q3 2025 notes and its chat guides say beta | value | Although still in beta, MCP is intended to become the primary framework for agentic AI in Halo; + HaloMCP (Beta) | 5, 6, 4 |
| HaloPSA | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 4 |
| HaloPSA | AI clients it names | OpenAI, in the guide we read | value | allows AI assistants such as OpenAI to connect | 4 |
| HaloPSA | Sign-in and access | A bearer token from a Halo API application | value | ensure the application used to generate the bearer token is granted only the permissions required by the available tools | 4 |
| HaloPSA | Write actions | Add a note to a ticket, log time, run a runbook | check | Add Note to Ticket - Allows you to add a private note to a ticket on behalf of the agent, allowing agents to update tickets.; Log Time - Logs a quick time entry for the agent.; Execute a Runbook - Executes a runbook of choice. | 4 |
| HaloPSA | Audit log | A setting turns on MCP request logging | check | it is advisable to enable the logs for this by navigating to Configuration > AI > Halo MCP Server; Enable MCP request logging | 6 |
| Thread | Connects your AI to | Thread’s service-desk tools | value | Connect Claude Desktop, ChatGPT, Cursor, or any MCP client to Thread's service-desk API via OAuth 2.1 to search tickets, manage contacts, and run write actions. | 7 |
| Thread | Status on its own page | Generally available to all partners | value | Availability: generally available to all partners. | 7 |
| Thread | Price of the MCP server | Open to all partners. Write tools need an Agentic plan | value | Availability: generally available to all partners.; Write and integration tools appear only on Agentic plans. | 7 |
| Thread | AI clients it names | Claude Desktop, ChatGPT, Cursor, or any MCP client | value | Connect Claude Desktop, ChatGPT, Cursor, or any MCP client to Thread's service-desk API via OAuth 2.1 to search tickets, manage contacts, and run write actions. | 7 |
| Thread | Sign-in and access | OAuth 2.1, one member for each connection. Admins choose who may connect | value | Workspace admins can decide who is allowed to connect at all, see every app that's connected, and cut off any app with one click.; Connect Claude Desktop, ChatGPT, Cursor, or any MCP client to Thread's service-desk API via OAuth 2.1 to search tickets, manage contacts, and run write actions. | 8, 9, 7 |
| Thread | Write actions | On Agentic plans. Other plans see read-only tools | check | Write and integration tools appear only on Agentic plans.; On a non-Agentic plan (no AI Pro), the connection sees only the seven read-only Super Magic tools | 7 |
| Thread | Audit log | Every tool call is logged under the member | check | every tool call is logged under the member identity behind the connection. | 7 |
| NinjaOne | Connects your AI to | NinjaOne | value | the NinjaOne MCP server, an agentic orchestration connector that extends the platform's governed access to the AI tools teams already use | 10 |
| NinjaOne | Status on its own page | Announced 2026-10-13 | value | Oct 13, 2026 | 10 |
| NinjaOne | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 10 |
| NinjaOne | AI clients it names | None named | not published | external AI agents; AI tools | 10 |
| NinjaOne | Sign-in and access | The same role-based permissions as the platform | value | governed by the same role-based permissions IT teams already manage | 10 |
| NinjaOne | Write actions | It can take action in the NinjaOne platform | check | It lets external AI agents interact with NinjaOne data and take action directly in the NinjaOne platform. | 10 |
| NinjaOne | Audit log | Not published for MCP | not published | The audit wording on its page is about Katana AI, not about the MCP server. | 11 |
| N-able | Connects your AI to | N-central and N-sight RMM | value | it provides AI assistants with real-time access to N-central, N-sight RMM, and other supported data sources | 12 |
| N-able | Status on its own page | Announced 2026-04-13. Its docs say active development. Its release notes say public preview | value | N-able MCP is in active development; April 13th, 2026; MCP Servers are available in public preview. | 12, 13, 14 |
| N-able | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 12, 13 |
| N-able | AI clients it names | Claude Desktop, Claude Code, Gemini CLI, Copilot in VS Code, Copilot Studio | value | Gemini CLI Google Workspace/CLI flow using httpUrl and token header.; Copilot Studio Step-by-step setup for adding N-able MCP inside a Copilot Studio agent | 12 |
| N-able | Sign-in and access | An API token. It inherits the permissions of the N-able SSO user who made it | value | API tokens inherit permissions from the N-able SSO user who created them | 12 |
| N-able | Write actions | Can run scripts and modify devices. A read-only endpoint is the suggested start | check | Can query data, execute scripts, and modify devices.; Recommended starting point for most use cases. | 12 |
| N-able | Audit log | Not published for MCP. Its release notes say an AI audit log comes later | not published | AI Audit Log, AI Conversation Log and AI Usage Dashboard will be implemented at a later stage and details will be provided. | 14 |
| Huntress | Connects your AI to | Huntress | value | The Huntress MCP Server provides read-only access to your account data, including agents, organizations, incident reports, signals, escalations, remediations, invoices, reports, and external recon. | 15 |
| Huntress | Status on its own page | Its own pages give three dates: 2026-04-22, 2026-04-27 and 2026-08-07 | value | April 22, 2026; August 7, 2026 | 16, 17, 15 |
| Huntress | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 15, 16 |
| Huntress | AI clients it names | Claude Desktop, Claude Code, ChatGPT. Its support article names Claude and OpenAI Codex | value | an MCP-compatible AI client like Claude Desktop, Claude Code, or ChatGPT; connects AI assistants like Claude and OpenAI Codex directly to your Huntress account data. | 15, 16 |
| Huntress | Sign-in and access | OAuth, or Base64-encoded credentials | value | Two authentication methods are supported: OAuth (recommended) and Base64-encoded credentials. | 15 |
| Huntress | Write actions | Read-only | X | the AI can surface your data, but it can't make changes to your account | 15 |
| Huntress | Audit log | Not published | not published | No audit wording on the pages read. | 15, 16 |
| MSP360 | Connects your AI to | MSP360 Managed Backup | value | It connects your MSP360 Managed Backup account to Claude and ChatGPT | 18 |
| MSP360 | Status on its own page | Announced 2026-11-03, for all its MSP backup users. Its help article says results are not for production | value | November 3, 2026; MSP360 MCP Server is available to all MSP backup Software users.; Results provided by the MCP server should be reviewed and verified before use and are not recommended for production environments. | 18, 19 |
| MSP360 | Price of the MCP server | Nothing more to buy. No paid Claude or ChatGPT plan is needed | check | There’s nothing to buy on either side: you don’t need a paid Claude or ChatGPT plan.; No paid ChatGPT or Claude subscription required. | 18, 20 |
| MSP360 | AI clients it names | Claude (desktop and mobile apps) and ChatGPT | value | It works with Claude (desktop and mobile apps) and ChatGPT. | 18, 20 |
| MSP360 | Sign-in and access | An MSP360 Managed Backup account. The provider sets access to Disabled, Read-only or Full | value | Add the MSP360 MCP Server to Claude or ChatGPT as a custom connector and sign in with your MSP360 Managed Backup account.; Set MCP access account-wide to Disabled, Read-only, or Full access | 18, 20 |
| MSP360 | Write actions | On Full access. Read-only mode blocks every change | check | It reads data and it makes changes, within what your account is allowed to do.; cannot create, modify, start, or delete anything. | 18, 20 |
| MSP360 | Audit log | Not published | not published | Searched the blog, product page, manage-MCP article and connect-Claude article. | 18, 20, 21, 19 |
| Hudu | Connects your AI to | Hudu | value | Hudu v2.42.0 introduces the Hudu MCP Server | 22 |
| Hudu | Status on its own page | In release 2.42.0, 2026-05-14 | value | 5/14/2026; Hudu v2.42.0 introduces the Hudu MCP Server | 22 |
| Hudu | Price of the MCP server | Included with a Hudu subscription, at no added cost | check | The Hudu MCP Server is included with your existing Hudu subscription at no additional cost. | 23 |
