SignalWire AI Webhooks API

openapi: 3.2.0
info:
  title: SignalWire REST AI Webhooks API
  version: 1.0.0
  contact:
    name: SignalWire
    url: https://support.signalwire.com/portal/en/newticket?departmentId=1029313000000006907&layoutId=1029313000000074011
    email: support@signalwire.com
  license:
    name: MIT
    url: https://github.com/signalwire/docs/blob/main/LICENSE
  termsOfService: https://signalwire.com/legal/signalwire-cloud-agreement
  description: Callbacks an AI agent sends to your server. The same payloads apply to every surface an agent runs on — voice calls, Amazon Bedrock agents, sidecar agents, and text conversations.
servers:
- url: https://{space_name}.signalwire.com
  description: SignalWire API
  variables:
    space_name:
      default: '{Your_Space_Name}'
      description: Your SignalWire Space name
security:
- SignalWireBasicAuth: []
tags:
- name: AI Webhooks
  description: Callbacks an AI agent sends to your server. The same payloads apply to every surface an agent runs on — voice calls, Amazon Bedrock agents, sidecar agents, and text conversations.
  externalDocs:
    url: https://signalwire.com/docs/apis
    description: Developer documentation on AI webhooks
paths: {}
webhooks:
  aiDebugWebhook:
    post:
      operationId: ai_debug_webhook
      summary: AI debug webhook
      description: 'A diagnostic feed for a call that is still in progress. Set `debug_webhook_url` on your agent and

        every step it takes is posted to that URL as it happens: speech recognized, model called, tool

        invoked, context switched, error hit. Use it to work out why a call went the way it did — which

        tool the agent reached for, what came back, where a turn went wrong — or to react while the call is

        still live, such as paging a supervisor.


        Each request carries `call_info` plus one or more event properties, where the property name is the

        event. One moment can produce several: evaluating a `data_map` webhook sends `webhook`, `input`,

        `output`, `error_keys`, and `match` together. Handle the properties you recognize and ignore the

        rest, since the set grows over time.


        Setting the URL is what enables the feed. `debug_webhook_level` only widens it: at `2` it also

        carries `conversation_add`, `llm_request`, and `llm_response`, which fire on every turn and every

        model call.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.AiDebugWebhookPayload'
      security:
      - {}
  aiPostPromptCallback:
    post:
      operationId: ai_post_prompt_callback
      summary: AI post-prompt callback
      description: 'Sent to your `ai.post_prompt_url` when the AI session ends. It

        carries the agent''s answer to your `post_prompt`

        alongside the full record of the call: the conversation, the tool calls, the timings, and the

        token counts. This is the one report you get per call, so store the body verbatim and extract only

        the fields you query. Nothing you return in the response is read.


        Read `action` first. It is `post_conversation` on the end-of-call report described here. The same

        URL also receives `fetch_conversation` when the agent starts with a stored conversation

        (`save_conversation` with a

        `conversation_id`), asking your endpoint to return that conversation; that request carries the

        call and session fields but none of the summary fields. Answer it with the stored

        `conversation_summary`.


        The conversation appears three times. `call_log` is the filtered view, with interrupted segments

        consolidated. `raw_call_log` is unfiltered and append-only, and is the only place barge-in detail

        survives. `call_timeline` is a flat stream of typed events aligned to `raw_call_log`.


        `amazon_bedrock` agents send a different report.

        Write your handler against the

        Bedrock post-prompt callback instead.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.AiPostPromptCallbackPayload'
      security:
      - {}
  aiSidecarCallback:
    post:
      operationId: ai_sidecar_callback
      summary: AI sidecar callback
      description: 'Sent to the sidecar''s `url` as an HTTP `POST` whenever you set one. The same event is always

        published in real time on the SignalWire Relay event channel (`calling.ai.sidecar`), so the

        webhook is optional. Each event is wrapped under `sidecar_event` — read that before checking its

        `type` and fields.


