TalkPush Agents API API

Collection of endpoints related to AI Agents (SmartCall Settings). Agents are configurable AI-powered voice/interview agents that can be linked to campaigns, question sets, and candidate attribute extractions.

OpenAPI Specification

talkpush-agents-api-api-openapi.yml Raw ↑
openapi: 3.0.0
info:
  description: '<b>Production Base URL to be used during implementation:</b> {your company workspace subdomain}.talkpush.com/api/talkpush_services


    With the Talkpush APIs you can instantly plug your lead source into out platform to enjoy the benefits of Talkpush. You can also synch your <b> existing HR / Recruitment technology stack </b> with your Talkpush account. To be able to use our API you will need pass an api key as a parameter for each API call.  In case you do not know your current API key you can contact our support team at cs@talkpush.com.

    '
  version: '2.0'
  title: Talkpush Agents API API
  contact:
    email: admin@talkpush.com
servers:
- url: https://company_subdomain.talkpush.com/api/talkpush_services
tags:
- name: Agents API
  description: Collection of endpoints related to AI Agents (SmartCall Settings). Agents are configurable AI-powered voice/interview agents that can be linked to campaigns, question sets, and candidate attribute extractions.
paths:
  /agents:
    get:
      tags:
      - Agents API
      summary: Retrieve a list of agents
      description: Returns a paginated list of AI agents configured in the company account, ordered by most recently updated. Supports optional filtering by category, provider, and active status.
      parameters:
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      - name: category
        in: query
        required: false
        description: 'Filter agents by category. Allowed values: `screen_interview`, `phone_call`, `on_site_interview`, `event_invitation`, `passive_candidate_outreach`, `status_update`, `reminders`.'
        schema:
          type: string
          enum:
          - screen_interview
          - phone_call
          - on_site_interview
          - event_invitation
          - passive_candidate_outreach
          - status_update
          - reminders
      - name: provider
        in: query
        required: false
        description: 'Filter agents by provider. Allowed values: `elevenlabs`, `vapi`.'
        schema:
          type: string
          enum:
          - elevenlabs
          - vapi
      - name: is_active
        in: query
        required: false
        description: Filter agents by active status. `true` returns only active agents; `false` returns only inactive agents. Omit to return all.
        schema:
          type: boolean
      - name: page
        in: query
        required: false
        description: Page number of results to return. Defaults to 1.
        schema:
          type: integer
          default: 1
      - name: per_page
        in: query
        required: false
        description: Number of agents per page. Defaults to 20, maximum 100.
        schema:
          type: integer
          default: 20
          maximum: 100
      responses:
        '200':
          description: Agents returned successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agents_list_response'
              example:
                data:
                - id: '6'
                  name: Daniel's Agent
                  description: Evaluates candidate English proficiency
                  category: screen_interview
                  prompt: ''
                  provider: elevenlabs
                  provider_agent_id: agent_abc123xyz456
                  question_set_id: '1'
                  data_extractions_count: 2
                  is_active: true
                  created_at: '2026-02-11T06:33:43+08:00'
                  updated_at: '2026-04-07T01:33:01+08:00'
                pagination:
                  page: 1
                  per_page: 20
                  total: 1
                  total_pages: 1
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
    post:
      tags:
      - Agents API
      summary: Create a new agent
      description: 'Creates a new AI agent configuration. On success, returns HTTP 201 with the newly created agent details.