| Hudu | AI clients it names | Claude, ChatGPT, Cursor, Copilot Studio | value | Works with Claude, ChatGPT, Cursor, and Copilot Studio | 23 |
| Hudu | Sign-in and access | OAuth-based | value | Every action through the Hudu MCP Server runs through OAuth-based authentication | 23 |
| Hudu | Write actions | Edit documents, update asset records, run process checklists, apply compliance labels | check | draft or edit documentation, look up and update asset records, create and run process checklists (onboarding, offboarding, maintenance), apply compliance labels | 23 |
| Hudu | Audit log | AI actions appear in Hudu’s activity logs | check | Article edits, asset changes, and other AI-initiated actions appear in Hudu's activity logs the same way manual changes do | 23 |
| Liongard | Connects your AI to | Liongard | value | Liongard partners can enable direct, secure access to Liongard's Model Context Protocol (MCP) server | 24 |
| Liongard | Status on its own page | Announced 2026-04-16. Its docs page gives no stage | value | April 16, 2026; The new MCP server opens secure, API-driven access to LiongardIQ asset intelligence | 25, 24 |
| Liongard | Price of the MCP server | No amount. Its own pages differ: the docs name an MCP subscription, and another page says MCP is included | value | An active LiongardIQ and Liongard MCP subscription; These tokens incur an additional cost and provide more limited, controlled access than the global token.; MCP is now included for all Liongard partners | 24, 26, 27 |
| Liongard | AI clients it names | Claude, Cursor, n8n, VS Code, OpenAI ChatKit, Zed, Copilot Studio | value | Claude (Chat, Cowork, Code) Cursor n8n VS Code OpenAI ChatKit Zed Microsoft Copilot Studio | 24 |
| Liongard | Sign-in and access | OAuth 2.0 sign-in, or a bearer token | value | OAuth 2.0 sign-in: Connect your MCP client with a browser sign-in and a one-click approval using your existing Liongard login.; Bearer Tokens: Generate authentication tokens through the Admin console. | 24 |
| Liongard | Write actions | Not published. Its tool list holds lookups | not published | enables conversational, chat-driven interaction with managed environments for real-time insight and action | 24 |
| Liongard | Audit log | Not published for MCP | not published | Logs AI-related operations and user management actions to provide full visibility into administrative and automation activity. | 28 |
| Pax8 | Connects your AI to | Pax8 | value | enabling AI assistants like Claude, Copilot, and Cursor to access your Pax8 data and perform actions on your behalf | 29 |
| Pax8 | Status on its own page | No stage in the docs | not published | Updated 6 months ago | 29 |
| Pax8 | Price of the MCP server | Not published | not published | No price wording on any page read (searched price, pricing, free, cost, no additional). | 29 |
| Pax8 | AI clients it names | Cursor, Claude, Copilot, VS Code, n8n | value | Cursor Claude Copilot VSCode n8n (OAuth 2.0 only) | 29 |
| Pax8 | Sign-in and access | OAuth 2.0 or a legacy token. You control which API endpoints are accessible | value | Choose your preferred connection method (OAuth 2.0 or Pax8 token (legacy)); You maintain control over which API endpoints are accessible and can revoke access at any time. | 29 |
| Pax8 | Write actions | It can perform actions on your behalf | check | Ask questions and initiate actions using simple, intuitive prompts.; enabling AI assistants like Claude, Copilot, and Cursor to access your Pax8 data and perform actions on your behalf | 29 |
| Pax8 | Audit log | All actions are logged and can be audited | check | All actions performed through the MCP Server are fully logged and auditable. | 29 |
Text outside the grid on this page repeats grid cells and carries the same source numbers.
StackJack next to gateways and platforms
Sources, each checked 2026-11-15
- 1WYRE Conduit. https://conduit.wyre.ai/pricing
- 2WYRE Conduit. https://conduit.wyre.ai/vendors.json
- 3WYRE Conduit. https://conduit.wyre.ai/features
- 4WYRE Conduit. https://conduit.wyre.ai/security
- 5WYRE Conduit. https://conduit.wyre.ai/docs/getting-started/
- 6WYRE Conduit. https://conduit.wyre.ai/docs/guides/workflows/
- 7WYRE Conduit. https://conduit.wyre.ai/docs/reference/agents-concepts/
- 8Stackyapper. https://stackyapper.dev/
- 9Stackyapper. https://stackyapper.dev/how-it-works
- 10Stackyapper. https://stackyapper.dev/pricing
- 11Stackyapper. https://stackyapper.dev/apps
- 12Stackyapper. https://docs.stackyapper.dev/
- 13Stackyapper. https://stackyapper.dev/trust
- 14Stackyapper. https://stackyapper.dev/llms.txt
- 15Pipedream. https://pipedream.com/docs/connect/mcp.md
- 16Pipedream. https://pipedream.com/
- 17Pipedream. https://pipedream.com/docs/conduit/cloud.md
- 18Pipedream. https://pipedream.com/apps
- 19Pipedream. https://pipedream.com/conduit
- 20Pipedream. https://pipedream.com/pricing
- 21Pipedream. https://pipedream.com/docs/workflows.md
- 22Pipedream. https://pipedream.com/docs/connect/components/triggers.md
- 23Pipedream. https://pipedream.com/docs/conduit/use/mcp-reference.md
- 24Zapier MCP. https://zapier.com/mcp
- 25Zapier MCP. https://docs.zapier.com/mcp/manage/security.md
- 26Zapier MCP. https://docs.zapier.com/mcp/get-started/build-a-workflow.md
- 27Zapier MCP. https://zapier.com/pricing
- 28Zapier MCP. https://docs.zapier.com/mcp/overview/how-connections-work.md
- 29Zapier MCP. https://docs.zapier.com/mcp/get-started/rollout/overview.md
- 30Zapier MCP. https://zapier.com/enterprise
- 31Zapier MCP. https://docs.zapier.com/mcp/embed/getting-started.md
- 32Composio. https://composio.dev/pricing
- 33Composio. https://composio.dev/mcp-gateway
- 34Composio. https://composio.dev/toolkits/halopsa
- 35Composio. https://composio.dev/pricing/platform
- 36Composio. https://docs.composio.dev/llms.txt
- 37Composio. https://docs.composio.dev/docs/triggers.md
- 38Composio. https://docs.composio.dev/docs/governance/approvals.md
- 39n8n. https://n8n.io/
- 40n8n. https://n8n.io/pricing/
- 41n8n. https://n8n.io/mcp/
- 42n8n. https://n8n.io/integrations/halopsa/
- 43n8n. https://n8n.io/integrations/
- 44n8n. https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-langchain.mcptrigger/
- 45n8n. https://n8n.io/legal/security/
- 46n8n. https://n8n.io/enterprise/
- 47Cloudflare. https://developers.cloudflare.com/changelog/post/2026-09-24-mcp-portals-ga/
- 48Cloudflare. https://developers.cloudflare.com/cloudflare-one/access-controls/ai-controls/mcp-portals/
- 49Cloudflare. https://blog.cloudflare.com/mcp-portal-writeguard-private-beta/
- 50Cloudflare. https://www.cloudflare.com/plans/zero-trust-services/
- 51Cloudflare. https://developers.cloudflare.com/cloudflare-one/traffic-controls/tiered-policies/tenant-api/
- 52Cloudflare. https://www.cloudflare.com/trust-hub/compliance-resources/
- 53Rewst. https://rewst.io/resources/news/rewst-unveils-suite-of-ai-capabilities-from-workflow-creation-using-natural-language-to-an-mcp-server-connecting-ai-agents-with-automations
- 54Rewst. https://rewst.io/platform/fabric
- 55Rewst. https://rewst.io/platform/integrations
- 56Rewst. https://rewst.io/pricing
- 57Rewst. https://rewst.io/
- 58Rewst. https://rewst.io/trust-center
- 59Rewst. https://rewst.io/platform/workflow-builder
- 60Hatz AI. https://www.hatz.ai/
- 61Hatz AI. https://docs.hatz.ai/en/articles/13722060-halo-psa-integration
- 62Hatz AI. https://docs.hatz.ai/en/articles/13727834-pax8-integration
- 63Hatz AI. https://docs.hatz.ai/en/articles/12430282-list-of-integrations
- 64Hatz AI. https://www.hatz.ai/partners
- 65Hatz AI. https://www.hatz.ai/credit-usage-faq
- 66Hatz AI. https://docs.hatz.ai/en/articles/14115976-custom-mcp-servers
- 67Hatz AI. https://docs.hatz.ai/en/articles/15394302-autotask-mcp-integration
- 68Hatz AI. https://www.hatz.ai/msps
- 69Hatz AI. https://hatz-ai.canny.io/changelog
- 70Pax8 (8os and 8bot). https://www.pax8.com/en-us/8os-agentic-workforce-operating-system/
- 71Pax8 (8os and 8bot). https://www.pax8.com/en-us/8bot-managed-ai-coworker/
- 72Pax8 (8os and 8bot). https://www.pax8.com/blog/introducing-8os-and-8bot/
- 73Kaseya. https://www.kaseya.com/press-release/kaseya-previews-next-evolution-of-kaseya-intelligence-at-connect-europe-2026/
- 74Kaseya. https://www.kaseya.com/platform/
- 75Kaseya. https://webinar.kaseya.com/kaseya-innovations-for-the-microsoft-ecosystem
- 76Kaseya. https://www.kaseya.com/platform/kaseya-365/
- 77Kaseya. https://www.kaseya.com/platform/intelligence/
- 78Kaseya. https://www.kaseya.com/trust-center/information-security/
- 79Thread. https://docs.getthread.com/super-magic/add-external-tools-to-super-magic-with-connectors-super-mcp.md
