Module airbyte_agent_sdk.connectors.zendesk_talk.connector

Zendesk-Talk connector.

Classes

AccountOverviewQuery(connector: ZendeskTalkConnector) : Query class for AccountOverview entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: AccountOverviewSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[AccountOverviewSearchData] : Search account_overview records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (AccountOverviewSearchFilter):

average_call_duration: Average call duration
average_callback_wait_time: Average callback wait time
average_hold_time: Average hold time per call
average_queue_wait_time: Average queue wait time
average_time_to_answer: Average time to answer
average_wrap_up_time: Average wrap-up time
current_timestamp: Current timestamp
max_calls_waiting: Max calls waiting in queue
max_queue_wait_time: Max queue wait time
total_call_duration: Total call duration
total_callback_calls: Total callback calls
total_calls: Total calls
total_calls_abandoned_in_queue: Total calls abandoned in queue
total_calls_outside_business_hours: Total calls outside business hours
total_calls_with_exceeded_queue_wait_time: Total calls exceeding max queue wait time
total_calls_with_requested_voicemail: Total calls requesting voicemail
total_embeddable_callback_calls: Total embeddable callback calls
total_hold_time: Total hold time
total_inbound_calls: Total inbound calls
total_outbound_calls: Total outbound calls
total_textback_requests: Total textback requests
total_voicemails: Total voicemails
total_wrap_up_time: Total wrap-up time

Args: query: Filter and sort conditions. Supports operators such as eq, neq, gt, gte, lt, lte, in, startswith, endswith, contains, array_contains, fuzzy, keyword, not, and, or. Example: {"filter": {"eq": {"status": "active"}}} limit: Maximum results to return (default 1000) cursor: Pagination cursor from previous response's meta.cursor fields: Field paths to include in results. Each path is a list of keys for nested access. Example: [["id"], ["user", "name"]] returns id and user.name fields.

Returns: AccountOverviewSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against account_overview records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResult[AccountOverview] : Returns overview statistics for the account for the current day

Returns: AccountOverviewListResult

AddressesQuery(connector: ZendeskTalkConnector) : Query class for Addresses entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: AddressesSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[AddressesSearchData] : Search addresses records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (AddressesSearchFilter):

city: City of the address
country_code: ISO country code
id: Unique address identifier
name: Name of the address
provider_reference: Provider reference of the address
province: Province of the address
state: State of the address
street: Street of the address
zip: Zip code of the address

Returns: AddressesSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

get(self, address_id: str, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.Address : Retrieves a single address by ID

Args: address_id: ID of the address **kwargs: Additional parameters

Returns: Address

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[Address], AddressesListResultMeta] : Returns a list of all addresses in the Zendesk Talk account

Returns: AddressesListResult

AgentsActivityQuery(connector: ZendeskTalkConnector) : Query class for AgentsActivity entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: AgentsActivitySearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[AgentsActivitySearchData] : Search agents_activity records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (AgentsActivitySearchFilter):

accepted_third_party_conferences: Accepted third party conferences
accepted_transfers: Total transfers accepted
agent_id: Agent ID
agent_state: Agent state: online, offline, away, or transfers_only
available_time: Total time agent was available to answer calls
avatar_url: URL to agent avatar
average_hold_time: Average hold time per call
average_talk_time: Average talk time per call
average_wrap_up_time: Average wrap-up time per call
away_time: Total time agent was set to away
call_status: Agent call status: on_call, wrap_up, or null
calls_accepted: Total calls accepted
calls_denied: Total calls denied
calls_missed: Total calls missed
calls_put_on_hold: Total calls placed on hold
forwarding_number: Forwarding number set by the agent
name: Agent name
online_time: Total online time
started_third_party_conferences: Started third party conferences
started_transfers: Total transfers started
total_call_duration: Total call duration
total_hold_time: Total hold time across all calls
total_talk_time: Total talk time (excludes hold)
total_wrap_up_time: Total wrap-up time
transfers_only_time: Total time in transfers-only mode
via: Channel the agent is registered on

Returns: AgentsActivitySearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against agents_activity records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[AgentActivity], AgentsActivityListResultMeta] : Returns activity statistics for all agents for the current day

