class langgraph_agent_toolkit.core.observability.langsmith.LangsmithObservability(remote_first=False)

Bases: BaseObservabilityPlatform

Langsmith implementation of observability platform.

Initialize LangsmithObservability.

Parameters:

remote_first (bool) – If True, prioritize remote prompts over local ones.

__init__(remote_first=False)

Initialize LangsmithObservability.

Parameters:

remote_first (bool) – If True, prioritize remote prompts over local ones.

property required_vars: List[str]

get_callback_handler(**kwargs)

Get the callback handler - LangSmith uses automatic tracing.

Return type:

None

before_shutdown()

Perform any necessary cleanup before shutdown.

Return type:

None

record_feedback(run_id, key, score, **kwargs)

Record feedback for a run to LangSmith.

Parameters:

• run_id (str)

• key (str)

• score (float)

Return type:

None

push_prompt(name, prompt_template, metadata=None, force_create_new_version=True)

Push a prompt to LangSmith.

Parameters:

• name (str) – Name of the prompt

• prompt_template (str | List[ChatMessageDict]) – The prompt template (string or list of message dicts)

• metadata (Dict[str, Any] | None) – Optional metadata (can include ‘model’ to create a chain)

• force_create_new_version (bool) – If True, always create a new version

Return type:

None

pull_prompt(name, template_format='f-string', **kwargs)

Pull a prompt from LangSmith.

Parameters:

• name (str) – Name of the prompt

• template_format (Literal['f-string', 'mustache', 'jinja2']) – Format for the template

• **kwargs – Additional parameters (e.g., version, label)

Returns:

ChatPromptTemplate

Return type:

ChatPromptTemplate | str | dict | None

delete_prompt(name)

Delete a prompt from LangSmith.

Parameters:

name (str) – Name of the prompt to delete

Return type:

None

async apull_prompt(name, template_format='f-string', **kwargs)

Async version of pull_prompt. Runs synchronous version in thread pool.

Parameters:

• name (str)

• template_format (Literal['f-string', 'mustache', 'jinja2'])

Return type:

ChatPromptTemplate | str | dict | None

property remote_first: bool

static requires_env_vars(func)

Validate environment variables before calling a method.

Parameters:

func (Callable[[...], T])

Return type:

Callable[[…], T]

trace_context(run_id, **kwargs)

Create a trace context for the execution.

Override in subclasses for platform-specific implementation.

Parameters:

• run_id (str) – The run ID to use as trace ID

• **kwargs – Additional context parameters (user_id, input, etc.)

Yields:

None (or platform-specific context object)

update_trace(trace, **attributes)

Set trace-level attributes (e.g. the final output) on a trace yielded by trace_context.

trace_context is entered before the run’s output is known, so the caller invokes this afterwards to record the output on the trace. No-op for platforms without a trace object; override per platform.

Parameters:

• trace (Any)

• attributes (Any)

Return type:

None

validate_environment()

Validate that required environment variables are set.

Return type:

bool