Claude connectors
Claude with AI tools: which connectors and MCP servers exist and what to decide first
A plain guide for teams who want Claude working with translation, voice, web data, meeting notes, memory and document tools. Every fact comes from Anthropic's connector directory, the MCP Registry or a vendor's own page, as read on 10 November 2026. Settings and availability change, so check the cited pages before you act.
Claude with AI tools works through connectors that Anthropic's directory, read on 10 November 2026, lists from DeepL, ElevenLabs, Firecrawl, Granola, Mem0, Otter.ai and Unstructured. Browser Use, corsair, FalkorDB and Perplexity have no directory listing that we found, but each publishes an MCP server in the MCP Registry, and some of those run on your own machine. Several of these tools store or read meeting content, memories or web data, so a team should decide what Claude may access. Each section below says what the listing says and what we could not read.
plain answer
Which AI tools connect to Claude?
This page is about AI tools and services rather than programming frameworks: translation, voice agents, web data, meeting notes, memory and document processing. A directory connector is one that Anthropic lists in its connector directory. A registry server is an MCP server the vendor published in the MCP Registry, which is not listed in Anthropic's directory. Each section says which one it is.
Sources: Anthropic connector directory (https://claude.com/connectors) and MCP Registry (https://registry.modelcontextprotocol.io), read 10 November 2026.
how to
Claude Browser Use integration: is there a Claude connector for Browser Use?
Claude Browser Use integration: we found no Browser Use connector in Anthropic's directory on 10 November 2026. Browser Use publishes an MCP server in the MCP Registry as com.browser-use/browser-use, described there as "Browser agents and stealth cloud browsers for AI: run web tasks, fill forms, extract data." The latest entry shows version 0.13.11, published 21 October 2026, status active, with a remote address that needs an API key header.
Browser Use's own documentation describes an MCP server that gives a client a browser, and says that in local mode it drives the Chrome you have open, with your logins, so it can reach your accounts. It also says to keep the exact version pin, and the version it names differs from the registry's latest, so check which one applies before you install. An operations or research team is the likely user, in our reading.
Next step: if you want Browser Use to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: MCP Registry, com.browser-use/browser-use (https://registry.modelcontextprotocol.io/v0/servers?search=browser-use); Browser Use docs, MCP server (https://docs.browser-use.com/customize/mcp-server), all read 10 November 2026.
how to
Claude corsair integration: is there a Claude connector for corsair?
Claude corsair integration: we found no corsair connector in Anthropic's directory on 10 November 2026. The MCP Registry lists dev.corsair/mcp, described as "Use your connected apps from your agent: search synced data, act in apps and set up automations." It shows version 0.1.2, published 28 October 2026, status active, at https://mcp.corsair.dev/mcp.
We did not read corsair's own pages, so this page does not describe setup or permissions. The registry description says it can act in apps, so a team should review what it is allowed to change before connecting it. Developers building agent workflows are the likely users, in our reading.
Next step: if you want corsair to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: MCP Registry, dev.corsair/mcp (https://registry.modelcontextprotocol.io/v0/servers?search=dev.corsair), all read 10 November 2026.
how to
Claude DeepL integration: what does the DeepL connector do?
Claude DeepL integration works through the DeepL connector, made by DeepL. Anthropic's directory describes it as "Translate text and documents across 100+ languages, with your preferred styles applied". Listed tools include translating text, uploading and downloading documents, glossaries and style rules. Sign-in is required, the listing shows June 2026 as the date added, and it shows no beta label.
We did not find a DeepL page on this connector, so nothing here comes from DeepL itself. DeepL's developer site mentions a separate MCP server for its documentation, which is not the translation connector. A translation, localization or communications team is the likely user, in our reading.
Next step: if you want DeepL to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, DeepL (https://claude.com/marketplace/connectors/deepl), all read 10 November 2026.
how to
Claude ElevenLabs integration: what does the ElevenLabs connector do?
Claude ElevenLabs integration works through the ElevenLabs connector, made by ElevenLabs. Anthropic's directory describes it as a way to "Connect Claude to your ElevenLabs workspace to build with ElevenAgents without leaving your chat." Visible tools include creating agents and calculating usage. Sign-in is required, the listing shows July 2026 as the date added, and it shows no beta label.
ElevenLabs' own documentation describes ElevenAgents as a platform for agents that "accomplish tasks through natural dialogue", and lists a Hosted MCP option that lets you create and update agents from any MCP client. Because the connector creates agents that talk to people, decide who may publish one. A customer service or product team is the likely user, in our reading.
Next step: if you want ElevenLabs to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, ElevenLabs (https://claude.com/marketplace/connectors/elevenlabs); ElevenLabs docs, Agents overview (https://elevenlabs.io/docs/agents-platform/overview), all read 10 November 2026.
how to
FalkorDB MCP server: is there a Claude connector for FalkorDB?