Returns: AgentsActivityListResult

AgentsOverviewQuery(connector: ZendeskTalkConnector) : Query class for AgentsOverview entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: AgentsOverviewSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[AgentsOverviewSearchData] : Search agents_overview records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (AgentsOverviewSearchFilter):

average_accepted_transfers: Average accepted transfers
average_available_time: Average available time
average_away_time: Average away time
average_calls_accepted: Average calls accepted
average_calls_denied: Average calls denied
average_calls_missed: Average calls missed
average_calls_put_on_hold: Average calls put on hold
average_hold_time: Average hold time
average_online_time: Average online time
average_started_transfers: Average started transfers
average_talk_time: Average talk time
average_transfers_only_time: Average transfers-only time
average_wrap_up_time: Average wrap-up time
current_timestamp: Current timestamp
total_accepted_transfers: Total accepted transfers
total_calls_accepted: Total calls accepted
total_calls_denied: Total calls denied
total_calls_missed: Total calls missed
total_calls_put_on_hold: Total calls put on hold
total_hold_time: Total hold time
total_started_transfers: Total started transfers
total_talk_time: Total talk time
total_wrap_up_time: Total wrap-up time

Returns: AgentsOverviewSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against agents_overview records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResult[AgentsOverview] : Returns overview statistics for all agents for the current day

Returns: AgentsOverviewListResult

CallLegsQuery(connector: ZendeskTalkConnector) : Query class for CallLegs entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: CallLegsSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[CallLegsSearchData] : Search call_legs records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (CallLegsSearchFilter):

agent_id: Agent ID
available_via: Channel agent was available through
call_charge: Call charge amount
call_id: Associated call ID
completion_status: Completion status
conference_from: Conference from time
conference_time: Conference duration
conference_to: Conference to time
consultation_from: Consultation from time
consultation_time: Consultation duration
consultation_to: Consultation to time
created_at: Creation timestamp
duration: Duration in seconds
forwarded_to: Number forwarded to
hold_time: Hold time in seconds
id: Call leg ID
minutes_billed: Minutes billed
quality_issues: Quality issues detected
talk_time: Talk time in seconds
transferred_from: Transferred from agent ID
transferred_to: Transferred to agent ID
type_: Type of call leg
updated_at: Last update timestamp
user_id: User ID
wrap_up_time: Wrap-up time in seconds

Returns: CallLegsSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, start_time: int, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[CallLeg], CallLegsListResultMeta] : Returns incremental call leg data. Requires a start_time parameter (Unix epoch timestamp).

Args: start_time: Unix epoch time to start from (e.g. 1704067200 for 2024-01-01) **kwargs: Additional parameters

Returns: CallLegsListResult

CallsQuery(connector: ZendeskTalkConnector) : Query class for Calls entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: CallsSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[CallsSearchData] : Search calls records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (CallsSearchFilter):

agent_id: Agent ID
call_charge: Call charge amount
call_group_id: Call group ID
call_recording_consent: Call recording consent status
call_recording_consent_action: Recording consent action
call_recording_consent_keypress: Recording consent keypress
callback: Whether this was a callback
callback_source: Source of the callback
completion_status: Call completion status
consultation_time: Consultation time
created_at: Creation timestamp
customer_requested_voicemail: Whether customer requested voicemail
default_group: Whether default group was used
direction: Call direction (inbound/outbound)
duration: Call duration in seconds
exceeded_queue_time: Whether queue time was exceeded
exceeded_queue_wait_time: Whether max queue wait time was exceeded
hold_time: Hold time in seconds
id: Call ID
ivr_action: IVR action taken
ivr_destination_group_name: IVR destination group name
ivr_hops: Number of IVR hops
ivr_routed_to: Where IVR routed the call
ivr_time_spent: Time spent in IVR
minutes_billed: Minutes billed
not_recording_time: Time not recording
outside_business_hours: Whether call was outside business hours
overflowed: Whether call overflowed
overflowed_to: Where call overflowed to
phone_number: Phone number used
phone_number_id: Phone number ID
quality_issues: Quality issues detected
recording_control_interactions: Recording control interactions count
recording_time: Recording time
talk_time: Talk time in seconds
ticket_id: Associated ticket ID
time_to_answer: Time to answer in seconds
updated_at: Last update timestamp
voicemail: Whether it was a voicemail
wait_time: Wait time in seconds
wrap_up_time: Wrap-up time in seconds

Returns: CallsSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, start_time: int, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[Call], CallsListResultMeta] : Returns incremental call data. Requires a start_time parameter (Unix epoch timestamp).