- 80Thread. https://docs.getthread.com/super-magic/thread-mcp-server-developer-guide.md
- 81Thread. https://docs.getthread.com/llms.txt
- 82Thread. https://www.getthread.com/thread-pricing
- 83Thread. https://docs.getthread.com/super-magic/control-who-can-connect-ai-apps-to-thread-mcp.md
- 84Thread. https://docs.getthread.com/super-magic/issue-mcp-automation-keys-for-team-members.md
- 85Thread. https://docs.getthread.com/security-billing/soc-2-and-compliance-status.md
- 86Thread. https://trust.getthread.com/
- 87Thread. https://docs.getthread.com/super-magic/faq.md
- 88Thread. https://docs.getthread.com/super-magic/meet-super-magic-your-ai-assistant-in-the-inbox.md
- 89Bifrost (gobifrost.com). https://gobifrost.com/features
- 90Bifrost (gobifrost.com). https://gobifrost.com/
- 91Bifrost (gobifrost.com). https://github.com/gobifrost/bifrost
- 92Bifrost (gobifrost.com). https://gobifrost.com/compare
- 93Bifrost (gobifrost.com). https://gobifrost.com/llms-full.txt
Grid cells
| Vendor | Row | The cell says | Mark | The vendor’s words | Source |
|---|---|---|---|---|---|
| WYRE Conduit | Connects your AI to | 230 vendor connectors | value | All 230 vendor connectors | 1, 2 |
| WYRE Conduit | Status on its own page | Shipped, as its features page says | value | Every card is a real, shipped capability. Nothing aspirational. | 3 |
| WYRE Conduit | MSP tools on its own pages | HaloPSA, ConnectWise PSA, Autotask PSA and NinjaOne are in its vendor list | value | Read from the vendor list that WYRE publishes. | 2 |
| WYRE Conduit | AI clients it names | Claude, ChatGPT, Microsoft Copilot, Cursor, Windsurf | value | Conduit works with any MCP-capable AI tool, including Claude, ChatGPT, Microsoft Copilot, Cursor, Windsurf, and more. | 1 |
| WYRE Conduit | Price on its public page | $39 for each seat a month, flat | check | $39 / seat / month | 1 |
| WYRE Conduit | Free plan | A 14-day trial | value | 14-day free trial. | 1 |
| WYRE Conduit | What counts against the plan | Seats, for people and AI agents. The first 2 agent seats are free | value | Your first 2 AI agent seats are free | 1 |
| WYRE Conduit | Sign-in and access | OAuth 2.1 with PKCE. A tool allowlist for each role, deny by default | value | Your AI clients connect over OAuth 2.1 + PKCE; Decide exactly which vendor tools each role is allowed to invoke. Deny-by-default; grant per-tool, per-role. | 3 |
| WYRE Conduit | Write actions | No read or write switch is stated. Its control is the tool allowlist | not published | The control it describes is the allowlist in the sign-in row. | 3 |
| WYRE Conduit | Audit log | Every tool call. Tool arguments are recorded only if you turn that on | check | Conduit logs every tool call your team makes: who ran what, against which vendor, when.; optionally record tool arguments and result summaries. Off by default | 3 |
| WYRE Conduit | A setup for each customer | Customers are kept apart, each with its own users, vendors and audit log | check | Run an estate of customers and sub-customers, each cleanly isolated with its own users, vendors, and audit log. | 3 |
| WYRE Conduit | SOC 2 | Not certified. Its page says it is preparing for SOC 2 Type 2 | X | SOC 2 Not certified Preparing for SOC 2 Type 2. | 4 |
| WYRE Conduit | Runs in your own cloud | Hosted by WYRE. Its docs mention self-hosted deployments | value | Conduit is a hosted MCP gateway.; Self-hosted deployments can skip the card field. | 5 |
| WYRE Conduit | Workflows or agents | Workflows that Conduit runs for you, and a library of pre-built agents | check | A workflow is a named, versioned definition; that Conduit runs for you, on a schedule or on demand, against your org's already-connected vendor tools; Conduit ships with a library of pre-built agents | 6, 7 |
| WYRE Conduit | Triggers | On a schedule, by hand, or from a webhook | check | A workflow can run on a schedule or when someone starts it, under one runner identity per deployment, with full run history in the console.; A run only knows its trigger's kind (schedule, manual, or webhook) | 6 |
| WYRE Conduit | A person approves an action | An admin approves a workflow’s tool list before it can run. Approval of each action is not published | value | That list is what an admin reviews and approves before a workflow is allowed to do anything. | 6 |
| Stackyapper | Connects your AI to | 5,200+ app listings for business tools | value | Connect your business tools to Claude, ChatGPT, and the AI your team already uses.; 5,200+ app listings | 8 |
| Stackyapper | Status on its own page | Live. Each app is marked Certified or Beta. Its MSP features are in preview | value | Apps labeled beta are not represented as fully supported.; MSP capabilities are in preview with staged availability | 9, 10 |
| Stackyapper | MSP tools on its own pages | ConnectWise PSA, NinjaOne, Hudu (Certified). HaloPSA, Autotask PSA (Beta) | value | ConnectWise PSA PSA Certified; NinjaOne RMM Certified; Hudu Documentation Certified; HaloPSA PSA Beta; Autotask PSA PSA Beta | 11 |
| Stackyapper | AI clients it names | Claude, ChatGPT, Codex, Claude Code, Hatz AI, Copilot Studio | value | Point a supported client at the Stackyapper remote MCP endpoint: Claude, ChatGPT, Codex, Claude Code, or Hatz AI.; Add Stackyapper to Claude, ChatGPT, Codex, Copilot Studio, or another MCP client. | 9, 12 |
| Stackyapper | Price on its public page | Standard: $99 a month. Business: $199 a month | check | Standard $99 / month; Business $199 / month | 10 |
| Stackyapper | Free plan | Not published | not published | No statement on the page read. | 10 |
| Stackyapper | What counts against the plan | The plan, with no charge for each user, app or call. A customer workspace is $119 a month more | value | Unlimited users, apps, and tool calls.; There is no per-user charge.; Each billable customer workspace $119 per month | 10 |
| Stackyapper | Sign-in and access | Microsoft and Google sign-in. SAML and SCIM on Business. Admins allow or block AI clients | value | Business workspaces can use managed SAML and Microsoft Entra SCIM for company sign-in and user lifecycle management.; An unapproved client is stopped during sign-in, before the consent screen, and never receives a grant or token | 13 |
| Stackyapper | Write actions | Reads, writes and deletes are treated differently. A risky action can need a confirmation | check | Reads, writes, deletes, financial actions, and credential-sensitive actions can receive different treatment.; Actions that require confirmation are bound to the arguments the user reviewed. | 13 |
| Stackyapper | Audit log | Metadata of each call. Export on Business | check | Administrators can review operational metadata such as the actor, workspace, app, tool, result, client, and duration.; Everything in Standard, plus managed employee access, audit export, and 365 days of activity history. | 13, 10 |
| Stackyapper | A setup for each customer | Customer administration and workspace switching, on Business. Its MSP features are in preview | not final | Business includes customer administration and workspace switching after MSP or consultant verification.; MSP capabilities are in preview with staged availability | 10 |
| Stackyapper | SOC 2 | Not certified. Its page says the status is preparation | X | Stackyapper is not currently SOC 2 certified, and no examination has been completed.; Current status: Preparation. | 13 |
| Stackyapper | Runs in your own cloud | Hosted. Credentials are kept on its servers | value | hosted MCP access and policy layer; Stackyapper keeps connected-app credentials encrypted on its servers | 14 |
| Stackyapper | Workflows or agents | A policy layer between AI clients and apps. It brings no agent of its own | value | It does not replace your AI assistant or require you to adopt one proprietary agent. | 9 |
| Stackyapper | Triggers | Not published | not published | Its pages describe a policy layer, not an automation product. | 9 |
| Stackyapper | A person approves an action | A confirmation for higher-risk actions | check | Actions that require confirmation are bound to the arguments the user reviewed. | 13 |
| Pipedream | Connects your AI to | 3,000+ apps, with an MCP server for each app | value | Pipedream offers a dedicated MCP (Model Context Protocol) server for all of our 3,000+ integrated apps. | 15 |
| Pipedream | Status on its own page | Conduit is in early access. Its docs list a hosted Conduit Cloud | not final | For companies · Early access; Conduit Cloud is free for up to 100 users. | 16, 17 |