FalkorDB MCP server: we found no FalkorDB connector in Anthropic's directory on 10 November 2026. The MCP Registry lists com.falkordb/QueryWeaver, described as "An MCP server for Text2SQL: transforms natural language into SQL using graph schema understanding." It is published as a container image with version 0.0.11 on 11 September 2025, status active, and the registry lists a local address and no public remote address.
That means the registry entry describes something you would run yourself, not a hosted connector. We did not read FalkorDB's own pages. A data or engineering team is the likely user, in our reading.
Next step: if you want FalkorDB to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: MCP Registry, com.falkordb/QueryWeaver (https://registry.modelcontextprotocol.io/v0/servers?search=com.falkordb), all read 10 November 2026.
how to
Claude Firecrawl integration: what does the Firecrawl connector do?
Claude Firecrawl integration works through the Firecrawl connector, made by Firecrawl. Anthropic's directory describes it as "Search and extract web data", and the listed tools include search, scrape and an agent tool. Sign-in is required, the listing shows July 2026 as the date added, and it shows no beta label.
Firecrawl's own page offers three ways to connect: keyless access with daily limits, sign-in with OAuth, or an API key, and says connections can be reviewed and revoked in MCP settings. The directory listing says sign-in is required, so the keyless option on Firecrawl's page may be a different entry point from the directory connector. A research or marketing team is the likely user, in our reading.
Next step: if you want Firecrawl to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, Firecrawl (https://claude.com/marketplace/connectors/firecrawl); Firecrawl docs, MCP server (https://docs.firecrawl.dev/mcp-server), all read 10 November 2026.
how to
Claude Granola integration: what does the Granola connector do?
Claude Granola integration works through the Granola connector, made by Granola. Anthropic's directory describes it as "The AI notepad for meetings, now in Claude. Ask about your notes, transcripts and meeting context." Listed tools include searching meetings, getting a transcript and creating a meeting note from a transcript. Sign-in is required, the listing shows January 2026 as the date added, and it shows no beta label.
We did not read a Granola page of its own. Meeting transcripts can hold sensitive content, so a team should decide whose notes Claude can search. Anyone who runs or follows up on meetings is the likely user, in our reading.
Next step: if you want Granola to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, Granola (https://claude.com/marketplace/connectors/granola), all read 10 November 2026.
how to
Claude Mem0 integration: what does the Mem0 connector do?
Claude Mem0 integration works through the Mem0 connector, made by Mem0. Anthropic's directory describes it as "Persistent memory for AI agents & assistants", and says it gives your agent persistent memory across sessions. Listed tools include adding, searching, updating and deleting memories, including deleting all memories. Sign-in is required, and the listing does not show a date added or a beta label.
Mem0's own page says the server is hosted by Mem0, that memories are stored in your Mem0 account and not locally, and that you connect with browser sign-in or an API key kept out of version control. Because the connector stores and deletes memories, decide what may be remembered and who can delete it. A developer or team building assistants is the likely user, in our reading.
Next step: if you want Mem0 to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, Mem0 (https://claude.com/marketplace/connectors/mem0); Mem0 docs, MCP (https://docs.mem0.ai/platform/mem0-mcp), all read 10 November 2026.
how to
Claude Otter.ai integration: what does the Otter.ai connector do?
Claude Otter.ai integration works through the Otter.ai connector, made by Otter.ai. Anthropic's directory describes it as "Otter's MCP Server integrates with Otter.ai's meeting intelligence platform". The three listed tools are get_user_info, search and fetch. Sign-in is required, the listing shows April 2026 as the date added, and it shows no beta label.
We did not read an Otter.ai page of its own. As with any meeting notes tool, a team should decide whose recordings Claude can search. Anyone who needs meeting summaries or follow-ups is the likely user, in our reading.
Next step: if you want Otter.ai to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, Otter.ai (https://claude.com/marketplace/connectors/otter-ai), all read 10 November 2026.
how to
Claude Perplexity integration: is there a Claude connector for Perplexity?
Claude Perplexity integration: we found no Perplexity connector in Anthropic's directory on 10 November 2026. The MCP Registry lists ai.perplexity/mcp-server, described as "Real-time web search, reasoning, and research through Perplexity's API". The latest entry shows version 1.2.1, published 27 August 2026, status active, as an npm package run locally, with no remote address listed.
We did not read Perplexity's own pages, so this page does not describe setup, keys or costs. A research or analyst team is the likely user, in our reading.
Next step: if you want Perplexity to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: MCP Registry, ai.perplexity/mcp-server (https://registry.modelcontextprotocol.io/v0/servers?search=ai.perplexity), all read 10 November 2026.
how to
Claude Unstructured integration: what does Unstructured Transform do?
Claude Unstructured integration works through Unstructured Transform, a connector made by Unstructured. Anthropic's directory describes it as "Turn any document into AI-ready structured data". Listed tools include starting a transform job, starting an extraction job and suggesting an extraction schema for a file. Sign-in is required, the listing shows July 2026 as the date added, and it shows no beta label.