Args: start_time: Unix epoch time to start from (e.g. 1704067200 for 2024-01-01) **kwargs: Additional parameters

Returns: CallsListResult

CurrentQueueActivityQuery(connector: ZendeskTalkConnector) : Query class for CurrentQueueActivity entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: CurrentQueueActivitySearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[CurrentQueueActivitySearchData] : Search current_queue_activity records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (CurrentQueueActivitySearchFilter):

agents_online: Current number of agents online
average_wait_time: Average wait time for callers in queue (seconds)
callbacks_waiting: Number of callers in callback queue
calls_waiting: Number of callers waiting in queue
current_timestamp: Current timestamp
embeddable_callbacks_waiting: Number of Web Widget callback requests waiting
longest_wait_time: Longest wait time for any caller (seconds)

Returns: CurrentQueueActivitySearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against current_queue_activity records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResult[CurrentQueueActivity] : Returns current queue activity statistics

Returns: CurrentQueueActivityListResult

GreetingCategoriesQuery(connector: ZendeskTalkConnector) : Query class for GreetingCategories entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: GreetingCategoriesSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[GreetingCategoriesSearchData] : Search greeting_categories records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (GreetingCategoriesSearchFilter):

id: Greeting category ID
name: Name of the greeting category

Returns: GreetingCategoriesSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against greeting_categories records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

get(self, greeting_category_id: str, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.GreetingCategory : Retrieves a single greeting category by ID

Args: greeting_category_id: ID of the greeting category **kwargs: Additional parameters

Returns: GreetingCategory

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[GreetingCategory], GreetingCategoriesListResultMeta] : Returns a list of all greeting categories

Returns: GreetingCategoriesListResult

GreetingsQuery(connector: ZendeskTalkConnector) : Query class for Greetings entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: GreetingsSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[GreetingsSearchData] : Search greetings records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (GreetingsSearchFilter):

active: Whether the greeting is associated with phone numbers
audio_name: Audio file name
audio_url: Path to the greeting sound file
category_id: ID of the greeting category
default: Whether this is a system default greeting
default_lang: Whether the greeting has a default language
has_sub_settings: Sub-settings for categorized greetings
id: Greeting ID
ivr_ids: IDs of IVRs associated with the greeting
name: Name of the greeting
pending: Whether the greeting is pending
phone_number_ids: IDs of phone numbers associated with the greeting
upload_id: Upload ID associated with the greeting

Returns: GreetingsSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

get(self, greeting_id: str, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.Greeting : Retrieves a single greeting by ID

Args: greeting_id: ID of the greeting **kwargs: Additional parameters

Returns: Greeting

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[Greeting], GreetingsListResultMeta] : Returns a list of all greetings in the Zendesk Talk account

Returns: GreetingsListResult

IvrsQuery(connector: ZendeskTalkConnector) : Query class for Ivrs entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: IvrsSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[IvrsSearchData] : Search ivrs records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (IvrsSearchFilter):

id: IVR ID
menus: List of IVR menus
name: Name of the IVR
phone_number_ids: IDs of phone numbers configured with this IVR
phone_number_names: Names of phone numbers configured with this IVR

Returns: IvrsSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

get(self, ivr_id: str, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.Ivr : Retrieves a single IVR configuration by ID

Args: ivr_id: ID of the IVR **kwargs: Additional parameters

Returns: Ivr

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[Ivr], IvrsListResultMeta] : Returns a list of all IVR configurations

Returns: IvrsListResult

PhoneNumbersQuery(connector: ZendeskTalkConnector) : Query class for PhoneNumbers entity operations.

Initialize query with connector reference.

Methods

context_store_search(self, query: PhoneNumbersSearchQuery, limit: int | None = None, cursor: str | None = None, fields: list[list[str]] | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[PhoneNumbersSearchData] : Search phone_numbers records from Airbyte cache.

This operation searches cached data from Airbyte syncs. Only available in hosted execution mode.

