Integration

Dust integration

Dust is a platform for company AI agents. Talk to a workspace's agents (start a conversation, post messages that mention an agent and read its answer, attach content, validate tool actions, leave feedback), manage agent configurations and skills, list spaces, apps and tools, run legacy Dust apps, and manage the knowledge agents read: data sources and their views, documents (upsert, read, delete, semantic search) and structured tables and rows. Not exposed: the server-sent event streams, the file upload flow, skill import from files, the client-side MCP server routes, the analytics file exports, the upsert-queue check (system keys only) and the inbound trigger webhook. Authentication is a Dust API key (created by a workspace admin under Admin, Developers, API keys) sent as a Bearer token; the connection also holds the workspace id (wId) and the region domain of the workspace: dust.tt for US workspaces, eu.dust.tt for European ones.

Connect Dust How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Dust.Built from the 53 Dust tools below, on the connection you add once.

What you can do with Dust

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.

Agent configuration, 4 operations

archive_agent_configurationArchive agent configurationDELETEArchives (soft-deletes) an agent configuration and disables its triggers.
get_agent_configurationGet agent configurationGETReturns one agent configuration; variant full also includes its actions (tools).
list_agent_configurationsList agent configurationsGETLists the workspace's agents (configurations) visible to the key, with sId, name, description, instructions and model.
update_agent_configurationUpdate agent configurationPATCHUpdates an agent configuration: name and settings in agent, instructions, generation settings, tags, editors, skills or toolset. Fields left out are kept.

Data source view, 4 operations

delete_data_source_viewDelete data source viewDELETEDeletes a data source view, removing that data source from the space.
get_data_source_viewGet data source viewGETReturns a data source view: the data source it exposes in a space, its category, kind and the parent nodes it is restricted to (parentsIn).
list_data_source_viewsList data source viewsGETLists the data source views of a space: the data sources exposed there and the nodes each view is restricted to.
update_data_source_viewUpdate data source viewPATCHChanges which nodes of the data source a view exposes: set parentsIn to replace the list, or use parentsToAdd and parentsToRemove. Returns the view.

Document, 4 operations

delete_documentDelete documentDELETEDeletes a document from a data source.
get_documentGet documentGETReturns a document of a data source with its text, chunks, tags, parents and source URL.
list_documentsList documentsGETLists the documents of a data source, paginated with limit and offset, or fetches given document_ids.
upsert_documentUpsert documentPOSTInserts or replaces a document in a data source: text (or a nested section), title, MIME type, source URL, tags and timestamp. Agents then retrieve it.

Row, 4 operations

delete_rowDelete rowDELETEDeletes one row of a table.
get_rowGet rowGETReturns one row of a table: its row_id and its values keyed by column name.
list_rowsList rowsGETLists the rows of a table in a data source, paginated with limit and offset. Each row has its row_id and its values keyed by column name.
upsert_rowsUpsert rowsPOSTInserts or replaces rows of a table, matched by row_id; with truncate true every existing row is deleted first. The table's columns are inferred from the values.

Search, 4 operations

search_agent_configurationsSearch agent configurationsGETFinds agents whose name contains the query, case-insensitively, among global and published agents the key can access.
search_data_sourceSearch data sourceGETSemantic search over a data source's documents: returns the best matching documents with their scored chunks.
search_data_source_viewSearch data source viewGETSemantic search over the documents a data source view exposes: returns the best matching documents with their scored chunks.
search_nodesSearch nodesPOSTSearches the titles of documents, tables and folders across the workspace's data sources (not their content: use search_data_source for that). Returns matching nodes.

Table, 4 operations

delete_tableDelete tableDELETEDeletes a table and its rows from a data source.
get_tableGet tableGETReturns a table of a data source: name, title, description, column schema, tags, parents and timestamp.
list_tablesList tablesGETLists the tables of a data source with their name, description, column schema, tags and parents.
upsert_tableUpsert tablePOSTCreates or updates a table in a data source (matched by table_id); its rows are added with upsert_rows. Returns the table.