        The `data_extractions` field in the request uses the candidate attribute **key** string (e.g. `"english_level"`) — unlike the update endpoint, which uses `{ "id": "..." }`. The response returns an array of extraction objects (not the keyed map returned by the show/update endpoints).'
      parameters:
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/agent_create_request'
            example:
              agent:
                name: My New Agent
                description: Evaluates candidate English proficiency
                category: screen_interview
                provider: elevenlabs
                provider_agent_id: agent_abc123xyz456
                question_set_id: '1'
                is_active: true
                prompt: "This is the prompt for my agent I need:\n {{messenger_interview_url}}"
                data_extractions:
                - candidate_attribute: english_level
                  data_type: string
                  description: Candidate English proficiency level
                  number_of_values: 1
                  allowed_values:
                  - basic
                  - intermediate
                  - advanced
      responses:
        '201':
          description: Agent created successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agent_create_response'
              example:
                id: '42'
                name: My New Agent
                description: Evaluates candidate English proficiency
                category: screen_interview
                prompt: 'This is the prompt for my agent I need: Messenger Interview URL'
                question_set_id: '1'
                provider: elevenlabs
                provider_agent_id: agent_abc123xyz456
                data_extractions:
                - id: '29'
                  candidate_attribute: english_level
                  data_type: string
                  description: Candidate English proficiency level
                  number_of_values: 1
                  allowed_values:
                  - basic
                  - intermediate
                  - advanced
                is_active: true
                created_at: '2026-04-24T10:00:00+08:00'
                updated_at: '2026-04-24T10:00:00+08:00'
        '422':
          description: Validation error. Returned when the request body contains invalid data — e.g. an invalid `category`, an invalid `provider`, malformed `data_extractions`, or invalid Handlebars in the prompt.
          content:
            application/json:
              schema:
                type: object
                properties:
                  errors:
                    type: array
                    items:
                      type: string
                    example:
                    - 'provider must be one of: elevenlabs, vapi'
                    - 'category must be one of: screen_interview, phone_call, on_site_interview, event_invitation, passive_candidate_outreach, status_update, reminders'
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
  /agents/{id}:
    get:
      tags:
      - Agents API
      summary: Retrieve details of a specific agent
      description: Returns the full detail of a single agent, including its prompt, provider info, linked question set, and the complete data extractions configuration.
      parameters:
      - name: id
        in: path
        required: true
        description: The ID of the agent.
        schema:
          type: integer
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      responses:
        '200':
          description: Agent returned successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agent_detail_response'
              example:
                data:
                  id: '6'
                  name: Daniel's Agent
                  description: Evaluates candidate English proficiency
                  category: screen_interview
                  prompt: This is the prompt for my agent...
                  provider: elevenlabs
                  provider_info:
                    name: elevenlabs
                    vendor: ElevenLabs
                  provider_agent_id: agent_abc123xyz456
                  question_set_id: '1'
                  question_set:
                    id: '1'
                    title: Set name
                  data_extractions:
                    English Level:
                      type: string
                      description: Candidate English proficiency level
                      enum:
                      - basic
                      - intermediate
                      - advanced
                      candidate_attribute:
                        id: '29'
                  is_active: true
                  created_at: '2026-02-11T06:33:43+08:00'
                  updated_at: '2026-04-07T01:33:01+08:00'
        '404':
          description: Agent not found.
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
    put:
      tags:
      - Agents API
      summary: Update an existing agent
      description: Updates an agent's configuration. All fields are optional — only the fields provided will be changed. **Note:** if `data_extractions` is included, it **replaces** the existing extractions entirely (not a partial merge). The same applies to `prompt`.
      parameters:
      - name: id
        in: path
        required: true
        description: The ID of the agent to update.
        schema:
          type: integer
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/agent_update_request'
            example:
              agent:
                name: My Agent
                description: Evaluates candidate English proficiency
                category: screen_interview
                provider: elevenlabs
                provider_agent_id: agent_abc123xyz456
                question_set_id: '1'
                is_active: true
                prompt: "This is the prompt for my agent I need:\n {{messenger_interview_url}}"
                data_extractions:
                - candidate_attribute:
                    id: '29'
                  description: Candidate English proficiency level
                  data_type: string
                  enum_values:
                  - basic
                  - intermediate
                  - advanced
      responses:
        '200':
          description: Agent updated successfully. Returns the full updated agent detail.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agent_detail_response'
        '404':
          description: Agent not found.
        '422':
          description: Validation error. Returned when the request body contains invalid data — e.g. an invalid `category`, an invalid `provider`, or malformed `data_extractions`.
          content:
            application/json:
              schema:
                type: object
                properties:
                  errors:
                    type: array
                    items:
                      type: string
                    example:
                    - 'provider must be one of: elevenlabs, vapi'
                    - 'category must be one of: screen_interview, phone_call, on_site_interview, event_invitation, passive_candidate_outreach, status_update, reminders'
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
    delete:
      tags:
      - Agents API
      summary: Delete an agent
      description: Permanently deletes the agent identified by `id`. On success, returns HTTP 204 with no response body.
      parameters:
      - name: id
        in: path
        required: true
        description: The ID of the agent to delete.
        schema:
          type: integer
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      responses:
        '204':
          description: Agent deleted successfully. No response body is returned.
        '404':
          description: Agent not found.
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
  /agents/{agent_id}/calls:
    get:
      tags:
      - Agents API
      summary: Retrieve a list of calls for a specific agent
      description: 'Returns a paginated list of call records handled by the given agent, ordered by most recent first. Each record includes the associated application (candidate / campaign invitation) context, extracted data, interview insights, transcript, and a link to the call recording when available.