        This payload covers the envelope shared by every callback. For the fields specific to each `type`

        (such as `insight.raw`, `turn.transcript_delta`, or `final.summary`), see the

        SWML ai_sidecar reference.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.AISidecarCallbackPayload'
      security:
      - {}
  aiSidecarSwaigToolWebhook:
    post:
      operationId: ai_sidecar_swaig_tool_webhook
      summary: AI sidecar SWAIG tool webhook
      description: 'Sent to a sidecar tool''s `web_hook_url` (or the SWAIG `defaults.web_hook_url`) when the sidecar

        calls one of your functions.


        Your endpoint runs the function and replies with a JSON object. Both fields are optional, so `{}`

        is a valid reply:


        - `response` — the result the model reads next. A plain string only; the

        `{tool_result, tool_prompt}` object form that `ai` agents

        accept is not read. Omit it and the model reads a default result.

        - `action` — a single action object or an array of them. See

        Supported SWAIG actions for

        what you can return.


        There is no `post_process` on a sidecar reply.


        The sidecar only listens to the call and never speaks on it, so a `say` action is reported back to

        you as a callback rather than being spoken aloud.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.AISidecarSwaigToolWebhookPayload'
      security:
      - {}
  aiSwaigToolWebhook:
    post:
      operationId: ai_swaig_tool_webhook
      summary: AI SWAIG tool webhook
      description: 'Sent to a tool''s `web_hook_url` (or the SWAIG `defaults.web_hook_url`) when an

        `ai` agent calls one of your functions.


        Your endpoint runs the function and replies with a JSON object — the same shape a `data_map`

        output produces, because the platform reads both the same way. Every field is optional, so `{}` is

        a valid reply and a handler that only steers the call can return `action` alone:


        - `response` — the result the agent reads next, written to the AI rather than spoken to the

        caller. A plain string becomes the tool message as-is; the object form

        `{tool_result, tool_prompt}` splits the data half from the steering half. Omit it and the agent

        reads a default result.

        - `action` — a single action object or an array of them, executed on the live call.

        - `post_process` — hold the actions until after the agent has spoken. **Default:** `false`.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.AiSwaigToolWebhookPayload'
      security:
      - {}
  bedrockPostPromptCallback:
    post:
      operationId: bedrock_post_prompt_callback
      summary: Amazon Bedrock post-prompt callback
      description: 'Sent to your `amazon_bedrock.post_prompt_url` when

        the agent''s session ends, carrying its answer to your `post_prompt` alongside the record of the

        call. Nothing you return in the response is read.


        Bedrock agents send a different report from `ai` agents. There

        is no `call_timeline`, `previous_contexts`, `hard_timeout`, `call_ended_by`, or `ai_id_tag`;

        `raw_call_log` is a copy of `call_log` rather than a separate unfiltered view; `swaig_log` is

        always empty; `conversation_summary` is always present; and the `total_*` fields arrive without

        your having to enable accounting. `post_prompt_url` also takes no separate credentials here — put

        them in the URL as `username:password@url`. Write your handler against this payload, not the

        AI post-prompt callback.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.BedrockPostPromptCallbackPayload'
      security:
      - {}
  bedrockSwaigToolWebhook:
    post:
      operationId: bedrock_swaig_tool_webhook
      summary: Amazon Bedrock SWAIG tool webhook
      description: 'Sent to a tool''s `web_hook_url` (or the SWAIG `defaults.web_hook_url`) when an

        `amazon_bedrock` agent calls one of your functions.

        Your endpoint runs the function and replies with a JSON object. Both fields are optional, so `{}`

        is a valid reply:


        - `response` — the result the agent reads next. A plain string only here; the

        `{tool_result, tool_prompt}` object form that `ai` agents

        accept is not read. Omit it and the agent reads a default result.

        - `action` — a single action object or an array of them, executed on the live call.


        There is no `post_process`: a Bedrock agent always defers actions until after it has spoken.


        Bedrock agents send a different payload from `ai` agents. Notably

        `content_type` is `text/json` rather than `text/swaig`, `argument` carries no `substituted` value,

        there is no `version`, `description`, or `argument_desc`, and the call''s timing and caller fields

        are named differently. Write your handler against this payload, not the

        AI SWAIG tool webhook.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.BedrockSwaigToolWebhookPayload'
      security:
      - {}
  swaigSignatureRequest:
    post:
      operationId: swaig_signature_request
      summary: SWAIG function signature request
      description: 'Sent once per `SWAIG.includes` entry when an AI

        agent loads, to discover the functions your server hosts. Every way of building an agent sends it —

        SWML you write yourself, SWML a Server SDK generates, or an agent you configure in your Dashboard —

        because they all resolve `includes` the same way.