| Pipedream | MSP tools on its own pages | ConnectWise PSA, Autotask PSA and NinjaOne are in its app catalog | value | Every NinjaOne action is exposed as an MCP tool on Pipedream's remote server.Read from Pipedream’s own app catalog. | 18 |
| Pipedream | AI clients it names | Claude, Cursor, ChatGPT, or any other AI assistant | value | Conduit gives you 3,000+ connectors for Claude, Cursor, ChatGPT, or any other AI assistant | 19, 15 |
| Pipedream | Price on its public page | Startup: $99 a month billed yearly, or $150 monthly | check | $99 per month; $150 per month | 20 |
| Pipedream | Free plan | Free to build. Hosted MCP servers are free for your own use. Conduit Cloud is free for up to 100 users | check | Free to build. Pay when your integrations go to production.; Anyone can use Pipedream's hosted MCP servers for their own use for free; Conduit Cloud is free for up to 100 users. | 20, 15, 17 |
| Pipedream | What counts against the plan | External users, then credits. Each MCP tool call uses credits | value | 100 users included, then $2 per user; 10,000 usage credits included; Running an action, calling a tool through MCP, sending a request through the Connect proxy | 20 |
| Pipedream | Sign-in and access | Connect: each user’s credentials are kept apart. Conduit: groups and policies for each tool | value | Each user's credentials are stored securely and isolated from other users; Groups and policies decide who can use which connectors and tools. Every call is authenticated, authorized, and logged. | 15, 19 |
| Pipedream | Write actions | Each NinjaOne action is an MCP tool, as one example | check | Every NinjaOne action is exposed as an MCP tool on Pipedream's remote server. | 18 |
| Pipedream | Audit log | Conduit: every sign-in, policy change and tool call. Connect: not stated | check | Every sign-in, policy change, and tool call is recorded. | 19 |
| Pipedream | A setup for each customer | Not published | not published | No statement on the page read. | 19 |
| Pipedream | SOC 2 | Type 2, on every plan | check | SOC 2 Type 2 and GDPR Included Included Included | 20 |
| Pipedream | Runs in your own cloud | Conduit: you can self-host it, or use its managed cloud. Connect: not published | check | Self-host a single Docker image in your own infrastructure, or use the managed cloud.; Self-hosted or in the cloud. Early access is open.; You can also self-host Conduit on your own infrastructure. | 19, 17 |
| Pipedream | Workflows or agents | Pipedream Workflows shuts down on 2027-03-31 | value | Pipedream Workflows and String will shut down on March 31, 2027. | 21 |
| Pipedream | Triggers | In Connect: event triggers that send to a webhook URL or a workflow. A schedule trigger is not published | check | Triggers (also called sources) are different from actions; When deploying a source, you can define either a webhook URL or a Pipedream workflow ID to consume those events.; Some app-based triggers poll for new events rather than using webhooks. | 22 |
| Pipedream | A person approves an action | Not published as a Pipedream control | not published | Its docs point to the approval setting of the AI client. | 23 |
| Zapier MCP | Connects your AI to | 9,000+ apps | value | your agent searches Zapier's 9,000+ apps and 66,000+ triggers and actions | 24 |
| Zapier MCP | Status on its own page | Generally available. Next Gen Zaps, which build workflows through MCP, are in early access | value | Zapier MCP is generally available and operates under Zapier's SOC 2 Type II certification.; Next Gen Zaps is in early access by request. | 25, 26 |
| Zapier MCP | MSP tools on its own pages | None named on its MCP page | not published | The page text names no PSA or RMM product. | 24 |
| Zapier MCP | AI clients it names | Claude, ChatGPT, Cursor, and any other AI client | value | Zapier MCP lets Claude, ChatGPT, Cursor, and any other AI client take real action in the apps that run your business | 24 |
| Zapier MCP | Price on its public page | Professional: from $19.99 a month, with yearly billing and 750 tasks | check | Starting from $19.99 /month | 27 |
| Zapier MCP | Free plan | MCP is in every plan. Free has 100 tasks a month | check | Zaps, Tables, and Forms included (100 tasks per month); Zapier MCP is included in every Zapier plan. | 27, 24 |
| Zapier MCP | What counts against the plan | Tasks. Each MCP tool call uses two | value | Each tool call uses two tasks from your existing task quota | 24 |
| Zapier MCP | Sign-in and access | OAuth from inside the MCP client, or a connection token. Each member signs in as themselves | value | The default is OAuth from inside the MCP client.; SOC 2 Type II certified with SAML SSO, OAuth, and connection token auth.; each member still signs in and runs tool calls as themselves | 28, 25, 29 |
| Zapier MCP | Write actions | It takes action in your apps | check | Zapier MCP lets Claude, ChatGPT, Cursor, and any other AI client take real action in the apps that run your business | 24 |
| Zapier MCP | Audit log | Agent actions show in the audit log. Its MCP page has an action history | check | Every agent action taken through Zapier is visible in the audit log.; MCP action History gives you a record of what ran. | 30, 24 |
| Zapier MCP | A setup for each customer | It can be embedded in another product, for that product’s customers | value | You want to embed Zapier MCP in your own product for your customers.; Embedding Zapier MCP gives your users governed access to 9,000+ apps without leaving your MCP client's interface. | 29, 31 |
| Zapier MCP | SOC 2 | Type II | check | SOC 2 Type II certified | 24 |
| Zapier MCP | Runs in your own cloud | Not published | not published | No statement on the pages read. | 27, 30 |
| Zapier MCP | Workflows or agents | Zaps, Tables and Forms are included on its Free plan | check | Zaps, Tables, and Forms included (100 tasks per month) | 27 |
| Zapier MCP | Triggers | In Next Gen Zaps: a schedule, an app event, or on demand | not final | Include the apps involved, what triggers it (a schedule, an app event, or on-demand only); Next Gen Zaps is in early access by request. | 26 |
| Zapier MCP | A person approves an action | A person approves the build plan of a Next Gen Zap. Approval of each tool call is not published | value | Review the plan before continuing.; Once you approve the plan, your agent writes the workflow's TypeScript code | 26 |
| Composio | Connects your AI to | 1,700+ apps, on Free and Standard | value | Connect 1,700+ apps to your AI for free. | 32 |
| Composio | Status on its own page | No stage label on its MCP gateway page | not published | The secure MCP gateway for enterprise AI agents | 33 |
| Composio | MSP tools on its own pages | HaloPSA: 85 tools in its toolkit | value | 85 Tools | 34 |
| Composio | AI clients it names | Claude, Codex, Cursor and more. Its HaloPSA page names Claude, ChatGPT, Cursor | value | Connect your apps to Claude, Codex, Cursor and more.; connect your AI agents and chatbots (Claude, ChatGPT, Cursor, etc) with HaloPSA MCP | 32, 34 |
| Composio | Price on its public page | Standard: $18 for each seat a month | check | $18 per seat / month | 32 |
| Composio | Free plan | $0. 10,000 tool calls a month, for up to 3 members | check | 10,000 Each action your AI takes in an app; Members Up to 3 | 32 |
| Composio | What counts against the plan | Tool calls. Standard adds 10,000 for each seat, and $18 in credits | value | Every seat adds 10,000 tool calls and $18 in credits to one pool that your whole workspace shares. | 32 |
| Composio | Sign-in and access | SSO for its gateway. Policies for the org, a team and a user | value | SSO verifies the user before any tool resolves; Org, team, and user policies | 33 |
| Composio | Write actions | A team policy can block destructive actions | check | Destructive actions blocked | 33 |
| Composio | Audit log | On Enterprise. Its Platform tab lists audit logs on every plan | check | SCIM provisioning and audit logs; Audit logs A record of who changed what in your organization. Hobby shows the last 30 days. | 32, 35 |
| Composio | A setup for each customer | An MCP endpoint for each team | value | Each team gets its own MCP endpoint carrying only the tools it is permitted to use. | 33 |
| Composio | SOC 2 | Type II, as stated on its HaloPSA toolkit page | check | Composio is SOC 2 Type II compliant | 34 |
| Composio | Runs in your own cloud | Bring your own cloud, on Enterprise | check | Bring your own cloud (BYOC); Choose Enterprise when you need SCIM, SIEM streaming, deployment in your own cloud, or dedicated support. | 32 |