        Optional filters:

        - `status`: filter by call status.

        - `start_date` + `end_date`: filter by creation date range. Both parameters must be provided together. Any format accepted by `Time.zone.parse` is supported (e.g. `2026-04-01`, `2026-04-01T00:00:00Z`).'
      parameters:
      - name: agent_id
        in: path
        required: true
        description: The ID of the agent whose calls should be returned.
        schema:
          type: integer
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      - name: status
        in: query
        required: false
        description: Filter by call status.
        schema:
          type: string
          enum:
          - pending
          - queued
          - in_progress
          - ended
          - completed
          - declined
          - voicemail
          - unfinished
          - not_taken
          - failed
          - dropped
      - name: start_date
        in: query
        required: false
        description: Start of the creation-date range filter. Must be provided together with `end_date`. Any format parseable by `Time.zone.parse` is accepted (e.g. `2026-04-01`, `2026-04-01T00:00:00Z`).
        schema:
          type: string
      - name: end_date
        in: query
        required: false
        description: End of the creation-date range filter. Must be provided together with `start_date`.
        schema:
          type: string
      - name: page
        in: query
        required: false
        description: Page number of the results to return. Default is 1.
        schema:
          type: integer
          default: 1
      - name: per_page
        in: query
        required: false
        description: Number of records per page. Default is 20, maximum is 100.
        schema:
          type: integer
          default: 20
          maximum: 100
      responses:
        '200':
          description: Calls returned successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agent_calls_response'
              example:
                data:
                - company: Acme
                  company_id: 123
                  company_tenant: acme
                  status: completed
                  summary: Candidate completed the screening interview successfully.
                  extracted_data:
                    English Level: advanced
                  interview_insights:
                    strengths:
                    - clear articulation
                    weaknesses: []
                  transcript: 'Agent: Hello...

                    Candidate: Hi...'
                  recording_url: https://cdn.talkpush.com/recordings/abc123.mp3
                  identifier: call_abc123xyz456
                  agent:
                    identifier: agent_abc123xyz456
                  application:
                    id: 987654
                    candidate_id: 123456
                    campaign_id: 4321
                    campaign_title: Customer Service Representative
                    first_name: Jane
                    last_name: Doe
                    email: jane.doe@example.com
                    msisdn: '639171234567'
                    others:
                      talkscore_overall: '85'
                      talkscore_cefr: C1
                    scheduled_at: '2026-04-10T10:00:00+08:00'
                pagination:
                  page: 1
                  per_page: 20
                  total: 42
                  total_pages: 3
        '400':
          description: 'Bad request. Possible messages:

            - `start_date and end_date are both required when filtering by date` (when only one of the two date parameters is supplied).

            - `Invalid date format for start_date or end_date` (when a date parameter cannot be parsed).

            - `status must be one of: <allowed values>` (when the provided status is not recognized).'
        '401':
          description: Invalid API key.
        '404':
          description: Agent not found.
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
  /agents/{agent_id}/changelog:
    get:
      tags:
      - Agents API
      summary: Retrieve the changelog for a specific agent
      description: 'Returns a paginated list of audit log entries for the given agent, ordered by most recent first. Each entry describes a single create, update, or destroy event — who made the change, when, and what fields were affected.


        For `created` entries the `changes` array is always empty (the initial state is not diffed). For `updated` entries each item in `changes` names the API field that changed and includes both the previous and new value.


        **Possible `field` values inside a change entry:**

        - `agent_name` — display name of the agent (stored as `title`)

        - `description` — short description

        - `prompt` — agent prompt, rendered as a plain-text string

        - `data_extractions` — array of `{ "candidate_attribute": "<key>" }` objects

        - `category` — category enum key (e.g. `talkscore_interview`, `phone_calls`, `onsite_interview`, `event_invitation`, `passive_candidate_outreach`, `status_update`, `reminder`)

        - `provider` — provider key (e.g. `elevenlabs`, `vapi`)