Unstructured's documentation says its Transform service "turns any locally hosted document into structured data". The page we read does not describe an MCP server, so it does not confirm the connector. A team that processes many documents is the likely user, in our reading; Kriv's page on AI document handling and data extraction (https://www.kriv.ai/solutions/ai-document-handling-data-extraction) covers that kind of work.
Next step: if you want Unstructured to be one part of a larger Claude workflow, Kriv's agentic AI and automation service looks at the whole process, and is scoped and priced after a short discovery call.
Sources: Anthropic directory, Unstructured Transform (https://claude.com/marketplace/connectors/unstructured-transform); Unstructured docs, API overview (https://docs.unstructured.io/api-reference/overview), all read 10 November 2026.
security
What should a team decide before connecting an AI tool to Claude?
Start with the data. Meeting tools such as Granola and Otter.ai hold transcripts, Mem0 stores memories in your Mem0 account, and Firecrawl and Browser Use reach the web. Decide what Claude may read or store, and who can delete it. Browser Use's own page says that in local mode it drives the Chrome you have open, with your logins, so it can reach your accounts.
Next, separate directory connectors from registry servers. A registry server that runs locally, as the FalkorDB and Perplexity entries do, runs on your own machine, so review it with your IT team before installing.
Anthropic's directory listings for third-party connectors carry the same warnings: only use connectors from developers you trust, and Anthropic does not control which tools developers make available and cannot verify that they will work as intended.
Sources: Browser Use docs (https://docs.browser-use.com/customize/mcp-server), Mem0 docs (https://docs.mem0.ai/platform/mem0-mcp), MCP Registry (https://registry.modelcontextprotocol.io), Anthropic directory listings, read 10 November 2026.
kriv role
Where Kriv helps
Kriv AI helps teams roll Claude out in the work they already do: who connects, which connector to pilot, what needs a human review, and how staff are trained on their own process. Kriv AI has achieved the Select partner tier in the Claude Partner Network Services Track. For a team that means choosing which of these tools to pilot first, writing down what Claude may access and training staff on their own process.
Kriv did not build any connector or server on this page and does not speak for any vendor named here. Your own admins hold the access and the final say.
limits
What this page could not confirm
We read the Anthropic directory page for each connector listed above. We read the vendor's own page only for Browser Use, ElevenLabs, Firecrawl, Mem0 and Unstructured, and Unstructured's page did not describe an MCP server. We found no DeepL page on the connector, and we did not look for vendor pages for Granola, Otter.ai, corsair, FalkorDB or Perplexity, so those sections rest on the directory listing or the MCP Registry entry alone. We tested no connector and did not check which Claude plans or regions each supports. For anything not cited, check each vendor's and Anthropic's current pages.
Sources
Cited sources
- Anthropic directory, DeepL, read 2026-11-10
- Anthropic directory, ElevenLabs, read 2026-11-10
- Anthropic directory, Firecrawl, read 2026-11-10
- Anthropic directory, Granola, read 2026-11-10
- Anthropic directory, Mem0, read 2026-11-10
- Anthropic directory, Otter.ai, read 2026-11-10
- Anthropic directory, Unstructured Transform, read 2026-11-10
- Browser Use docs, MCP server, read 2026-11-10
- ElevenLabs docs, Agents overview, read 2026-11-10
- Firecrawl docs, MCP server, read 2026-11-10
- Mem0 docs, MCP, read 2026-11-10
- Unstructured docs, API overview, read 2026-11-10
- MCP Registry, com.browser-use/browser-use, read 2026-11-10
- MCP Registry, dev.corsair/mcp, read 2026-11-10
- MCP Registry, com.falkordb/QueryWeaver, read 2026-11-10
- MCP Registry, ai.perplexity/mcp-server, read 2026-11-10
Straight answers
Frequently asked questions about Claude with AI tools: which connectors and MCP servers exist and what to decide first
Does Claude work with DeepL?
Anthropic's directory lists a DeepL connector made by DeepL, described as translating text and documents across 100+ languages, with sign-in required (read 10 November 2026).
Is there a Claude connector for Perplexity?
We found none in Anthropic's directory on 10 November 2026. Perplexity publishes an MCP server in the MCP Registry, listed as an npm package run locally.
Does Claude work with Granola or Otter.ai?
Yes. Anthropic's directory lists a connector made by each, with sign-in required (read 10 November 2026). Decide whose meeting notes Claude can search.
Where does Mem0 store memories?
Mem0's own page says memories are stored in your Mem0 account, not locally (read 10 November 2026).
Can Kriv set up these connectors for us?
Kriv AI helps plan the rollout, the choice of tool and the access rules, and trains your team. Your admins authorize each connection in your own accounts.
Talk to the team that would do the work
Bring your requirements to a working session with the person who'll actually deliver.
Book a Discovery Call