Skip to content

fix(openai-agents): span attribute handling for tool calls and results #3422

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

Merged
doronkopit5 merged 4 commits into from
Oct 29, 2025
Merged

fix(openai-agents): span attribute handling for tool calls and results #3422

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2ba1445
fix(openai-agents): span attribute handling for tool calls and results
doronkopit5 Oct 26, 2025
b0d7b2b
lint
doronkopit5 Oct 26, 2025
adbc139
add tool call ID to span attributes
doronkopit5 Oct 26, 2025
949e3fa
Merge branch 'main' into dk/openai-agents-tool-responses
nirga Oct 29, 2025
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
114 changes: 103 additions & 11 deletions ...metry-instrumentation-openai-agents/opentelemetry/instrumentation/openai_agents/_hooks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -239,19 +239,111 @@ def on_span_end(self, span):
input_data = getattr(span_data, 'input', [])
if input_data:
for i, message in enumerate(input_data):
if hasattr(message, 'role') and hasattr(message, 'content'):
otel_span.set_attribute(f"{SpanAttributes.LLM_PROMPTS}.{i}.role", message.role)
content = message.content
prefix = f"{SpanAttributes.LLM_PROMPTS}.{i}"

# Convert message to dict for unified handling
Copy link
Copy Markdown
Contributor

@ellipsis-dev ellipsis-dev Bot Oct 26, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Consider extracting the object-to-dict conversion (lines 244–253) into a helper function to improve readability and reduce duplication. Using something like vars(message) or a dedicated converter might simplify the logic.

if isinstance(message, dict):
msg = message
else:
# Convert object to dict
msg = {}
for attr in [
"role",
"content",
"tool_call_id",
"tool_calls",
"type",
"name",
"arguments",
"call_id",
"output",
]:
if hasattr(message, attr):
msg[attr] = getattr(message, attr)

# Determine message format and extract data
role = None
content = None
tool_call_id = None
tool_calls = None

if 'role' in msg:
# Standard OpenAI chat format
role = msg['role']
content = msg.get('content')
tool_call_id = msg.get('tool_call_id')
tool_calls = msg.get('tool_calls')
elif 'type' in msg:
# OpenAI Agents SDK format
msg_type = msg['type']
if msg_type == 'function_call':
# Tool calls are assistant messages
role = 'assistant'
# Create tool_calls structure matching OpenAI SDK format
tool_calls = [{
'id': msg.get('id', ''),
'name': msg.get('name', ''),
'arguments': msg.get('arguments', '')
}]
Comment thread
doronkopit5 marked this conversation as resolved.
elif msg_type == 'function_call_output':
# Tool outputs are tool messages
role = 'tool'
content = msg.get('output')
tool_call_id = msg.get('call_id')

# Set role attribute
if role:
otel_span.set_attribute(f"{prefix}.role", role)

# Set content attribute
if content is not None:
if not isinstance(content, str):
content = json.dumps(content)
otel_span.set_attribute(f"{SpanAttributes.LLM_PROMPTS}.{i}.content", content)
elif isinstance(message, dict):
if 'role' in message and 'content' in message:
otel_span.set_attribute(f"{SpanAttributes.LLM_PROMPTS}.{i}.role", message['role'])
content = message['content']
if isinstance(content, dict):
content = json.dumps(content)
otel_span.set_attribute(f"{SpanAttributes.LLM_PROMPTS}.{i}.content", content)
otel_span.set_attribute(f"{prefix}.content", content)

# Set tool_call_id for tool result messages
if tool_call_id:
otel_span.set_attribute(f"{prefix}.tool_call_id", tool_call_id)

# Set tool_calls for assistant messages with tool calls
if tool_calls:
for j, tool_call in enumerate(tool_calls):
# Convert to dict if needed
if not isinstance(tool_call, dict):
tc_dict = {}
if hasattr(tool_call, 'id'):
tc_dict['id'] = tool_call.id
if hasattr(tool_call, 'function'):
func = tool_call.function
if hasattr(func, 'name'):
tc_dict['name'] = func.name
if hasattr(func, 'arguments'):
tc_dict['arguments'] = func.arguments
elif hasattr(tool_call, 'name'):
tc_dict['name'] = tool_call.name
if hasattr(tool_call, 'arguments'):
tc_dict['arguments'] = tool_call.arguments
tool_call = tc_dict

# Extract function details if nested (standard OpenAI format)
if 'function' in tool_call:
function = tool_call['function']
tool_call = {
'id': tool_call.get('id'),
'name': function.get('name'),
'arguments': function.get('arguments')
}

# Set tool call attributes
if tool_call.get('id'):
otel_span.set_attribute(f"{prefix}.tool_calls.{j}.id", tool_call['id'])
if tool_call.get('name'):
otel_span.set_attribute(f"{prefix}.tool_calls.{j}.name", tool_call['name'])
if tool_call.get('arguments'):
args = tool_call['arguments']
if not isinstance(args, str):
args = json.dumps(args)
otel_span.set_attribute(f"{prefix}.tool_calls.{j}.arguments", args)

# Add function/tool specifications to the request using OpenAI semantic conventions
response = getattr(span_data, 'response', None)
Expand Down
Loading