Skip to main content
UserChatSettings is a TypeScript type in the Relevance AI Node.js SDK that lets you configure the chat interface when starting or managing an agent conversation programmatically. It controls which mounts (built-in capabilities like file upload or code execution) and agent skills (other agents callable from within the conversation) are enabled or disabled.

​
Concepts

Mounts are named capabilities that can be attached to a chat session. Each mount maps to a feature the agent can use during a conversation — for example, accessing files, running code, or calling external services. Disabling a mount prevents the agent from using that capability in a given session. Agent skills are agents that have been configured as callable skills from within another agent’s conversation. Each skill is identified by its agent ID. You can control whether a given skill is available to the agent in a specific session.

​
API structure

UserChatSettings takes an object where keys are mount names or agent IDs. Each value is an object with a single boolean field. 
type UserChatSettings = {
  // Mount configuration: mountName -> { is_enabled: boolean }
  [mountName: string]: { is_enabled: boolean };
} | {
  // Agent skill configuration: agentId -> { is_skill: boolean }
  [agentId: string]: { is_skill: boolean };
}

​
Mount format

{
  [mountName: string]: {
    is_enabled: boolean; // true = mount is active, false = mount is disabled
  }
}

​
Agent skill format

{
  [agentId: string]: {
    is_skill: boolean; // true = agent is available as a skill, false = agent is not available
  }
}

​
Usage examples

​
Enable and disable mounts

import { RelevanceAI } from "@relevanceai/chain";

const client = new RelevanceAI({
  apiKey: process.env.RELEVANCE_API_KEY,
  projectId: process.env.RELEVANCE_PROJECT_ID,
  region: process.env.RELEVANCE_REGION,
});

// Enable a specific mount, disable another
const chatSettings = {
  file_upload: { is_enabled: true },
  code_execution: { is_enabled: false },
};

await client.agents.trigger({
  agentId: "your-agent-id",
  message: { role: "user", content: "Hello" },
  userChatSettings: chatSettings,
});

​
Control agent skills

import { RelevanceAI } from "@relevanceai/chain";

const client = new RelevanceAI({
  apiKey: process.env.RELEVANCE_API_KEY,
  projectId: process.env.RELEVANCE_PROJECT_ID,
  region: process.env.RELEVANCE_REGION,
});

// Make a specific agent available as a skill, disable another
const chatSettings = {
  "agent-id-for-research": { is_skill: true },
  "agent-id-for-outreach": { is_skill: false },
};

await client.agents.trigger({
  agentId: "your-agent-id",
  message: { role: "user", content: "Start the research workflow" },
  userChatSettings: chatSettings,
});

​
Combine mounts and agent skills

const chatSettings = {
  // Mounts
  file_upload: { is_enabled: true },
  web_search: { is_enabled: true },
  code_execution: { is_enabled: false },

  // Agent skills
  "a1b2c3d4-research-agent": { is_skill: true },
  "e5f6g7h8-outreach-agent": { is_skill: false },
};

​
Deprecated format

The following format is deprecated. It still works but will be removed in a future release. Migrate to the new format above.
The old format used two separate list fields: 
// Deprecated — do not use in new code
{
  disabled_mounts: string[];  // list of mount names to disable
  disabled_skills: string[];  // list of agent IDs to disable
}
Why the new format is preferred: The object format makes each entry’s intent explicit — you can see at a glance whether a mount is enabled or disabled without inferring from inclusion in a list. It also removes the ambiguity around what the default state is when a mount or skill is not listed.

​
Migrating from the old format

Old code using disabled_mounts:
// Old format (deprecated)
const chatSettings = {
  disabled_mounts: ["code_execution", "file_upload"],
};
Equivalent new format:
// New format
const chatSettings = {
  code_execution: { is_enabled: false },
  file_upload: { is_enabled: false },
};
Mounts not listed in the new format retain their default state. If you previously relied on the absence from disabled_mounts to mean “enabled”, explicitly set is_enabled: true for those entries to be unambiguous.
Old code using disabled_skills:
// Old format (deprecated)
const chatSettings = {
  disabled_skills: ["a1b2c3d4-research-agent"],
};
Equivalent new format:
// New format
const chatSettings = {
  "a1b2c3d4-research-agent": { is_skill: false },
};
Old code with both fields:
// Old format (deprecated)
const chatSettings = {
  disabled_mounts: ["code_execution"],
  disabled_skills: ["a1b2c3d4-research-agent", "e5f6g7h8-outreach-agent"],
};
Equivalent new format:
// New format
const chatSettings = {
  code_execution: { is_enabled: false },
  "a1b2c3d4-research-agent": { is_skill: false },
  "e5f6g7h8-outreach-agent": { is_skill: false },
};

SDK & API triggers

Trigger agents programmatically using the SDK or HTTP API

Conversation API

Create and continue conversations with agents via the API