        Reply with the function definitions you host, each shaped like an entry in

        `SWAIG.functions` — `function`,

        `description`, and `parameters`. Three forms are accepted:


        - an array of definitions, the usual one;

        - a single definition on its own;

        - an object `{functions, defaults}`, where `defaults` sets SWAIG defaults across the definitions

        it carries — useful when they share a `web_hook_url` or auth.


        Every definition you return is registered, whether or not it was named in the request''s

        `functions` list.


        This is not the payload a function call sends. It goes to the `includes` entry''s `url`, using

        `auth_user` and `auth_password` when set. Your endpoint can also receive it outside of a call, as a

        check that it answers, so answer it the same way. When your project has a signing key, the request

        carries an `X-SignalWire-Signature` header you can verify.'
      parameters: []
      responses:
        '200':
          description: Webhook received
      tags:
      - AI Webhooks
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/Webhooks.AI.SwaigSignatureRequestPayload'
      security:
      - {}
components:
  schemas:
    Webhooks.AI.AISidecarCallbackPayload:
      type: object
      required:
      - call_info
      - sidecar_event
      properties:
        call_info:
          allOf:
          - $ref: '#/components/schemas/Webhooks.AI.AICallInfo'
          description: Envelope describing the call. `project_id` and `space_id` are included when available.
        sidecar_event:
          allOf:
          - $ref: '#/components/schemas/Webhooks.AI.AISidecarEvent'
          description: The sidecar callback.
      unevaluatedProperties:
        not: {}
    Webhooks.AI.AISidecarSwaigToolWebhookPayload:
      type: object
      required:
      - function
      - argument
      - channel_data
      properties:
        function:
          type: string
          description: The name of the function the model is calling.
          examples:
          - lookup_competitor
        argument:
          type: object
          properties:
            parsed:
              type: array
              items:
                type: object
                unevaluatedProperties: {}
              description: The arguments parsed into objects. Usually a single-element array.
              examples:
              - - competitor: ACME
            raw:
              type: string
              description: The raw argument string, exactly as the model produced it.
              examples:
              - '{"competitor":"ACME"}'
            substituted:
              type: string
              description: 'Any text that surrounded the JSON, with the JSON itself removed. Omitted when the whole

                argument was JSON, which is the usual case.'
              examples:
              - ''
          required:
          - parsed
          - raw
          unevaluatedProperties:
            not: {}
          description: The arguments the model passed to your function.
        call_id:
          type: string
          description: The ID of the call the sidecar is attached to.
          examples:
          - 2e1e66e5-5d07-413d-9668-55542992eec0
        global_data:
          type: object
          unevaluatedProperties: {}
          description: The sidecar's current `global_data`. Present when the sidecar has any.
        channel_data:
          allOf:
          - $ref: '#/components/schemas/Webhooks.AI.AISidecarChannelData'
          description: Identifies the call the sidecar is attached to.
      unevaluatedProperties:
        not: {}
    Webhooks.AI.AiSwaigToolWebhookPayload:
      type: object
      required:
      - function
      - argument
      - argument_desc
      - description
      - call_id
      - ai_session_id
      - app_name
      - meta_data_token
      - meta_data
      - channel_active
      - channel_offhook
      - channel_ready
      - content_type
      - version
      - content_disposition
      - conversation_type
      properties:
        function:
          type: string
          description: The name of the function the AI is calling.
          examples:
          - get_weather
        argument:
          type: object
          properties:
            parsed:
              type: array
              items:
                type: object
                unevaluatedProperties: {}
              description: The arguments parsed into objects. Usually a single-element array.
              examples:
              - - city: San Francisco
            raw:
              type: string
              description: The raw argument string, exactly as the AI produced it.
              examples:
              - '{"city":"San Francisco"}'
            substituted:
              type: string
              description: 'Any text that surrounded the JSON, with the JSON itself removed. Omitted when the whole

                argument was JSON, which is the usual case.'
              examples:
              - ''
          required:
          - parsed
          - raw
          unevaluatedProperties:
            not: {}
          description: The arguments the AI passed to your function.
        argument_desc:
          type: object
          unevaluatedProperties: {}
          description: The function's parameter definition, as you declared it in `parameters`.
          examples:
          - type: object
            properties:
              city:
                type: string
                description: Name of the city
            required:
            - city
        description:
          type: string
          description: 'The description you gave the function in

