Integration

AnythingLLM integration

AnythingLLM is a self-hosted document workspace that turns files into a searchable knowledge base and answers questions over it. The API lets an agent upload files, links and raw text as documents, browse the document folders, read a document and its metadata schema, create and remove folders and move files between them, create workspaces, pin or embed documents into a workspace, ask a workspace or one of its threads a question in chat or query mode and read the sources it answered from, run a raw vector search over a workspace, read and delete chat history, manage threads, manage users, invites and workspace membership, publish and inspect embeddable chat widgets, and read instance settings and vector counts. The host is part of the credential because every install is its own deployment. Authentication is an API key generated in the AnythingLLM settings under Tools, Developer API, sent as a bearer token.

Connect AnythingLLM How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in AnythingLLM.Built from the 43 AnythingLLM tools below, on the connection you add once.

What you can do with AnythingLLM

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Chat, 6 operations

chat_with_threadChat with threadPOSTChat with a workspace thread.
chat_with_workspaceChat with workspacePOSTExecute a chat with a workspace.
get_embed_session_chatsGet embed session chatsGETGet chats for a specific embed and session.
list_embed_chatsList embed chatsGETGet all chats for a specific embed.
list_thread_chatsList thread chatsGETGet chats for a workspace thread.
list_workspace_chatsList workspace chatsGETGet a workspaces chats regardless of user by its unique slug.

User, 5 operations

delete_userDelete userDELETEDelete existing user by id. Methods are disabled until multi user mode is enabled via the UI.
list_admin_usersList admin usersGETCheck to see if the instance is in multi-user-mode first. Methods are disabled until multi user mode is enabled via the UI.
list_usersList usersGETList all users.
list_workspace_usersList workspace usersGETRetrieve a list of users with permissions to access the specified workspace.
update_userUpdate userPOSTUpdate existing user settings. Methods are disabled until multi user mode is enabled via the UI.

Embed, 3 operations

delete_embedDelete embedDELETEDelete an existing embed configuration.
list_embedsList embedsGETList all active embeds.
update_embedUpdate embedPOSTUpdate an existing embed configuration.

Workspace, 3 operations

delete_workspaceDelete workspaceDELETEDeletes a workspace by its slug.
get_workspaceGet workspaceGETGet a workspace by its unique slug.
list_workspacesList workspacesGETList all current workspaces.

Document, 2 operations

get_documentGet documentGETGet a single document by its unique AnythingLLM document name.
list_documentsList documentsGETList of all locally-stored documents in instance. Optionally, pass ?folder=name to fetch the contents of a single folder, paginated with offset and limit (limit=all returns every.

Invite, 2 operations

delete_inviteDelete inviteDELETEDeactivates (soft-delete) invite by id. Methods are disabled until multi user mode is enabled via the UI.
list_invitesList invitesGETList all existing invitations to instance regardless of status. Methods are disabled until multi user mode is enabled via the UI.

Update, 2 operations

update_threadUpdate threadPOSTUpdate thread name by its unique slug.
update_workspaceUpdate workspacePOSTUpdate workspace settings by its unique slug.

Upload, 2 operations

upload_documentUpload documentPOSTUpload a new file to AnythingLLM to be parsed and prepared for embedding, with optional metadata.
upload_document_to_folderUpload document to folderPOSTUpload a new file to a specific folder in AnythingLLM to be parsed and prepared for embedding. If the folder does not exist, it will be created.

Accepted file type, 1 operation

list_accepted_file_typesList accepted file typesGETCheck available filetypes and MIMEs that can be uploaded.

Auth, 1 operation

verify_api_keyVerify api keyGETVerify the attached Authentication header contains a valid API token.

Export chat, 1 operation

export_chatsExport chatsGETExport all of the chats from the system in a known format. Output depends on the type sent. Will be send with the correct header for the output.

Folder, 1 operation

get_document_folderGet document folderGETGet all documents stored in a specific folder. Returns every document by default; pass offset and limit to paginate.

Generated file, 1 operation

get_generated_fileGet generated fileGETDownload a file generated by an agent skill (e.g., PDF, DOCX, XLSX, PPTX). The filename is returned in the `outputs` array of a chat response when an agent generates a file. Use.

Is multi user mode, 1 operation

get_multi_user_modeGet multi user modeGETCheck to see if the instance is in multi-user-mode first. Methods are disabled until multi user mode is enabled via the UI.

Manage user, 1 operation

manage_workspace_usersManage workspace usersPOSTSet workspace permissions to be accessible by the given user ids and admins. Methods are disabled until multi user mode is enabled via the UI.

Metadata schema, 1 operation

get_metadata_schemaGet metadata schemaGETGet the known available metadata schema for when doing a raw-text upload and the acceptable type of value for each key.

New, 1 operation

create_threadCreate threadPOSTCreate a new workspace thread.

Remove document, 1 operation

delete_documentsDelete documentsDELETEPermanently remove documents from the system.

Remove folder, 1 operation

remove_document_folderRemove document folderDELETERemove a folder and all its contents from the documents storage directory.

System, 1 operation

get_system_settingsGet system settingsGETGet all current system settings that are defined.

Thread, 1 operation

delete_threadDelete threadDELETEDelete a workspace thread.

Update embedding, 1 operation

update_workspace_embeddingsUpdate workspace embeddingsPOSTAdd or remove documents from a workspace by its unique slug.

Update pin, 1 operation

update_workspace_pinUpdate workspace pinPOSTAdd or remove pin from a document in a workspace by its unique slug.

Update user, 1 operation

update_workspace_usersUpdate workspace usersPOSTOverwrite workspace permissions to only be accessible by the given user ids and admins. Methods are disabled until multi user mode is enabled via the UI.

Vector count, 1 operation

get_vector_countGet vector countGETNumber of all vectors in connected vector database.

Vector search, 1 operation

search_workspace_vectorsSearch workspace vectorsPOSTPerform a vector similarity search in a workspace.

Use AnythingLLM from Claude Code, Cursor or Codex

Connect AnythingLLM in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its AnythingLLM operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect AnythingLLM

Create your workspace
Sign up free. Nothing to install.
Connect AnythingLLM once
Add your AnythingLLM connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.