6sense
Ask an AI agent about 6sense account insights, buying signals and keyword trends.
Generate, edit, merge, split, convert and organize PDFs in a cloud library.
Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.
Paste it into any MCP client. Setup docs
The PDF.net MCP server connects a PDF.net cloud document library to Claude, ChatGPT and any MCP-compatible agent. Anthropic's directory lists 20 tools that generate PDFs from a prompt, edit text inside them, merge, split, compress and convert files, and reorganise folders. Sign-in is OAuth with a single service-wide scope.
Verified connector
Listed by Anthropic as a partner connector in its Connectors Directory.
Connection checked by Agentman on .
Anthropic states this reflects the level of review a connector received, not a security audit.
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
pdfnet.mcp and offline_access, and its authorization server metadata repeats the same pair. Neither names a verb or an object. No boundary falls between listing your library and deleting a folder from it — the one place a scope split would have mattered on this surface.delete_resource is ambiguous about its target. The name says resource; PDF.net's capability table groups documents and folders together under Organize. A single call may remove a folder.has_mcp_app: true; the bundle is served over the authenticated session and we could not read it. Treat the click-driven surface as unknown.download_pdf_document returning a direct link and does not state whether it is session-scoped, time-limited or openable by anyone. We did not call the tool.get_profile is the only account-reading tool and PDF.net does not document it exposing quota or plan state.429 handling for the MCP server or its API.It works on documents already there, and it can bring in more. Every tool acts on your PDF.net cloud library rather than your disk. Two tools add to it: import_file_from_url fetches a URL server-side, and generate_pdf_document writes a new file. PDF.net states documents are encrypted at rest with AES-256 and are not used to train models.
Indefinitely, until you delete it. PDF.net's retention policy states documents are retained while the account is active and the user has not deleted them. Deleting moves a file to a recycle bin recoverable for thirty days, then permanent erasure. There is no automatic expiry for documents in an active account, and no tool sets a retention period.
It can delete them; recovery is a separate step you take yourself. delete_resource is the one destructive tool among the twenty and its name does not say whether it hits a folder or a file. PDF.net's policy puts deleted documents in a recycle bin for thirty days, so an agent's mistake is recoverable — but only from the web app, since no tool restores.
Not from the connector, on the evidence available. download_pdf_document returns a download link, and PDF.net's help article says the assistant gives you one. No tool carries a share verb, and PDF.net's separate Share PDF product, whose links open without an account, is not represented in the tool list. Whether a download link authenticates was not established.
No, not today. PDF.net states the MCP connector is available on all plans including free, and that the MCP server is free to use today. No tool name contains credit, balance, usage, quota or billing, so nothing observes spend either. The word today is PDF.net's own, so treat free as current policy rather than a commitment.
One service-wide permission and a refresh token. The server's RFC 9728 descriptor advertised pdfnet.mcp and offline_access on 2026-08-23. Neither names a verb or an object, so no boundary falls between listing your library and deleting a folder from it. Consent is all-or-nothing across every document in the account.
Twenty, according to Anthropic's directory, which is the only enumeration that exists. PDF.net's help article publishes a nine-row capability table describing what the connector does, and never prints a single tool name. So there is no second list of names to diff, and a rename between the two would be invisible.
E-signing, annotations, form filling and password protection. PDF.net's own troubleshooting section names those four as unavailable through the connector, and states the assistant will instead hand you a link to open the document in the web editor. Two tools, open_in_web_editor and open_pdf_document_manager, appear to be that handoff path.
Yes, and that is worth pricing in before approving it. list_document_tree walks the library and find_in_documents searches across documents, so an agent that can call either sees everything in the account, not only the file you named. PDF.net states the connector reaches your library only, and never another user's account.
.md twin). Source of the "free to use today" statement. · retrieved 2026-08-23llms.txt (retrieved 2026-08-23). PDF.net's robots.txt carries Content-Signal: search=yes, ai-train=yes, ai-input=yes, expressly permitting this synthesis. · retrieved 2026-08-23initialize against https://mcp.pdf.net/mcp returned HTTP 401 with a www-authenticate challenge naming scope pdfnet.mcp offline_access (2026-08-23). No tool was called. · retrieved 2026-08-23https://mcp.pdf.net/.well-known/oauth-protected-resource and authorization server metadata at https://mcp.pdf.net/.well-known/oauth-authorization-server (2026-08-23). A control request to a non-existent sibling path returned 404, so these 200s are genuine rather than a catch-all. · retrieved 2026-08-23Connect PDF.net once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.