            [`SWAIG.functions`](/docs/swml/reference/calling/ai/swaig/functions#properties).'
          examples:
          - Look up the current weather for a city.
        call_id:
          type: string
          description: The ID of the call. On a chat conversation this carries the conversation id instead.
          examples:
          - 2e1e66e5-5d07-413d-9668-55542992eec0
        ai_session_id:
          type: string
          description: The ID of the AI session on the call.
          examples:
          - a0d4e6e5-5d07-413d-9668-55542992eec0
        conversation_id:
          type: string
          description: 'The conversation ID, when the AI session has one. A text conversation held over the

            [AI chat endpoint](/docs/apis/rest/ai-chat/chat-methods) always has one, and it matches

            `call_id`.'
        app_name:
          type: string
          description: The name of your AI application.
          examples:
          - ai
        global_data:
          type: object
          unevaluatedProperties: {}
          description: The AI session's current `global_data`, when it has any.
          examples:
          - customer_tier: premium
            pickup_address: 123 Main St, Springfield
        meta_data_token:
          type: string
          description: 'The token that scopes `meta_data`. This is the `meta_data_token` you set on the function, or a

            value derived from the function''s `web_hook_url` and credentials when you did not set one.'
          examples:
          - my-token
        meta_data:
          type: object
          unevaluatedProperties: {}
          description: Metadata scoped to `meta_data_token`. An empty object when the function has none yet.
          examples:
          - order_number: '12345'
        caller_id_name:
          type: string
          description: The caller's name, when available.
          examples:
          - Jane Doe
        caller_id_num:
          type: string
          description: The caller's number, when available.
          examples:
          - '+15555550100'
        channel_active:
          type: boolean
          description: Whether the call is still up.
          examples:
          - true
        channel_offhook:
          type: boolean
          description: Whether the call is answered.
          examples:
          - true
        channel_ready:
          type: boolean
          description: Whether the AI session is ready to take actions.
          examples:
          - true
        content_type:
          type: string
          description: The content type of the request body. Always `text/swaig`.
          examples:
          - text/swaig
        version:
          type: string
          description: The SWAIG protocol version.
          examples:
          - '2.0'
        content_disposition:
          type: string
          description: How the body is delivered. Always `SWAIG Function`.
          examples:
          - SWAIG Function
        conversation_type:
          type: string
          enum:
          - voice
          - chat
          description: 'The kind of conversation the agent is running: a call, or a text conversation held over the

            [AI chat endpoint](/docs/apis/rest/ai-chat/chat-methods).'
          examples:
          - voice
        project_id:
          type: string
          description: Your project ID, when available.
          examples:
          - 4d0d6f16-5881-4fcc-92a4-02c51a91954d
        space_id:
          type: string
          description: Your Space ID, when available.
          examples:
          - 451ed9ff-e568-4222-8af9-4f9ab7428d09
        fatal_error:
          type: boolean
          description: '`true` when the AI session has hit an unrecoverable error. Included only in that case.'
        error_reason:
          type: string
          description: A description of the error. Included only when `fatal_error` is set.
        SWMLVars:
          type: object
          unevaluatedProperties: {}
          description: SWML variables for the call. Included when you enable `swaig_post_swml_vars`.
        SWMLCall:
          type: object
          unevaluatedProperties: {}
          description: SWML call state. Included when you enable `swaig_post_swml_vars`.
        call_log:
          type: array
          items:
            $ref: '#/components/schemas/Webhooks.AI.AICallLogEntry'
          description: The conversation so far, with sensitive values redacted. Included when you enable `swaig_post_conversation`.
        raw_call_log:
          type: array
          items:
            $ref: '#/components/schemas/Webhooks.AI.AICallLogEntry'
          description: The full, unredacted conversation so far. Included when you enable `swaig_post_conversation`.
      unevaluatedProperties:
        not: {}
    Webhooks.AI.AISwaigLogEntry:
      type: object
      required:
      - command_name
      - command_arg
      - epoch_time
      properties:
        command_name:
          type: string
          description: The name of the function the agent called.
          examples:
          - get_weather
        command_arg:
          type: string
          description: The arguments it passed, as the raw string the model produced.
          examples:
          - '{"city":"San Francisco"}'
        epoch_time:
          type: integer
          description: When the call was made, as a Unix timestamp in seconds.
          examples:
          - 1694541334
        native:
          type: boolean
          description: Always `true` when present, meaning the function ran inside the platform rather than on your server.
          examples:
          - true
        active_count:
          anyOf:
          - type: integer
          - type: string
          description: 'How many times the function may still be called, or `endless` when it has no limit. Present only