Conversation, 3 operations

create_conversationCreate conversationPOSTStarts a conversation with a first message; mention an agent in message.mentions to make it answer. With blocking true the call waits and the answer is in conversation.content.
get_conversationGet conversationGETReturns a conversation with its messages in order: user messages, agent messages (with their content, status and actions) and content fragments. Use it to read an agent's answer.
update_conversationUpdate conversationPATCHRenames a conversation (title) or marks it read or unread (read).

Feedback, 3 operations

create_message_feedbackCreate message feedbackPOSTLeaves thumbs up or down feedback, with optional text, on an agent message.
delete_message_feedbackDelete message feedbackDELETERemoves the caller's feedback from an agent message.
list_conversation_feedbacksList conversation feedbacksGETLists the thumbs up and down feedback left on a conversation's agent messages.

Run, 2 operations

get_app_runGet app runGETReturns a Dust app run: status, results per block and traces.
run_appRun appPOSTRuns a legacy Dust app with inputs; with blocking true the call waits for the run to finish and returns its results.

Skill, 2 operations

archive_skillArchive skillDELETEArchives (soft-deletes) a custom skill.
list_skillsList skillsGETLists the workspace's custom skills (active by default) with name, description and availability.

Suggestion, 2 operations

suggest_conversation_mentionsSuggest conversation mentionsGETGet suggestions for mentions (agents and users) based on a query string, scoped to a specific conversation.
suggest_mentionsSuggest mentionsGETGet suggestions for mentions (agents and users) based on a query string.

Trigger, 2 operations

get_triggerGet triggerGETReturns one agent trigger (scheduled run or webhook). Requires a workspace admin API key.
list_triggersList triggersGETLists agent triggers (scheduled runs and webhooks) across the workspace. Requires a workspace admin API key.

Answer question, 1 operation

answer_message_questionAnswer message questionPOSTAnswers a question an agent asked through a tool action in a message, so the agent can continue.

App, 1 operation

list_appsList appsGETLists the legacy Dust apps of a space with their sId, name and specification hash.

Cancel, 1 operation

cancel_message_generationCancel message generationPOSTStops agents still generating the given messages in a conversation.

Consumption, 1 operation

get_conversation_consumptionGet conversation consumptionGETReturns the credits a conversation billed, broken down by agent, model and tool, for finished agent messages.

Content fragment, 1 operation

create_content_fragmentCreate content fragmentPOSTAttaches content to a conversation for its agents to read: inline text (content plus contentType), a previously uploaded file (fileId) or a data source node (nodeId plus nodeDataSourceViewId).

Data source, 1 operation

list_data_sourcesList data sourcesGETLists the data sources of a space (connections, folders, websites) with their sId, name and connector.

Edit, 1 operation

edit_messageEdit messagePOSTReplaces the content and mentions of a user message; the mentioned agents answer the new version.

File, 1 operation

download_conversation_fileDownload conversation fileGETDownloads a file an agent generated in a conversation, by the filePath its message action reported in generatedFiles.

Import, 1 operation

import_agent_configurationImport agent configurationPOSTCreates an agent from a full configuration: agent settings, instructions, generation settings, tags, editors and toolset.

Mcp server view, 1 operation

list_mcp_server_viewsList mcp server viewsGETLists the tools (MCP server views) enabled in a space, with each server's name, description and tools.

Message, 1 operation

create_messageCreate messagePOSTPosts a user message in a conversation; mention an agent to make it answer. Returns the message at once: read the agent's answer later with get_conversation.

Parent, 1 operation

update_document_parentsUpdate document parentsPOSTSets a document's place in the data source hierarchy: its direct parent_id and the full parents list (the document id first, then its ancestors).

Parse, 1 operation

parse_mentionsParse mentionsPOSTConverts @name mentions in pasted markdown into Dust's mention format, matching agent and user names.

Space, 1 operation

list_spacesList spacesGETLists the spaces the key can access (company, open and restricted spaces) with sId, name and kind.

Validate action, 1 operation

validate_message_actionValidate message actionPOSTApproves or rejects a tool action an agent is waiting on in a message (blocked actions need a person's validation unless tools validation was skipped).

Use Dust from Claude Code, Cursor or Codex

Connect Dust in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Dust 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 Dust

Create your workspace
Sign up free. Nothing to install.
Connect Dust once
Add your Dust 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.