| Composio | Workflows or agents | Agents. Its gateway can start background agents from events | check | Build and operate agents with authentication, tool discovery, execution, and 1,700+ toolkits.; TRIGGERS Start background agents from events across your connected apps | 36, 33 |
| Composio | Triggers | Event triggers, sent to your webhook URL. A schedule trigger is not published | check | sends that event to your app as a structured payload; Composio delivers every event to one destination you control: your webhook URL. | 37 |
| Composio | A person approves an action | A session setting asks the user before a tool runs | not final | You can configure the session to require the user's approval before any tool runs; Approvals are experimental. | 38 |
| n8n | Connects your AI to | Apps through nodes, any API through HTTP, any agent through MCP | value | 500+ native integrations. Any API via HTTP. Any agent or coding assistant via MCP. | 39 |
| n8n | Status on its own page | Its MCP features are on every cloud plan, with no preview label. Its Agents are labelled preview | value | Instance-level MCP Included Included Included; Included on all n8n plans; Agents (preview) Included Included Included | 40, 41 |
| n8n | MSP tools on its own pages | HaloPSA and SyncroMSP have native nodes | value | HaloPSA integrations | Workflow automation with n8n | 42, 43 |
| n8n | AI clients it names | Claude, ChatGPT, Gemini, Cursor | value | Claude, ChatGPT, Gemini, or Cursor. | 41 |
| n8n | Price on its public page | Starter: 20 euros a month, billed yearly, with 2.5K executions | check | 20€ /mo, billed annually; 2.5K workflow executions with unlimited steps | 40 |
| n8n | Free plan | A free self-hosted edition. The cloud has a free trial with no card | check | Start free trial No credit card required; A standard, self-hosted version of n8n is available on GitHub. | 40, 39 |
| n8n | What counts against the plan | Workflow executions a month, not steps or users | value | All plans include unlimited users and workflows and every integration. Pricing is based on monthly workflow executions, regardless of complexity. | 40 |
| n8n | Sign-in and access | Its MCP server trigger can use n8n OAuth2 | value | n8n User Auth (OAuth2) : available on node version 2 and later. | 44 |
| n8n | Write actions | It depends on the workflow tools you expose. Human approval for tool calls is included | value | Human approval for tool calls Included Included Included | 40 |
| n8n | Audit log | Audit logging is on Enterprise | check | Audit logging Not included Not included Included | 40 |
| n8n | A setup for each customer | Not published | not published | No statement on the page read. | 40 |
| n8n | SOC 2 | Type 2 is listed on its enterprise page. The report is for enterprise customers | check | n8n aligns its security program to SOC 2, a standard framework for security compliance.; Our SOC 2 report is available to enterprise customers.; GDPR SOC 2 Type 2 | 45, 46 |
| n8n | Runs in your own cloud | Self-host with Docker or on-prem | check | Deploy on-prem or with Docker to keep your data in your own environment.; A standard, self-hosted version of n8n is available on GitHub. | 39, 40 |
| n8n | Workflows or agents | Every plan has unlimited workflows | check | All plans include unlimited users and workflows and every integration. Pricing is based on monthly workflow executions, regardless of complexity. | 40 |
| n8n | Triggers | An MCP Server Trigger is on every cloud plan | check | MCP Server Trigger Included Included Included | 40 |
| n8n | A person approves an action | Human approval for tool calls is included | check | Human approval for tool calls Included Included Included | 40 |
| Cloudflare | Connects your AI to | One endpoint for the MCP servers you approve | value | A portal gives users one endpoint for approved Model Context Protocol (MCP) servers. | 47 |
| Cloudflare | Status on its own page | Generally available since 2026-09-24. Its finer write controls are in private beta | value | MCP server portals are now generally available to all Cloudflare customers.; September 24, 2026; We are now bringing those controls to Cloudflare MCP server portals through a private beta. | 47, 48, 49 |
| Cloudflare | MSP tools on its own pages | The MCP servers you add to the portal | value | A portal gives users one endpoint for approved Model Context Protocol (MCP) servers. | 47 |
| Cloudflare | AI clients it names | Workers AI Playground and other clients that support remote MCP servers | value | using Workers AI Playground; or other MCP clients that support remote MCP servers | 48 |
| Cloudflare | Price on its public page | No price of its own. Cloudflare One is $7 for each user a month, pay as you go | check | Pay-as-you-go; $7/user/month | 50 |
| Cloudflare | Free plan | $0, for teams under 50 users | check | Free Plan; Best for teams under 50 users or enterprise proof-of-concept tests.; $0 forever | 50 |
| Cloudflare | What counts against the plan | Users | value | Pay-as-you-go; $7/user/month | 50 |
| Cloudflare | Sign-in and access | Access policies decide who sees each server. Service tokens for agents | value | The MCP server link will only appear in the portal for users who match an Allow policy.; Service token authentication for autonomous agents and machine-to-machine access | 48, 47 |
| Cloudflare | Write actions | Admins can rename tools and prompts. Finer write controls are in private beta | value | Admins can rename tools and prompts and edit their descriptions at the portal or server level without modifying the upstream MCP server.; WriteGuard is a shared policy, attribution, and auditing layer.; We are now bringing those controls to Cloudflare MCP server portals through a private beta. | 48, 49 |
| Cloudflare | Audit log | Tool, prompt and resource activity. Log export is on Enterprise plans | check | Cloudflare Access logs tool, prompt, and resource activity.; Export logs with Logpush Availability Only available on Enterprise plans. | 47, 48 |
| Cloudflare | A setup for each customer | A Tenant API for partner MSPs. Its page does not tie it to MCP portals | value | which allows Cloudflare-partnered managed service providers (MSPs) to set up and manage Cloudflare accounts and services for their customers | 51 |
| Cloudflare | SOC 2 | Type II | check | SOC 2 Type II; Certification to attest to Security, Confidentiality, and Availability controls in place in accordance to the AICPA Trust Service Criteria. | 52 |
| Cloudflare | Runs in your own cloud | Not published | not published | No statement on the page read. | 48 |
| Cloudflare | Workflows or agents | Not published | not published | A portal is an access and policy layer. | 48 |
| Cloudflare | Triggers | Not published | not published | A portal is an access and policy layer. | 48 |
| Cloudflare | A person approves an action | Not published | not published | A portal is an access and policy layer. | 48 |
| Rewst | Connects your AI to | Rewst workflows, through its MCP Server. Through Fabric, the tools connected to Rewst | value | giving agents like ChatGPT, Claude Desktop, and custom copilots secure, structured access to workflows, variables, and metadata; So can another MCP-compatible AI tool your team uses, connected to Rewst. | 53, 54 |
| Rewst | Status on its own page | MCP Server and MCP Trigger announced 2025-06-25. No stage on its Fabric page | value | Date: 25 June 2025; Rewst also introduced its Model Context Protocol (MCP) Server | 53, 54 |
| Rewst | MSP tools on its own pages | HaloPSA, ConnectWise PSA, NinjaOne, Datto RMM, Pax8 and more | value | HaloPSA Enables automation of PSA tasks.; NinjaOne Enables automation of remote monitoring and management.; Datto RMM Enables automation of RMM tasks.; ConnectWise PSA Enables automation of PSA tasks.; Pax8 Enables automation of cloud marketplace operations. | 55 |
| Rewst | AI clients it names | ChatGPT, Claude Desktop, custom copilots | value | giving agents like ChatGPT, Claude Desktop, and custom copilots secure, structured access to workflows, variables, and metadata | 53 |
| Rewst | Price on its public page | Not published. Priced for each client organization, by quote | not published | We price per client organization you run on the platform.; Demo the platform and get your quote | 56 |
| Rewst | Free plan | Not published | not published | No statement on the page read. | 56 |
| Rewst | What counts against the plan | Client organizations. AI use has an allowance, then your own AI key | value | Set at 5% of your monthly invoice and included by default.; Connect the API key from your preferred AI provider | 56 |