            on a function you limited.'
          examples:
          - 2
        url:
          type: string
          description: The URL the function was served from. Present when the call reached your server.
          examples:
          - https://example.com/tools/get_weather
        post_data:
          type: object
          unevaluatedProperties: {}
          description: The body sent to your server. Present when the call reached it.
        post_response:
          type: object
          unevaluatedProperties: {}
          description: The response your server returned. Present when the call reached it and got a reply.
        delayed_post_response:
          type: object
          unevaluatedProperties: {}
          description: A response that arrived after the agent had already moved on. Present only when that happened.
        mcp_url:
          type: string
          description: The MCP server the tool was served from. Present only for an MCP-backed tool.
        mcp_tool:
          type: string
          description: The tool name on that MCP server. Present only for an MCP-backed tool.
        mcp_response:
          type: string
          description: What the MCP server returned. Present only for an MCP-backed tool.
        mcp_error:
          type: boolean
          description: Always `true` when present, meaning the MCP call failed.
          examples:
          - true
      unevaluatedProperties:
        not: {}
      description: One tool call the agent made, in the order it was made.
    Webhooks.AI.AISidecarChannelData:
      type: object
      properties:
        call_id:
          type: string
          description: ID of the call the sidecar is attached to.
          examples:
          - 2e1e66e5-5d07-413d-9668-55542992eec0
        caller_id_name:
          type: string
          description: The caller's name.
          examples:
          - Jane Doe
        caller_id_number:
          type: string
          description: The caller's number.
          examples:
          - '+15555550100'
        destination_number:
          type: string
          description: The number that was called.
          examples:
          - '+15555550199'
      unevaluatedProperties:
        not: {}
      description: Identifies the call a sidecar callback came from. Each field is included when the call has it.
    Webhooks.AI.BedrockPostPromptCallbackPayload:
      type: object
      required:
      - content_type
      - content_disposition
      - conversation_type
      - call_id
      - app_name
      - ai_session_id
      - action
      - call_log
      - raw_call_log
      - post_prompt_data
      - global_data
      - swaig_log
      - conversation_summary
      properties:
        project_id:
          type: string
          description: Your project ID, when available.
          examples:
          - 4d0d6f16-5881-4fcc-92a4-02c51a91954d
        space_id:
          type: string
          description: Your Space ID, when available.
          examples:
          - 451ed9ff-e568-4222-8af9-4f9ab7428d09
        content_type:
          type: string
          description: The content type of the request body. Always `text/json`.
          examples:
          - text/json
        content_disposition:
          type: string
          description: How the body is delivered. Always `agent.summary` for the end-of-call report.
          examples:
          - agent.summary
        conversation_type:
          type: string
          description: The kind of conversation the agent ran. Always `voice`.
          examples:
          - voice
        call_id:
          type: string
          description: The ID of the call.
          examples:
          - 2e1e66e5-5d07-413d-9668-55542992eec0
        app_name:
          type: string
          description: The name of your Bedrock application. Defaults to `bedrock`.
          examples:
          - bedrock
        ai_session_id:
          type: string
          description: The ID of the AI session on the call. Matches `call_id` for Bedrock agents.
          examples:
          - 2e1e66e5-5d07-413d-9668-55542992eec0
        conversation_id:
          type: string
          description: The conversation ID, when the agent was configured with one.
          examples:
          - support-thread-4821
        action:
          type: string
          description: What the request is asking of you. Always `post_conversation` for the end-of-call report.
          examples:
          - post_conversation
        call_log:
          type: array
          items:
            $ref: '#/components/schemas/Webhooks.AI.AICallLogEntry'
          description: 'The conversation. Each entry carries a `role` of `system`, `user`, or `assistant` and its