Available filter fields (PhoneNumbersSearchFilter):

call_recording_consent: What call recording consent is set to
capabilities: Phone number capabilities (sms, mms, voice)
categorised_greetings: Greeting category IDs and names
categorised_greetings_with_sub_settings: Greeting categories with associated settings
country_code: ISO country code for the number
created_at: Date and time the phone number was created
default_greeting_ids: Names of default system greetings
default_group_id: Default group ID
display_number: Formatted phone number
external: Whether this is an external caller ID number
failover_number: Failover number associated with the phone number
greeting_ids: Custom greeting IDs associated with the phone number
group_ids: Array of associated group IDs
id: Unique phone number identifier
ivr_id: ID of IVR associated with the phone number
line_type: Type of line (phone or digital)
location: Geographical location of the number
name: Nickname if set, otherwise the display number
nickname: Nickname of the phone number
number: Phone number digits
outbound_enabled: Whether outbound calls are enabled
priority: Priority level of the phone number
recorded: Whether calls are recorded
schedule_id: ID of schedule associated with the phone number
sms_enabled: Whether SMS is enabled
sms_group_id: Group associated with SMS
token: Generated token unique for the phone number
toll_free: Whether the number is toll-free
transcription: Whether voicemail transcription is enabled
voice_enabled: Whether voice is enabled

Returns: PhoneNumbersSearchResult with typed records, pagination metadata, and optional search metadata

Raises: NotImplementedError: If called in local execution mode

context_store_sql_query(self, sql: str, limit: int | None = None) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.AirbyteSearchResult[dict[str, Any]] : Run a SQL query against phone_numbers records in the Airbyte Context Store.

Only available in hosted execution mode.

Args: sql: SQL query to execute. limit: Maximum results to return.

Returns: AirbyteSearchResult containing the projected rows and query metadata.

Raises: NotImplementedError: If called in local execution mode.

get(self, phone_number_id: str, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.PhoneNumber : Retrieves a single phone number by ID

Args: phone_number_id: ID of the phone number **kwargs: Additional parameters

Returns: PhoneNumber

list(self, **kwargs) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkExecuteResultWithMeta[list[PhoneNumber], PhoneNumbersListResultMeta] : Returns a list of all phone numbers in the Zendesk Talk account

Returns: PhoneNumbersListResult

Auto-generated from OpenAPI specification with full type safety.

Initialize a new zendesk-talk connector instance.

Supports both local and hosted execution modes:

Local mode: Provide connector-specific auth config (e.g., ZendeskTalkAuthConfig)
Hosted mode: Provide AirbyteAuthConfig with client credentials and either connector_id or workspace_name

Args: auth_config: Either connector-specific auth config for local mode, or AirbyteAuthConfig for hosted mode on_token_refresh: Optional callback for OAuth2 token refresh persistence. Called with new_tokens dict when tokens are refreshed. Can be sync or async. Example: lambda tokens: save_to_database(tokens) subdomain: Your Zendesk subdomain (the part before .zendesk.com in your Zendesk URL) Examples:

Local mode (direct API calls)

connector = ZendeskTalkConnector(auth_config=ZendeskTalkAuthConfig(access_token="...", refresh_token="...", client_id="...", client_secret="..."), subdomain="...")

Hosted mode with explicit connector_id (no lookup needed)

connector = ZendeskTalkConnector( auth_config=AirbyteAuthConfig( airbyte_client_id="client_abc123", airbyte_client_secret="secret_xyz789", connector_id="existing-source-uuid" ) )

Hosted mode with lookup by workspace_name

connector = ZendeskTalkConnector( auth_config=AirbyteAuthConfig( workspace_name="user-123", organization_id="00000000-0000-0000-0000-000000000123", airbyte_client_id="client_abc123", airbyte_client_secret="secret_xyz789" ) )

Class variables

connector_name : The type of the None singleton.

connector_version : The type of the None singleton.

sdk_version : The type of the None singleton.

Static methods

agent_tool(role: AgentToolRole | None = None, *, inspect_tool: str | None = None, docs_tool: str | None = None, max_output_chars: int | None | Unset = UNSET, framework: FrameworkName = 'none', internal_retries: int = 0, should_internal_retry: Callable[[Exception, tuple[Any, ...], dict[str, Any]], bool] | None = None, exhausted_runtime_failure_message: Callable[[Exception, tuple[Any, ...], dict[str, Any]], str | None] | None = None) ‑> Callable[[~_F], ~_F] : Decorator for new user-written connector tool functions.

Use this when a tool needs a custom body or the framework lacks a native strategy. Instead of baking the full entity/action reference into the docstring, it instructs the agent to call this connector's inspect and docs tools before executing. Tool failures raise :class:airbyte_agent_sdk.AirbyteToolError by default (framework="none", no auto-detection) — pass framework=... to translate to a supported framework's signal instead.

