Skip to content

feat(langchain): Change LLM span operation to generate_text #5705

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ericapisani wants to merge 3 commits into
base: master
Choose a base branch
from
+17 −7
Open

feat(langchain): Change LLM span operation to generate_text #5705

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
bce8be6
feat(langchain): Change LLM span operation to generate_text and captu…
ericapisani Mar 19, 2026
c4d5f03
ref(langchain): Remove redundant docstring from on_llm_start
ericapisani Mar 19, 2026
30e22ba
ref(langchain): Remove agent name from on_llm_start and clean up tests
ericapisani Mar 20, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 8 additions & 3 deletions sentry_sdk/integrations/langchain.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -351,7 +351,6 @@
metadata: "Optional[Dict[str, Any]]" = None,
**kwargs: "Any",
) -> "Any":
"""Run when LLM starts running."""
with capture_internal_exceptions():
if not run_id:
return
Expand All @@ -369,12 +368,18 @@
watched_span = self._create_span(
run_id,
parent_run_id,
op=OP.GEN_AI_PIPELINE,
name=kwargs.get("name") or "Langchain LLM call",
op=OP.GEN_AI_GENERATE_TEXT,
name=f"generate_text {model}".strip(),
origin=LangchainIntegration.origin,
)
span = watched_span.span

span.set_data(SPANDATA.GEN_AI_OPERATION_NAME, "generate_text")

pipeline_name = kwargs.get("name")
if pipeline_name:
span.set_data(SPANDATA.GEN_AI_PIPELINE_NAME, pipeline_name)

Check warning on line 381 in sentry_sdk/integrations/langchain.py

View check run for this annotation

@sentry/warden / warden: code-review

GEN_AI_AGENT_NAME not captured in on_llm_start despite PR description 

The PR description states that "when an LLM is invoked within an agent context, the agent name is now captured on the span via GEN_AI_AGENT_NAME", but the `on_llm_start` method does not call `_get_current_agent()` or set `SPANDATA.GEN_AI_AGENT_NAME`. The parallel method `on_chat_model_start` (lines 463-465) does capture the agent name correctly. This means LLM calls via `on_llm_start` will not have agent context attached, reducing observability as promised.

if model:
span.set_data(
SPANDATA.GEN_AI_REQUEST_MODEL,
Expand Down
13 changes: 9 additions & 4 deletions tests/integrations/langchain/test_langchain.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -851,12 +851,15 @@ def test_langchain_integration_with_langchain_core_only(sentry_init, capture_eve
assert tx["type"] == "transaction"

llm_spans = [
span for span in tx.get("spans", []) if span.get("op") == "gen_ai.pipeline"
span
for span in tx.get("spans", [])
if span.get("op") == "gen_ai.generate_text"
]
assert len(llm_spans) > 0

llm_span = llm_spans[0]
assert llm_span["description"] == "Langchain LLM call"
assert llm_span["description"] == "generate_text gpt-3.5-turbo"
assert llm_span["data"]["gen_ai.operation.name"] == "generate_text"
assert llm_span["data"]["gen_ai.request.model"] == "gpt-3.5-turbo"
assert (
llm_span["data"]["gen_ai.response.text"]
Expand Down Expand Up @@ -1062,11 +1065,12 @@ def test_langchain_message_truncation(sentry_init, capture_events):
assert tx["type"] == "transaction"

llm_spans = [
span for span in tx.get("spans", []) if span.get("op") == "gen_ai.pipeline"
span for span in tx.get("spans", []) if span.get("op") == "gen_ai.generate_text"
]
assert len(llm_spans) > 0

llm_span = llm_spans[0]
assert llm_span["data"]["gen_ai.operation.name"] == "generate_text"
assert SPANDATA.GEN_AI_REQUEST_MESSAGES in llm_span["data"]

messages_data = llm_span["data"][SPANDATA.GEN_AI_REQUEST_MESSAGES]
Expand Down Expand Up @@ -1776,11 +1780,12 @@ def test_langchain_response_model_extraction(
assert tx["type"] == "transaction"

llm_spans = [
span for span in tx.get("spans", []) if span.get("op") == "gen_ai.pipeline"
span for span in tx.get("spans", []) if span.get("op") == "gen_ai.generate_text"
]
assert len(llm_spans) > 0

llm_span = llm_spans[0]
assert llm_span["data"]["gen_ai.operation.name"] == "generate_text"

if expected_model is not None:
assert SPANDATA.GEN_AI_RESPONSE_MODEL in llm_span["data"]
Expand Down
Loading