> ## Documentation Index
> Fetch the complete documentation index at: https://docs.threetone.in/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent configuration

> Configure an agent's first message, prompt, language, models, and resource references, then review the complete saved version.

An agent configuration combines instructions, models, language settings, and references to workspace resources. Review these settings as one version so the prompt does not promise actions or information the configuration cannot provide.

<Frame caption="Recording pending: Configure an agent. Follow the written guide below.">
  <img src="https://mintcdn.com/threetonelabs-c2efc9c8/RV32i5440UZIaFHA/images/tutorials/recording-pending.webp?fit=max&auto=format&n=RV32i5440UZIaFHA&q=85&s=55f66d2eef49d8c92c9ac46f84bace35" alt="Recording pending for Configure an agent. No playable video is available yet." width="1600" height="900" className="w-full h-auto aspect-video" loading="lazy" noZoom data-path="images/tutorials/recording-pending.webp" />
</Frame>

<Note>
  The dashboard submits prompt, first-message, LLM, soft-timeout, and dynamic-variable edits through the configuration publication flow. Publication currently requires an eligible managed text-only configuration; voice and custom-model publication are blocked. Its **Voices**, **Language**, **ASR**, and **Agent behavior** pickers are previews whose changes are not saved; the Security page also displays disabled example settings. Use a complete server draft to preserve those settings without publication. See [voice and language](/agents-platform/configure/voice-and-language/overview) and [conversation flow](/agents-platform/configure/agent-behavior/conversation-flow).
</Note>

## Configuration structure

| Section | Purpose | Guide |
| - | - | - |
| `conversation_config.agent` | First message, base language, response-language policy, and prompt. | [Prompting](/agents-platform/configure/agent-behavior/prompting) |
| `conversation_config.asr` | Recognition model and language mode for voice input. | [Models and voices](/overview/models) |
| `conversation_config.tts` | Speech model and saved workspace voice. | [Models and voices](/overview/models) |
| `conversation_config.agent.prompt` | LLM selection, tool and MCP identifiers, and knowledge attachments. | [Tools](/agents-platform/configure/tools/overview) and [knowledge](/agents-platform/configure/knowledge-base/overview) |
| `platform_settings` | Platform-level configuration separate from the conversational instructions. | [Update agent schema](/api-reference/agents-platform/update-agent) |
| `workflow` | The workflow included in the configuration snapshot. | [Versions and drafts](/agents-platform/deploy/versioning) |

## Configure and review

<Steps>
  <Step title="Set the purpose and opening">
    Define the task in `agent.prompt.prompt` and write a matching `agent.first_message`. Keep the first message short enough to establish the purpose without asking several questions at once.
  </Step>

  <Step title="Choose language and models">
    Set `agent.language` and review `agent.response_language_policy`. Select an available LLM with `agent.prompt.llm`.

    For voice configurations, include ASR and TTS settings with compatible models and a saved voice. Check the catalog rather than assuming a provider supports every language and recognition mode.
  </Step>

  <Step title="Attach workspace resources">
    Review each tool, MCP server, and knowledge attachment. Use the intended identifiers and check access in the active workspace.

    Keep secrets in supported credential storage. Resource identifiers and prompt instructions do not replace authorization.
  </Step>

  <Step title="Review the entire version">
    Inspect the configuration diff, including `platform_settings` and `workflow`. Preserve unrelated fields when preparing a full draft snapshot.

    Save a complete server draft on the intended branch and read it back. Only publish when the configuration is eligible; managed text-only publication has model, resource, and workflow checks. See [versions and drafts](/agents-platform/deploy/versioning).
  </Step>
</Steps>

Paths in the steps above are relative to `conversation_config`. The generated [Create agent](/api-reference/agents-platform/create-agent), [Get agent](/api-reference/agents-platform/get-agent), and [Update agent](/api-reference/agents-platform/update-agent) pages define the request and response shapes.

## Configuration and execution

<Note>
  The documented implementation currently gates signed-token session initiation, outbound calling, traffic deployment, and automated response-test invocation. A valid configuration can be saved and reviewed without proving that those execution paths work. See [current availability](/agents-platform/overview#current-availability).
</Note>

Privacy, retention, evaluation, and related settings must be assessed separately from their enforcement. A saved field is not evidence of automatic deletion, a completed evaluation, or a compliance guarantee.

## Troubleshooting

| Problem | What to inspect |
| - | - |
| Configuration is rejected | The reported schema field, model compatibility, and workflow validation errors. |
| A referenced resource is missing | The active workspace, resource ID, and access granted to the editor or API key. |
| An edited field was lost | Whether a full snapshot omitted fields from the selected version. |
| A saved agent does not start a session | The separate execution path and its current availability. |

Start with [Create your first agent](/agents-platform/get-started/first-agent) for the dashboard workflow or [API quickstart](/agents-platform/get-started/api-quickstart) for a server-side request example.

## Frequently asked questions

<AccordionGroup>
  <Accordion title="Do resource IDs grant access to the resource?" id="faq-configuration-resource-access">
    No. Each referenced voice, tool, MCP server, or knowledge source must be available to the intended workspace and caller. Inspect access separately from the configuration reference.
  </Accordion>

  <Accordion title="How do I confirm which settings were saved?" id="faq-configuration-read-back">
    Read the intended branch or version, compare its complete configuration with the reviewed diff, and retain the version ID. A saved setting does not prove that its runtime behavior is available or enforced.
  </Accordion>
</AccordionGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.