Decorate three functions per connector — execute, inspect and docs. The role is inferred from each function's signature (extra parameters are allowed); a signature matching more than one role, a generic (*args, **kwargs) wrapper, or a callable whose signature cannot be read must pass the role explicitly:

(entity, action, ...) -> "execute"
(section, ...) -> "read_skill_docs"
() -> "inspect_connector"

Usage: connector = ZendeskTalkConnector(...)

@ZendeskTalkConnector.agent_tool() async def execute(entity: str, action: str, params: dict | None = None): return await connector.execute(entity=entity, action=action, params=params or {})

@ZendeskTalkConnector.agent_tool() async def inspect_connector(): return await connector.inspect_connector()

@ZendeskTalkConnector.agent_tool() async def read_skill_docs(section: str | None = None): return await connector.read_skill_docs(section)

Args: role: "execute" | "inspect_connector" | "read_skill_docs". None (default) infers the role from the decorated function's signature; an explicit role validates the canonical parameters are present (functions accepting **kwargs, or callables whose signature cannot be read, pass validation). inspect_tool: Exact registered name of the sibling inspect tool, woven into the execute docstring for tighter steering. Defaults to generic phrasing. docs_tool: Exact registered name of the sibling docs tool (see inspect_tool). max_output_chars: Max serialized output size before failing. Defaults per role: execute -> DEFAULT_MAX_OUTPUT_CHARS, docs tools -> None. framework: Translation target for tool failures. Defaults to "none" (raise AirbyteToolError); never auto-detects. internal_retries: How many transient runtime failures (429/5xx, network, timeout) to retry silently before surfacing. Forwarded to :func:airbyte_agent_sdk.translation.translate_exceptions. should_internal_retry: Optional predicate (error, args, kwargs) -> bool further restricting which retryable errors are safe for this specific tool. Forwarded to :func:airbyte_agent_sdk.translation.translate_exceptions. exhausted_runtime_failure_message: Optional callback (error, args, kwargs) -> str | None invoked after internal retries are exhausted or skipped. Forwarded to :func:airbyte_agent_sdk.translation.translate_exceptions.

tool_utils(func: _F | None = None, *, update_docstring: bool = True, max_output_chars: int | None = 100000, framework: FrameworkName | None = None, internal_retries: int = 0, should_internal_retry: Callable[[Exception, tuple[Any, ...], dict[str, Any]], bool] | None = None, exhausted_runtime_failure_message: Callable[[Exception, tuple[Any, ...], dict[str, Any]], str | None] | None = None) ‑> ~_F | Callable[[~_F], ~_F] : Deprecated. Add connector-specific documentation and runtime safeguards to one tool.

Kept for backwards compatibility with existing single-tool integrations; it is not removed and does not warn at runtime, but new code should use build_connector_tools or agent_tool below.

For new agents, prefer build_connector_tools. It returns progressive inspect_connector, read_skill_docs, and execute tools so the agent can load only the connector guidance it needs:

from airbyte_agent_sdk import build_connector_tools
from pydantic_ai import Agent

tools = build_connector_tools(connector, framework="pydantic_ai")
agent = Agent("openai:gpt-4o", tools=tools.as_list())

When a new integration needs custom tool bodies or a framework without native support, use agent_tool instead.

Legacy: one generated-description tool

Existing integrations can keep using tool_utils for one broad execute tool with the connector's full generated catalog in its description:

from fastmcp import FastMCP

connector = ZendeskTalkConnector()
mcp = FastMCP("Connector Agent")

@mcp.tool()
@ZendeskTalkConnector.tool_utils
async def execute(entity: str, action: str, params: dict):
    ...

Configure documentation, output limits, framework translation, and retries when needed:

@mcp.tool()
@ZendeskTalkConnector.tool_utils(update_docstring=False, max_output_chars=None)
async def execute(entity: str, action: str, params: dict):
    ...

@mcp.tool()
@ZendeskTalkConnector.tool_utils(framework="pydantic_ai", internal_retries=2)
async def execute(entity: str, action: str, params: dict):
    ...

This decorator composes translate_exceptions for runtime wrapping, output-size checks, framework signal translation, and optional internal retries, then adds connector-specific docstring augmentation.