| Rewst | Sign-in and access | Fabric access is off by default and follows each integration’s permissions | value | Access is off by default and limited to the permissions already set for that integration. | 54 |
| Rewst | Write actions | The Rewst Agent can take action. AI tools can start workflows | check | The Rewst Agent can take action, too; MCP Trigger: Enables AI tools to initiate Rewst workflows via a lightweight API | 57, 53 |
| Rewst | Audit log | Actions show in the client tenant’s activity log. 30-day execution history | check | Every action shows up in the client tenant's activity log.; 30-day execution history | 54, 56 |
| Rewst | A setup for each customer | Each AI connection is scoped to one client tenant | check | Each connection is scoped to one client tenant and can't cross into another. | 54 |
| Rewst | SOC 2 | A SOC 2 report under NDA. The type is not stated | check | After signing an NDA, you will be able to access our SOC 2 Report and additional information security program resources. | 58 |
| Rewst | Runs in your own cloud | Not published | not published | No statement on the page read. | 56 |
| Rewst | Workflows or agents | An automation platform built for MSPs, with the Rewst Agent | check | One secure automation platform, built for MSPs, where the work actually runs. | 57 |
| Rewst | Triggers | Manual, on a schedule, or on an event such as a webhook | check | Start an automation manually, on a schedule, or when something happens elsewhere in your stack; a form submission, new ticket, status change, or webhook | 59 |
| Rewst | A person approves an action | Approval is a stated step, with a record of who approved | check | Describe it. Approve it. Run it.; you can see what ran, when, and who approved it | 57 |
| Hatz AI | Connects your AI to | Hatz is the AI application. It connects out to MCP servers and integrations | value | Connect to MCP servers, web search, document analysis, data connectors and integrations. | 60 |
| Hatz AI | Status on its own page | Live. Its Halo PSA and Pax8 integrations are labelled beta | value | Beta: This integration is currently available in Hatz beta. | 61, 62 |
| Hatz AI | MSP tools on its own pages | ConnectWise Manage, Autotask, HaloPSA | value | PSA and service management integrations such as ConnectWise Manage, Autotask, HaloPSA | 63 |
| Hatz AI | AI clients it names | Hatz is the client. It offers ChatGPT, Claude, Gemini and more | value | Access ChatGPT, Claude, Gemini and more in one place. | 60 |
| Hatz AI | Price on its public page | Not published | not published | No statement on the page read. | 64 |
| Hatz AI | Free plan | Not published | not published | No statement on the page read. | 64 |
| Hatz AI | What counts against the plan | Credits tied to AI use. They reset each month | value | Hatz Credits are units of measure directly tied to AI usage in the Hatz AI platform.; Unused credits from the previous month are not rolled into the next. | 65 |
| Hatz AI | Sign-in and access | Role-based access. A custom MCP server can use OAuth | value | Role-based access, spending limits and audit logs give you full visibility and control across your entire client base.; OAuth The server requires browser-based OAuth. | 60, 66 |
| Hatz AI | Write actions | For Autotask, always behind a confirmation step | check | always behind an explicit, reviewable confirmation step | 67 |
| Hatz AI | Audit log | Audit logs are named on its home page | check | Role-based access, spending limits and audit logs give you full visibility and control across your entire client base. | 60 |
| Hatz AI | A setup for each customer | A multi-tenant administrative portal | check | our multi-tenant administrative portal | 64 |
| Hatz AI | SOC 2 | Type 2 | check | SOC 2 Type 2 Independently audited controls | 60 |
| Hatz AI | Runs in your own cloud | Not published for Hatz. A custom MCP server is hosted by the customer | not published | Hatz AI only supports HTTP-based MCP servers. Customers must host and maintain these servers on their own infrastructure. | 66 |
| Hatz AI | Workflows or agents | Agents, workflows and automations | check | Build an Agent Create a custom AI agent; Hatz is purpose-built for MSPs to deploy AI tools, workflows, and agents both internally and to their customers. | 60, 68 |
| Hatz AI | Triggers | On a schedule. A webhook trigger is not published | check | Run it on a schedule, or click Run with new inputs whenever you need it. | 69 |
| Hatz AI | A person approves an action | Smart mode asks before a risky action. Manual mode asks more often | check | Smart runs low-risk actions on its own and asks before anything risky. Manual asks before more actions. | 69 |
| Pax8 (8os and 8bot) | Connects your AI to | 8os: supported AI, through a built-in gateway. 8bot: Microsoft 365 | value | Connect supported AI through the built-in gateway. Manage usage and budgets by customers.; At beta, 8bot connects to Microsoft 365: Outlook mail, Calendar, Slack, Teams and OneDrive and SharePoint. | 70, 71 |
| Pax8 (8os and 8bot) | Status on its own page | Its pages give two timings: early access now, and access from January 2027 | not final | 8bot is currently Early Access.; 8bot access begins in January 2027. | 70, 71 |
| Pax8 (8os and 8bot) | MSP tools on its own pages | Not published | not published | No statement on the pages read. | 70, 71 |
| Pax8 (8os and 8bot) | AI clients it names | None named for 8os. 8bot is its own AI coworker | not published | No statement on the page read. | 70 |
| Pax8 (8os and 8bot) | Price on its public page | Not published | not published | No statement on the pages read. | 70, 71 |
| Pax8 (8os and 8bot) | Free plan | Not published | not published | No statement on the pages read. | 70, 71 |
| Pax8 (8os and 8bot) | What counts against the plan | 8bot: a seat is one AI worker for one person. 8os: budgets by customer | value | Each seat is one always-on AI worker for one person, running on its own dedicated computer in the cloud.; You can also set budgets and spending limits by customer to help avoid unexpected costs. | 71, 72 |
| Pax8 (8os and 8bot) | Sign-in and access | 8bot uses the person’s Microsoft work account. Each app connection is approved one by one | value | Sign in with the Microsoft work account you already have.; Approve each 8bot app connection individually, starting with read-only access to the work it needs. | 71, 70 |
| Pax8 (8os and 8bot) | Write actions | With approval. Connections start read-only | check | Connections start read-only.; every one of those actions still asks for your approval | 71 |
| Pax8 (8os and 8bot) | Audit log | Not stated for the MSP. Pax8 support can read the activity record | not published | Pax8 support can access your AI coworker's activity record to resolve issues | 71 |
| Pax8 (8os and 8bot) | A setup for each customer | A multi-tenant provider model, in each client’s context | check | 8os is designed for a multi-tenant provider model, helping MSPs manage agents, access, usage and budgets in each client's context. | 70 |
| Pax8 (8os and 8bot) | SOC 2 | Not on the 8os and 8bot pages. The Pax8 footer shows ISO 27001 | not published | ISO 27001 Certified | 70, 71 |
| Pax8 (8os and 8bot) | Runs in your own cloud | Hosted by Pax8. In the U.S. during the beta | value | during the beta your data is hosted in the U.S. | 71 |
| Pax8 (8os and 8bot) | Workflows or agents | 8os builds agents. 8bot runs routines | check | Build agents from scratch. Turn proven workflows into repeatable client services.; Anything you ask for regularly becomes a routine with a name and a schedule. | 70, 71 |
| Pax8 (8os and 8bot) | Triggers | Schedules and triggers. A webhook trigger is not published | check | Anything you ask for regularly becomes a routine with a name and a schedule.; acts on schedules and triggers | 71 |
| Pax8 (8os and 8bot) | A person approves an action | Each action asks for your approval | check | every one of those actions still asks for your approval | 71 |
| Kaseya | Connects your AI to | The Kaseya platform, Claude and Microsoft Copilot Cowork, through APIs | value | an open platform built on APIs that delivers insights and automation wherever IT teams work; including the Kaseya platform, Claude and Microsoft Copilot Cowork | 73 |
| Kaseya | Status on its own page | Previewed 2026-06-17. Early access late 2026, general availability in 2027 | not final | Early access to these new Kaseya Intelligence capabilities is expected in late 2026, with general availability expected in 2027.; 17 June 2026 | 73 |