        - `provider_agent_id` — provider-side agent identifier

        - `question_set_id` — ID of the linked question set

        - `is_active` — boolean active/inactive flag


        **`changed_by` values:**

        - Manager email address when the change was made from the Talkpush CRM

        - `"TalkPush API"` when the change was made via the API'
      parameters:
      - name: agent_id
        in: path
        required: true
        description: The ID of the agent whose changelog should be returned.
        schema:
          type: integer
      - name: api_key
        in: query
        required: true
        description: The API key provided for your application.
        schema:
          type: string
      - name: page
        in: query
        required: false
        description: Page number of the results to return. Default is 1.
        schema:
          type: integer
          default: 1
      - name: per_page
        in: query
        required: false
        description: Number of records per page. Default is 20, maximum is 100.
        schema:
          type: integer
          default: 20
          maximum: 100
      responses:
        '200':
          description: Changelog returned successfully.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/agent_changelog_response'
              example:
                agent_id: '6'
                data:
                - id: '4917'
                  action: updated
                  changed_at: '2026-04-07T01:33:01Z'
                  changed_by: TalkPush API
                  changes:
                  - field: agent_name
                    previous_value: Daniel's agent
                    new_value: Daniel's Agent Updated via Postman
                  - field: prompt
                    previous_value: ''
                    new_value: 'This is the prompt for my agent I need: Messenger Interview URL then Interview Link'
                  - field: data_extractions
                    previous_value:
                    - candidate_attribute: admired_brands
                    new_value:
                    - candidate_attribute: admired_brands
                    - candidate_attribute: bpo_experience
                  - field: category
                    previous_value: event_invitation
                    new_value: talkscore_interview
                  - field: provider
                    previous_value: vapi
                    new_value: elevenlabs
                  - field: is_active
                    previous_value: false
                    new_value: true
                - id: '4852'
                  action: created
                  changed_at: '2026-03-26T10:30:00Z'
                  changed_by: user@example.com
                  changes: []
                pagination:
                  page: 1
                  per_page: 20
                  total: 2
                  total_pages: 1
        '401':
          description: Invalid API key.
        '404':
          description: Agent not found.
        '500':
          description: Internal server error, we will provide you with the exception message of the application.
components:
  schemas:
    changelog_entry:
      type: object
      description: A single audit event recorded for an agent.
      properties:
        id:
          type: string
          description: Unique identifier of the audit entry.
        action:
          type: string
          description: Type of event. `created` — agent was first created; `updated` — one or more fields were changed; `destroyed` — agent was deleted.
          enum:
          - created
          - updated
          - destroyed
        changed_at:
          type: string
          format: date-time
          description: ISO 8601 timestamp of when the change occurred.
        changed_by:
          type: string
          description: Who made the change. Contains the manager's email address when the change originated from the Talkpush CRM, or `"TalkPush API"` when the change was made via the API.
        changes:
          type: array
          description: Field-level diff for this event. Always empty for `created` entries.
          items:
            $ref: '#/components/schemas/changelog_change'
      required:
      - id
      - action
      - changed_at
      - changed_by
      - changes
    agent_create_request:
      type: object
      required:
      - agent
      properties:
        agent:
          type: object
          description: Agent creation payload. All fields except `name` are optional.
          properties:
            name:
              type: string
              description: Display name of the agent.
            description:
              type: string
              description: Short description of the agent's purpose.
            category:
              type: string
              description: Use-case category of the agent.
              enum:
              - screen_interview
              - phone_call
              - on_site_interview
              - event_invitation
              - passive_candidate_outreach
              - status_update
              - reminders
            provider:
              type: string
              description: AI provider powering the agent.
              enum:
              - elevenlabs
              - vapi
            provider_agent_id:
              type: string
              description: The agent identifier on the external provider's platform.
            question_set_id:
              type: string
              description: ID of the question set to link to this agent.
            is_active:
              type: boolean
              description: Set to `true` to create the agent in active (published) state; `false` for inactive (draft). Defaults to inactive when omitted.
            prompt:
              type: string
              description: 'The agent''s prompt. Supports Handlebars-style variables: `{{messenger_interview_url}}`, `{{interview_page_url}}`, `{{messenger_interview_link}}`.'
            data_extractions:
              type: array
              description: 'List of data extraction rules to configure on the agent. Unlike the update endpoint (which uses `{ candidate_attribute: { id } }`), here `candidate_attribute` is the attribute **key** string as returned by the Candidate Attributes API.'
              items:
                type: object
                required:
                - candidate_attribute
                - data_type
                properties:
                  candidate_attribute:
                    type: string
                    description: The `key` of the candidate attribute to map the extracted value to (e.g. `"english_level"`). Use the Candidate Attributes API to look up valid keys.
