Feature Requests

When trying to connect to the EVI ChatSocket, I'm getting this error: { "type": "error", "code": "I0100", "slug": "uncaught", "message": "Unexpected service error occurred. Please visit https://dev.hume.ai/support for assistance.", "requestId": "0155a95c-f670-44b9-8a13-d69bba6fc5945155824", "receivedAt": "2025-11-20T12:45:29.030Z" }

ws.mjs:389 WebSocket connection to 'wss: //api.hume.ai/v0/evi/chat?config_id=9e3b3048-4f27-420c-86b6-7d84642853c…experienceLevel%5D=senior&fernSdkLanguage=JavaScript&fernSdkVersion=0.15.4 ' failed:

I need to access the number that was called, which is a twilio number, as well as the originating number. The caller number is supplied in the chat_started webhook, but not the <Parameter name="to_number" value="+1xxxxxxxxxxx" /> which is posted when twilio connects to Hume. Is there a way to access that parameter? This is so the tools for a configuration and prompt parameters can be dynamically assigned.

Hi I am setting up the example here: https://github.com/HumeAI/hume-api-examples/tree/eeccd72e5a200f658354669f81cbf7b27b0a487f/evi/evi-python-phone-calling-proxy-server . It worked yesterday but today, I am getting this error: ❌ Error: 38 validation errors for union[AssistantEnd,AssistantMessage,AssistantProsody,AudioOutput,ChatMetadata,WebSocketError,UserInterruption,UserMessage,ToolCallMessage,ToolResponseMessage,ToolErrorMessage] AssistantEnd.type Input should be 'assistant_end' [type=literal_error, input_value='session_settings', input_type=str] For further information visit https://errors.pydantic.dev/2.12/v/literal_error AssistantMessage.from_text Field required [type=missing, input_value={'type': 'session_setting...es': {'name': 'Joshua'}}, input_type=dict] For further information visit https://errors.pydantic.dev/2.12/v/missing AssistantMessage.message Field required [type=missing, input_value={'type': 'session_setting...es': {'name': 'Joshua'}}, input_type=dict] For further information visit https://errors.pydantic.dev/2.12/v/missing AssistantMessage.models Field required [type=missing, input_value={'type': 'session_setting...es': {'name': ..... I don't think this should happen since this is on your official github and was working yesterday. Can someone please look into this.

I’d like to suggest adding a user-facing webhook configuration tool directly within the Speech-to-Speech (EVI) configuration editor. Currently, the API supports webhooks for events such as chat_started and chat_ended, but there’s no way to configure these directly through the dashboard UI. Implementing a simple “Add Webhook Tool” interface would be tremendously helpful for creators. Why this matters: Reliability: Webhooks tied directly to EVI lifecycle events would eliminate the need for fragile, client-side workarounds that break when the Vercel app or frontend code is updated. Ease of Use: Many creators rely on no-code tools like Make.com , Hubitat, or IFTTT. Allowing webhooks to be added from the Config page (under “Tools” or a new “Webhooks” tab) would make event-based automation accessible without requiring deployment-level edits. Consistent integration: For many use cases — like triggering lighting changes, analytics logs, or IoT events at the end of a session — the most natural and secure place to connect these automations is directly from the EVI backend, not from the app UI. This small UI addition would make EVI integrations far easier to maintain and expand, while also reducing error rates from custom code changes. Thank you for considering this — it would be a great enhancement for developers and creators who use EVI in interactive or physical environments.