| Kaseya | MSP tools on its own pages | Its own stack: Datto RMM, Autotask PSA, IT Glue | value | Remote monitoring & Endpoint management Datto RMM Helpdesk & Service desk automation Autotask PSA IT documentation & knowledge management IT Glue | 74 |
| Kaseya | AI clients it names | Claude and Microsoft Copilot | value | Access Kaseya Intelligence through Microsoft Copilot, Claude and other AI tools | 75 |
| Kaseya | Price on its public page | Not published. Kaseya 365 offers a quote | not published | Request a quote | 76 |
| Kaseya | Free plan | Not published | not published | No statement on the page read. | 76 |
| Kaseya | What counts against the plan | Not published | not published | No statement on the page read. | 76 |
| Kaseya | Sign-in and access | Not published | not published | No statement on the page read. | 73 |
| Kaseya | Write actions | With technician approval | check | with technician approval, execute tasks, update tickets and generate reports automatically | 73 |
| Kaseya | Audit log | Not published | not published | No statement on the pages read. | 73, 77 |
| Kaseya | A setup for each customer | Not published | not published | No statement on the page read. | 73 |
| Kaseya | SOC 2 | Type II audits for its IT Complete modules. The scope for Intelligence is not stated | value | a history of exception-free SOC 2 Type II audits for a wide set of IT Complete modules | 78 |
| Kaseya | Runs in your own cloud | Not published | not published | No statement on the page read. | 73 |
| Kaseya | Workflows or agents | Digital Specialists: AI specialists for a role | check | Digital Specialists Free up time with role-level AI specialists | 77 |
| Kaseya | Triggers | Not published | not published | No statement on the page read. | 77 |
| Kaseya | A person approves an action | Technician approval | check | with technician approval, execute tasks, update tickets and generate reports automatically | 73 |
| Thread | Connects your AI to | Its own service-desk tools. Its connectors bring in NinjaOne, Notion, Zapier and Linear | value | Use Super Magic connectors (Super MCP) to bring Notion, Zapier, Linear, NinjaOne, and other tools into Inbox so you can act across them from any ticket. | 79 |
| Thread | Status on its own page | Its MCP server is generally available to all partners | value | Availability: generally available to all partners. | 80 |
| Thread | MSP tools on its own pages | Built on ConnectWise, Autotask and HaloPSA | value | Thread is the AI service desk for MSPs on ConnectWise, Autotask, and HaloPSA | 81 |
| Thread | AI clients it names | Claude Desktop, ChatGPT, Cursor, or any MCP client | value | Connect Claude Desktop, ChatGPT, Cursor, or any MCP client to Thread's service-desk API via OAuth 2.1 to search tickets, manage contacts, and run write actions. | 80 |
| Thread | Price on its public page | AI Essentials: $19 for each managed customer a month. AI Pro: $34 | check | $19 per managed customer/mo; $34 per managed customer/mo | 82 |
| Thread | Free plan | A 60-day money-back guarantee | value | 60-Day Money-Back Guarantee | 82 |
| Thread | What counts against the plan | Managed customers, not tokens | value | Your pricing is based on the number of customers you manage; there are no token meters or usage surprises | 82 |
| Thread | Sign-in and access | OAuth 2.1, one member for each connection. Admins choose who may connect | value | Workspace admins can decide who is allowed to connect at all, see every app that's connected, and cut off any app with one click.; Connect Claude Desktop, ChatGPT, Cursor, or any MCP client to Thread's service-desk API via OAuth 2.1 to search tickets, manage contacts, and run write actions. | 83, 84, 80 |
| Thread | Write actions | On Agentic plans. Other plans see read-only tools | check | Write and integration tools appear only on Agentic plans.; On a non-Agentic plan (no AI Pro), the connection sees only the seven read-only Super Magic tools | 80 |
| Thread | Audit log | Every tool call is logged under the member | check | every tool call is logged under the member identity behind the connection. | 80 |
| Thread | A setup for each customer | Priced and licensed for each managed customer | check | Your pricing is based on the number of customers you manage | 82 |
| Thread | SOC 2 | Not published in its docs. Its Trust Center has a SOC 2 section that lists an engagement letter | not published | Thread's certification status, continuous control monitoring, and the security documents available for request are published on the Thread Trust Center.; Compliance; SOC 2; Resources; Engagement Letter | 85, 86 |
| Thread | Runs in your own cloud | Not published | not published | No statement on the pages read. | 82, 81 |
| Thread | Workflows or agents | A Triage Agent and a Reminder Agent, on AI Pro | check | Currently, we offer a Triage Agent and a Reminder Agent | 82 |
| Thread | Triggers | A Flow trigger starts an agent | check | an agent is that same skill given a Flow trigger and a write scope so it runs unattended | 81 |
| Thread | A person approves an action | It asks for confirmation before any write. An agent in a Flow runs with no confirmation of each step | check | It reads freely, and before any write it shows a Confirm action card; It acts only with your confirmation.; nobody confirms each step there | 87, 88 |
| Bifrost (gobifrost.com) | Connects your AI to | Any API, through code. It has no vendor catalog | value | Reach any API directly instead of waiting for a vendor connector to exist. | 89 |
| Bifrost (gobifrost.com) | Status on its own page | Not stated on its site | not published | No statement on the page read. | 90 |
| Bifrost (gobifrost.com) | MSP tools on its own pages | Its README names NinjaOne, HaloPSA, Pax8 and Microsoft CSP as example integrations | value | Create integration modules for common platforms (NinjaOne, HaloPSA, Pax8, Microsoft CSP) | 91 |
| Bifrost (gobifrost.com) | AI clients it names | Claude Code, Codex, Cursor, Copilot | value | Install the Bifrost plugin for Claude Code or Codex, or point Cursor, Copilot, or any other agent at Bifrost. | 90 |
| Bifrost (gobifrost.com) | Price on its public page | Free and open source (AGPL). You host it | check | AGPL, self-hosted, no per-seat pricing, no lock-in. | 89 |
| Bifrost (gobifrost.com) | Free plan | It is free. No hosted plan is stated | check | AGPL, self-hosted, no per-seat pricing, no lock-in. | 89 |
| Bifrost (gobifrost.com) | What counts against the plan | No per-seat pricing. No meter is stated | value | AGPL, self-hosted, no per-seat pricing, no lock-in. | 89 |
| Bifrost (gobifrost.com) | Sign-in and access | Google, Microsoft and OIDC sign-in | value | Google, Microsoft, and OIDC sign-in. | 89 |
| Bifrost (gobifrost.com) | Write actions | Not published | not published | No statement on the page read. | 89 |
| Bifrost (gobifrost.com) | Audit log | Execution history, traces and cost | value | Execution history, traces, and cost across everything you run. | 89 |
| Bifrost (gobifrost.com) | A setup for each customer | A customer signs in and sees only its own data | check | Put an agent in Global and give every customer access, or deploy an app to one organization and one role. Customers sign in and see only their own data. | 90 |
| Bifrost (gobifrost.com) | SOC 2 | Not published. Its compare page says that self-hosting is your responsibility | not published | self-host = your responsibility | 92, 93 |
| Bifrost (gobifrost.com) | Runs in your own cloud | You self-host it | check | Self-host it, read it, fork it. | 90 |
| Bifrost (gobifrost.com) | Workflows or agents | Workflows, agents and apps, built in code | check | Develop and run workflows, agents, and apps, not only for your own team but for all of your customers. | 90 |
| Bifrost (gobifrost.com) | Triggers | Webhooks, schedules and topic events | check | Event sources, subscriptions, schedules, and topics to trigger reusable logic.; an HTTP endpoint that an external system (HaloPSA, Microsoft 365, a generic provider) calls.; a cron expression that fires on a recurring timer, with a timezone and an overlap policy | 89, 93 |
| Bifrost (gobifrost.com) | A person approves an action | Not published for agent or workflow runs | not published | The one approval in its docs is an administrator who approves a permission request at install. | 93 |
Text outside the grid on this page repeats grid cells and carries the same source numbers.