                  data_type:
                    type: string
                    description: Data type of the extracted value (e.g. `string`, `integer`).
                  description:
                    type: string
                    description: Instructions to the AI on what to extract.
                  number_of_values:
                    type: integer
                    description: Expected number of values to extract. Defaults to the length of `allowed_values` when provided, otherwise 1.
                  allowed_values:
                    type: array
                    items:
                      type: string
                    description: Optional list of allowed values for the extraction. When provided, the extraction is constrained to these values and `number_of_values` defaults to their count.
    agent_calls_pagination:
      type: object
      properties:
        page:
          type: integer
          description: Current page number.
        per_page:
          type: integer
          description: Number of records per page.
        total:
          type: integer
          description: Total number of call records matching the query.
        total_pages:
          type: integer
          description: Total number of pages available.
    agent_update_request:
      type: object
      required:
      - agent
      properties:
        agent:
          type: object
          description: All fields are optional. Only the fields provided will be updated. If `data_extractions` is sent, it fully replaces the existing extractions.
          properties:
            name:
              type: string
              description: Display name of the agent.
            description:
              type: string
              description: Short description of the agent's purpose.
            category:
              type: string
              description: Use-case category of the agent.
              enum:
              - screen_interview
              - phone_call
              - on_site_interview
              - event_invitation
              - passive_candidate_outreach
              - status_update
              - reminders
            provider:
              type: string
              description: AI provider powering the agent.
              enum:
              - elevenlabs
              - vapi
            provider_agent_id:
              type: string
              description: The agent identifier on the external provider's platform.
            question_set_id:
              type: string
              description: ID of the question set to link to this agent.
            is_active:
              type: boolean
              description: Set to `true` to activate the agent (published); `false` to deactivate it (draft).
            prompt:
              type: string
              description: 'The agent''s prompt. Supports Handlebars-style variables: `{{messenger_interview_url}}`, `{{interview_page_url}}`, `{{messenger_interview_link}}`.'
            data_extractions:
              type: array
              description: Fully replaces all existing data extraction rules when provided. Send an empty array to remove all extractions.
              items:
                type: object
                required:
                - candidate_attribute
                - description
                - data_type
                properties:
                  candidate_attribute:
                    type: object
                    required:
                    - id
                    properties:
                      id:
                        type: string
                        description: ID of the candidate attribute to map the extracted value to.
                  description:
                    type: string
                    description: Instructions to the AI on what to extract.
                  data_type:
                    type: string
                    description: Data type of the extracted value (e.g. `string`, `integer`).
                  enum_values:
                    type: array
                    items:
                      type: string
                    description: Optional list of allowed values. Send an empty array for unconstrained extraction.
    agent_create_response:
      type: object
      description: Response body returned after successfully creating an agent (HTTP 201). The `data_extractions` field is an array — unlike the keyed map returned by the show/update endpoints.
      properties:
        id:
          type: string
          description: Unique identifier of the newly created agent.
        name:
          type: string
          nullable: true
          description: Display name of the agent.
        description:
          type: string
          nullable: true
          description: Short description of the agent's purpose.
        category:
          type: string
          nullable: true
          description: Use-case category of the agent.
          enum:
          - screen_interview
          - phone_call
          - on_site_interview
          - event_invitation
          - passive_candidate_outreach
          - status_update
          - reminders
        prompt:
          type: string
          description: The agent's prompt string.
        question_set_id:
          type: string
          nullable: true
          description: ID of the linked question set, if any.
        provider:
          type: string
          nullable: true
          description: AI provider powering the agent.
          enum:
          - elevenlabs
          - vapi
        provider_agent_id:
          type: string
          nullable: true
          description: The agent identifier on the external provider's platform.
        data_extractions:
          type: array
          description: List of data extraction rules configured on the agent.
          items:
            type: object
            properties:
              id:
                type: string
                nullable: true
                description: ID of the candidate attribute record.
              candidate_attribute:
                type: string
                nullable: true
                description: Key of the candidate attribute this extraction maps to.
              data_type:
                type: string
                description: Data type of the extracted value.
              description:
                type: string
                nullable: true
                description: Instr

# --- truncated at 32 KB (44 KB total) ---
# Full source: https://raw.githubusercontent.com/api-evangelist/talkpush/refs/heads/main/openapi/talkpush-agents-api-api-openapi.yml