Support Ticket: Hume Prompt Version Endpoint Failure Subject: Unable to create new prompt versions with /prompts/{id}/version (returns 404) ## Summary We can no longer create prompt versions via the documented endpoint POST /v0/evi/prompts/{id}/version . The call worked previously but now consistently returns 404 – No static resource prompts/{id}/version. even when the prompt was created moments earlier using the same API key. ## Environment API base: https://api.hume.ai/v0/evi Date observed: November 5–6, 2025 Issue reproduces across multiple newly created prompts ## Steps to Reproduce Create a prompt (works): ```bash curl -X POST https://api.hume.ai/v0/evi/prompts \ -H "Content-Type: application/json" \ -d '{ "name": "Test Prompt 1762399433623", "text": "<system_prompt>Initial test prompt 1762399433623</system_prompt>" }' ``` Response returns ID 2267531d-3ae7-4640-a102-65a659954174 , version 0 . Attempt to create a new version (fails): ```bash curl -X POST https://api.hume.ai/v0/evi/prompts/2267531d-3ae7-4640-a102-65a659954174/version \ -H "Content-Type: application/json" \ -d '{ "text": "<system_prompt>Updated prompt text</system_prompt>", "version_description": "Test update" }' ``` Actual response: ```json { "timestamp": "2025-11-06T03:24:00.488+00:00", "status": 404, "error": "Not Found", "message": "No static resource prompts/2267531d-3ae7-4640-a102-65a659954174/version.", "path": "/prompts/2267531d-3ae7-4640-a102-65a659954174/version" } ``` Listing prompts with GET /v0/evi/prompts still shows the prompt at version 0 ; no new version is created. Fetching GET /v0/evi/prompts/{id}/versions also returns 404. When we reuse an existing prompt name we get a 409 Conflict plus a prompt ID (example dba3da91-2c85-44ce-90c4-f919b770b703 ), but attempting to access that prompt with this API key returns 404/unauthorized. So we cannot reuse names or create versions; creating an entirely new prompt is the only option. ## Expected Behavior POST /prompts/{id}/version should create a new prompt version and return the new version details so we can iterate on the same prompt ID. ## Actual Behavior The endpoint returns 404 for every prompt ID created with the same API key. Prompt version APIs appear unusable. ## Impact We are forced to create a new prompt ID for every update, which breaks our editor workflow, complicates config updates, and loses prompt history. ## Scripts / References test-create-hume-assets.js — programmatically creates a prompt, config, and attempts the version endpoint (fails with 404). test-access-hume-prompt.js — helper to inspect prompt accessibility; duplicated prompt IDs reported by the API are not readable with the same key. ## Request Please confirm whether POST /prompts/{id}/version is still supported. If this is a regression, we’d appreciate a fix or guidance on how to update prompt text without minting new prompt IDs each time. Thanks!

Add a parameter and method to force the maximum duration message to be sent 30-60 seconds before the duration has been reached. Currently (as of Oct 2025), Hume sends this message less than 10 seconds before the maximum duration has been reached. That's not sufficient time for a user to wrap up a conversation.

raw_message:📨 Raw message from HumeAI: {"type":"chat_metadata","chatGroupId":"4df7d513-d93a-4b92-89ac-64712cd8e670","chatId":"92ec8037-4bad-4e55-a858-1ad04b49dc28","requestId":"97e3c2eb-5558-4cbf-85e6-529bb50b08543309936","receivedAt":"2025-10-25T09:53:45.207Z"} raw_message:📨 Raw message from HumeAI: {"type":"error","code":"I0100","slug":"uncaught","message":"Unexpected service error occurred. Please visit https://dev.hume.ai/support for assistance.","requestId":"97e3c2eb-5558-4cbf-85e6-529bb50b08543309936","receivedAt":"2025-10-25T09:53:46.496Z"} connection:🔌 HumeAI WebSocket connection closed Getting the above error and not able to make a socket succession.

I'm building a consumer app with EVI where users need time to reflect and recall (often around 5 seconds of silence while thinking) - think therapy. I see users facing a challenges currently with EVI3, Quick interruptions User pauses to think → EVI generates response too quickly (400-600ms after last word) Need: Allow longer natural pauses without triggering response generation Could you please add new dashboard settings or API configs to: Increase turn-taking delay/sensitivity? Adjust VAD (Voice Activity Detection) to allow longer thinking pauses? Configure "end-of-turn" detection threshold?

It would be great to integrate EVI with an avatar and have realistic lipsync. However, to do this we would need to have the same Word or Phoneme timestamps that are provided in your TTS functionality - https://dev.hume.ai/docs/text-to-speech-tts/timestamps