Args: update_docstring: When True, append connector capabilities to __doc__. max_output_chars: Max serialized output size before raising. Use None to disable. framework: One of "pydantic_ai" | "langchain" | "openai_agents" | "mcp" | "none". Defaults to None, which auto-detects each framework's canonical import in order and falls back to "none" with a warning when no supported framework is installed. Explicit always wins, and an explicit framework whose package is missing raises RuntimeError. internal_retries: How many transient runtime failures (429/5xx, network, timeout) to retry silently before surfacing. Default 0. Forwarded to airbyte_agent_sdk.translation.translate_exceptions. should_internal_retry: Optional predicate (error, args, kwargs) -> bool further restricting which retryable errors are safe for this specific tool. Forwarded to airbyte_agent_sdk.translation.translate_exceptions. exhausted_runtime_failure_message: Optional callback (error, args, kwargs) -> str | None. Invoked after internal retries are exhausted or were skipped because should_internal_retry returned False. Forwarded to airbyte_agent_sdk.translation.translate_exceptions.

Instance variables

connector_id: str | None : Get the connector/source ID (only available in hosted mode).

Returns: The connector ID if in hosted mode, None if in local mode.

Methods

check(self) ‑> airbyte_agent_sdk.connectors.zendesk_talk.models.ZendeskTalkCheckResult : Perform a health check to verify connectivity and credentials.

Executes a lightweight list operation (limit=1) to validate that the connector can communicate with the API and credentials are valid.

Returns: ZendeskTalkCheckResult with status ("healthy" or "unhealthy") and optional error message

Example: result = await connector.check() if result.status == "healthy": print("Connection verified!") else: print(f"Check failed: {result.error}")

close(self) : Close the connector and release resources.

entity_schema(self, entity: str) ‑> dict[str, typing.Any] | None : Get the JSON schema for an entity.

Args: entity: Entity name (e.g., "contacts", "companies")

Returns: JSON schema dict describing the entity structure, or None if not found.

Example: schema = connector.entity_schema("contacts") if schema: print(f"Contact properties: {list(schema.get('properties', {}).keys())}")

execute(self, entity: str, action: "Literal['list', 'get', 'context_store_search', 'context_store_sql_query']", params: Mapping[str, Any] | None = None, *, select_fields: list[str] | None = None, exclude_fields: list[str] | None = None, skip_truncation: bool = True) ‑> Any : Execute an entity operation with full type safety.

This is the recommended interface for blessed connectors as it:

Uses the same signature as non-blessed connectors
Provides full IDE autocomplete for entity/action/params
Makes migration from generic to blessed connectors seamless

Args: entity: Entity name (e.g., "customers") action: Operation action (e.g., "create", "get", "list") params: Operation parameters (typed based on entity+action) select_fields: Optional allowlist of dot-notation fields to include exclude_fields: Optional blocklist of dot-notation fields to remove skip_truncation: Disable long-text truncation for collection actions

Returns: Typed response based on the operation

Example: customer = await connector.execute( entity="customers", action="get", params={"id": "cus_123"} )

inspect_connector(self) ‑> dict[str, typing.Any] : Inspect this connector's hosted metadata/readiness and resolve its docs skill id.

Call this before read_skill_docs in the normal hosted flow. For local/offline connectors this returns a local-mode payload with a warning instead of a hosted inspection.

Example: info = await connector.inspect_connector() print(info["docs_skill_id"])

list_entities(self) ‑> list[dict[str, typing.Any]] : Get structured data about available entities, actions, and parameters.

Returns a list of entity descriptions with:

entity_name: Name of the entity (e.g., "contacts", "deals")
description: Entity description from the first endpoint
available_actions: List of actions (e.g., ["list", "get", "create"])
parameters: Dict mapping action -> list of parameter dicts

Example: entities = connector.list_entities() for entity in entities: print(f"{entity['entity_name']}: {entity['available_actions']}")

read_skill_docs(self, section: str | None = None) ‑> str : Read this connector's usage docs, rendered to text.

Omit section for the outline and general guidance; pass an exact section id from the outline for full details. For local/offline connectors the full generated docs are returned and section is ignored.

Example: outline = await connector.read_skill_docs() details = await connector.read_skill_docs(section="entity:contacts")

Classes​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Methods​

Local mode (direct API calls)

Hosted mode with explicit connector_id (no lookup needed)

Hosted mode with lookup by workspace_name

Class variables​

Static methods​

Legacy: one generated-description tool​

Instance variables​

Methods​

Classes

Methods

Methods

Methods

Methods

Methods

Methods

Methods

Methods

Methods

Methods

Methods

Class variables

Static methods

Legacy: one generated-description tool

Instance variables

Methods