AI Governance vs MSP shadow-AI tools
Sources, each checked 2026-11-15
- 1ShadowLock. https://shadowlock.io/pricing
- 2ShadowLock. https://shadowlock.io/features/ai-governance
- 3Shadow Defend. https://shadowdefend.com/pricing/
- 4Shadow Defend. https://shadowdefend.com/solutions/msp/
- 5Shadow Defend. https://shadowdefend.com/security/
- 6DefensX. https://www.defensx.com/product/
- 7DefensX. https://www.defensx.com/blog/everyones-talking-about-shadow-ai/
- 8dope.security. https://dope.security/post/shadow-ai-tools-for-msps
- 9dope.security. https://www.dope.security/pricing
- 10N-able Shadow AI Visibility. https://www.n-able.com/blog/n-able-shadow-ai-visibility-is-now-available-to-help-govern-ai-usage-n-central-n-sight-and-adlumin
Grid cells
| Vendor | Row | The cell says | Mark | The vendor’s words | Source |
|---|---|---|---|---|---|
| ShadowLock | How it is priced | For each managed Windows device. 25-device minimum | value | priced per managed device per month, billed monthly with no annual commitment and a 25-device minimum per account; A device is any managed Windows endpoint that runs the ShadowLock agent. | 1 |
| ShadowLock | Price on its public page | Not published. A rate card by email | not published | we will email your rate card | 1 |
| ShadowLock | Free plan | A 14-day trial | value | Free 14-day trial. | 2 |
| ShadowLock | Client organizations | No cap on organizations. No fee for each user | check | Unlimited organizations; There is no per-user fee | 1 |
| ShadowLock | How it sees AI use | An endpoint agent, a browser extension and a Microsoft 365 OAuth scanner | value | Every tier includes the endpoint agent, browser extension, and Microsoft 365 OAuth scanner. | 1 |
| ShadowLock | Endpoint agent | Windows | check | A device is any managed Windows endpoint that runs the ShadowLock agent. | 1 |
| ShadowLock | Exposure check for a client | A Microsoft 365 OAuth scanner | check | Microsoft 365 OAuth app detection | 1 |
| ShadowLock | Checks or blocks AI use | An allow, warn or block policy engine | check | Allow / Warn / Block policy engine | 1 |
| ShadowLock | How long it keeps events | 90 days | value | 90-day event retention | 1 |
| ShadowLock | Your own storage | Not published | not published | No statement on the pages read. | 1, 2 |
| Shadow Defend | How it is priced | For each device | value | Per device. Each enrolled device is $4/month | 3 |
| Shadow Defend | Price on its public page | Pro is $4 for each device a month | check | $4 per device / month | 3 |
| Shadow Defend | Free plan | Starter is free, for up to 10 devices | check | Up to 10 devices | 3 |
| Shadow Defend | Client organizations | A policy set for each client | check | Every client gets its own policy set, risk rules, and audit trail. | 4 |
| Shadow Defend | How it sees AI use | A browser extension, deployed by MDM | value | A lightweight browser extension deploys across a client's fleet via MDM. | 4 |
| Shadow Defend | Endpoint agent | Covers AI use in the browser | value | We govern browser-based AI use on managed devices. | 4 |
| Shadow Defend | Exposure check for a client | Not published | not published | No statement on the pages read. | 3, 4, 5 |
| Shadow Defend | Checks or blocks AI use | Prompt inspection, with warn and block actions | check | Full prompt inspection; Warn & block actions | 3 |
| Shadow Defend | How long it keeps events | 7 days on Starter, 90 on Pro, 365 on Enterprise | value | Data retention 7 days 90 days 365 days | 3 |
| Shadow Defend | Your own storage | Storage location not stated. It stores prompt text itself | not published | Non-sensitive prompt text is stored to give you a usable audit record. | 5 |
| DefensX | How it is priced | Not published. Its page says it has flexible pricing tiers | not published | Yes! We offer flexible pricing tiers | 6 |
| DefensX | Price on its public page | Not published | not published | No statement on the page read. | 6 |
| DefensX | Free plan | Trial subscriptions are available | value | Trial subscriptions are also available. | 6 |
| DefensX | Client organizations | One multi-tenant console | check | from one multi-tenant console | 7 |
| DefensX | How it sees AI use | An agent and a browser extension, on managed and unmanaged devices | value | It runs as a lightweight agent and a browser extension, across managed and unmanaged devices alike | 7 |
| DefensX | Endpoint agent | A lightweight agent | check | It runs as a lightweight agent and a browser extension, across managed and unmanaged devices alike | 7 |
| DefensX | Exposure check for a client | A self-assessment scorecard, with no sign-up | check | Seven questions, about sixty seconds, no sign-up. | 7 |
| DefensX | Checks or blocks AI use | Keeps a record of AI prompts. Prompt logging is in its Premium package | value | keep a record of AI prompts; LLM Prompt Logging | 7, 6 |
| DefensX | How long it keeps events | 30 days on Core packages, 90 days on Premium | value | 30-day Log Retention; 90 Day Log Retention | 6 |
| DefensX | Your own storage | Not published | not published | No statement on the page read. | 6 |
| dope.security | How it is priced | Not published | not published | No statement on the pages read. | 8, 9 |
| dope.security | Price on its public page | Not published | not published | No statement on the pages read. | 8, 9 |
| dope.security | Free plan | A free trial with no card. The length is not stated | value | dope.security's instant trial needs no payment method; Feel confident before you buy and try it completely free. | 8, 9 |
| dope.security | Client organizations | One MSP login switches between customer tenants | check | one MSP login switches between customer tenants | 8 |
| dope.security | How it sees AI use | An endpoint agent, with the proxy on the device | value | dope.security puts the proxy on the device. | 8 |
| dope.security | Endpoint agent | Mac and Windows | check | it's an endpoint agent for Mac and Windows | 8 |
| dope.security | Exposure check for a client | Not published | not published | No statement on the page read. | 8 |
| dope.security | Checks or blocks AI use | Blocks personal accounts. DLP in Monitor, then Block | check | Dopamine DLP in Monitor, then Block; personal-account blocking | 8 |
| dope.security | How long it keeps events | Not published | not published | Its seven-day figure is a dashboard view, not a retention period. | 8 |
| dope.security | Your own storage | Not published. It can stream to a SIEM or to an AWS S3 bucket | not published | every client tenant can stream web transactions to Splunk, QRadar, Microsoft Sentinel, CrowdStrike, Taegis, or Datadog over HTTP, or to an AWS S3 bucket for any SIEM | 8 |
| N-able Shadow AI Visibility | How it is priced | Not published. Its post says to contact N-able | not published | Contact your N-able representative or visit n-able.com for details. | 10 |
| N-able Shadow AI Visibility | Price on its public page | Not published | not published | No statement on the page read. | 10 |
| N-able Shadow AI Visibility | Free plan | Not published | not published | No statement on the page read. | 10 |
| N-able Shadow AI Visibility | Client organizations | A view by tool, user, device and organization | value | A dedicated AI Usage tab with views by tool, user, device, and organization | 10 |
| N-able Shadow AI Visibility | How it sees AI use | An endpoint scan and DNS traffic analysis, with no new agents | value | Detection runs through existing technology, with no new agents, no new tools, and no additional infrastructure required.; Network-based detection: DNS traffic analysis | 10 |
| N-able Shadow AI Visibility | Endpoint agent | No new agents | value | Detection runs through existing technology, with no new agents, no new tools, and no additional infrastructure required. | 10 |
| N-able Shadow AI Visibility | Exposure check for a client | Not published | not published | No statement on the page read. | 10 |
| N-able Shadow AI Visibility | Checks or blocks AI use | Not published | not published | No statement on the page read. | 10 |
| N-able Shadow AI Visibility | How long it keeps events | Not published | not published | No statement on the page read. | 10 |
| N-able Shadow AI Visibility | Your own storage | Not published | not published | No statement on the page read. | 10 |
Text outside the grid on this page repeats grid cells and carries the same source numbers.
Left out, and why.
- Lists of MSP tools that a search did not find in a vendor’s catalog (n8n, Pipedream, Cloudflare). A missing search result is not a fact about the vendor.
- Whether Kaseya has an MCP server. No Kaseya page that was read uses the term, so the grid says nothing either way.
- Bifrost by Maxim AI. It is another product with the same name. The column is Bifrost at gobifrost.com.
- Counts that a vendor’s own pages disagree on: WYRE’s vendor count (two pages, two numbers), n8n’s integration count and Zapier’s action count. Each cell uses the figure from the page it cites.
- Action counts for single apps in a catalog (Pipedream). A count shows coverage and is not a score.
- Hatz AI’s cap on active tools in one chat. It comes from a document dated 2026-10-01, and a later release note does not say whether it changed.
- Roadmap items, such as N-able’s planned client registration.
- Add-on prices (Thread), a second plan table (Composio’s Platform tab) and plan prices above the first paid plan, except where a cell needs them.
- MSP360’s extra controls (IP allowlist, time-limited access). There is no matching row.
- The line in WYRE’s getting-started FAQ that says Conduit is not a workflow tool. Its workflows guide describes workflows that Conduit runs, and the cell cites the guide.
- Halo’s warning about its “Include all system tools” option, NinjaOne’s Katana AI controls, and every tool count on the vendor-server page. These stay out.
- DefensX’s own caveat about its scorecard, and Shadow Defend’s wording about a committed minimum (its own pages disagree).
- A SOC 2 row on the AI Governance page, a worked price example, and the products outside each comparison (Acronis, Check Point, FireTail, Microsoft Purview, MSPlex, MCPconnect, MSPbots).