            `content`.'
          examples:
          - - role: system
              content: You dispatch taxis.
            - role: user
              content: I need a ride to the airport.
        raw_call_log:
          type: array
          items:
            $ref: '#/components/schemas/Webhooks.AI.AICallLogEntry'
          description: A copy of `call_log`. Bedrock agents do not keep a separate unfiltered log.
        call_start_date:
          type: integer
          format: int64
          description: When the call was created, as a Unix timestamp in microseconds.
          examples:
          - 1694541295773508
        call_answer_date:
          type: integer
          format: int64
          description: When the call was answered, as a Unix timestamp in microseconds. `0` when it never was.
          examples:
          - 1694541296799504
        call_end_date:
          type: integer
          format: int64
          description: When the call ended, as a Unix timestamp in microseconds. `0` while the call is still up.
          examples:
          - 1694541335435503
        ai_start_date:
          type: integer
          format: int64
          description: When the agent started, as a Unix timestamp in microseconds.
          examples:
          - 1694541297950440
        ai_end_date:
          type: integer
          format: int64
          description: When the agent stopped, as a Unix timestamp in microseconds. Omitted while it is still running.
          examples:
          - 1694541335425164
        caller_id_name:
          type: string
          description: The caller's name, when available.
          examples:
          - Jane Doe
        caller_id_number:
          type: string
          description: The caller's number, when available.
          examples:
          - '+15555550100'
        times:
          type: array
          items:
            $ref: '#/components/schemas/Webhooks.AI.AIResponseTiming'
          description: Per-response performance metrics, one entry per generated reply. Included once the agent has any.
        total_minutes:
          type: number
          description: 'The number of billable minutes, with a minimum of one. Fractional durations are kept as they are;

            only a call under a minute is raised to `1`. Included once the agent has stopped.'
          examples:
          - 3.14
        total_input_tokens:
          type: integer
          description: Input tokens the session consumed. Included once the agent has stopped.
          examples:
          - 5627
        total_output_tokens:
          type: integer
          description: Output tokens the session produced. Included once the agent has stopped.
          examples:
          - 119
        total_wire_input_tokens:
          type: integer
          description: A copy of `total_input_tokens`. Included once the agent has stopped.
          examples:
          - 5627
        total_wire_input_tokens_per_minute:
          type: number
          description: '`total_input_tokens` divided by `total_minutes`. Included once the agent has stopped.'
          examples:
          - 1792.04
        total_wire_output_tokens:
          type: integer
          description: A copy of `total_output_tokens`. Included once the agent has stopped.
          examples:
          - 119
        total_wire_output_tokens_per_minute:
          type: number
          description: '`total_output_tokens` divided by `total_minutes`. Included once the agent has stopped.'
          examples:
          - 37.9
        total_tts_chars:
          type: integer
          description: Characters sent to text-to-speech. Included once the agent has stopped.
          examples:
          - 842
        total_tts_chars_per_min:
          type: number
          description: '`total_tts_chars` divided by `total_minutes`. Included once the agent has stopped.'
          examples:
          - 268.15
        total_asr_minutes:
          type: number
          description: Minutes of audio sent to speech recognition. Included once the agent has stopped.
          examples:
          - 2.41
        total_asr_cost_factor:
          type: number
          description: Always `1`. Bedrock agents do not vary the factor. Included once the agent has stopped.
          examples:
          - 1
        SWMLVars:
          type: object
          unevaluatedProperties: {}
          

# --- truncated at 32 KB (103 KB total) ---
# Full source: https://raw.githubusercontent.com/api-evangelist/signalwire/refs/heads/main/openapi/signalwire-ai-webhooks-api-openapi.yml