Issues

• 🐛 Bug Report: Bedrock invoke_model_with_response_stream records streamed tool-call arguments as {} (input_json_delta appended to the dict placeholder, error swallowed)

  traceloop/openllmetry#4558 · Filter by author armanunix opened on Oct 5, 2026
  2

• 🐛 Bug Report: Bedrock instrumentation does not record top_k / stop_sequences, and records no request parameters at all when the call fails

  traceloop/openllmetry#4557 · Filter by author armanunix opened on Oct 5, 2026

• 🐛 Bug Report: Bedrock instrumentation: converse_stream and invoke_model_with_response_stream spans are never ended on early exit (break, close(), abandoned stream)

  traceloop/openllmetry#4556 · Filter by author armanunix opened on Oct 5, 2026
  1

• 🐛 Bug Report: Google GenAI instrumentation: generate_content_stream spans are never ended on early exit (break, close(), abandoned stream)

  traceloop/openllmetry#4555 · Filter by author armanunix opened on Oct 4, 2026

• 🐛 Bug Report: Google GenAI streaming records only the last chunk as the output (earlier text is dropped from gen_ai.output.messages)

  traceloop/openllmetry#4554 · Filter by author armanunix opened on Oct 4, 2026

• 🐛 Bug Report: Google GenAI instrumentation records no request parameters (temperature, top_p, top_k, max_tokens, seed, stop_sequences) because it reads them from top-level kwargs instead of config; system instruction missing for dict config

  traceloop/openllmetry#4553 · Filter by author armanunix opened on Oct 4, 2026

• 🐛 Bug Report: OpenAI instrumentation uninstrument() is a no-op: unwrap() is called with "module", "Class.method", so nothing is unwrapped; re-instrumenting doubles spans and can leave spans that never end

  traceloop/openllmetry#4552 · Filter by author armanunix opened on Oct 4, 2026

• 🐛 Bug Report: async with_streaming_response records no response (parse() coroutine never awaited) in the OpenAI and Anthropic instrumentations

  traceloop/openllmetry#4551 · Filter by author armanunix opened on Oct 4, 2026

• 🚀 Feature: Support for deepagents

  traceloop/openllmetry#4550 · Filter by author AbdealiLoKo opened on Oct 4, 2026

• 🐛 Bug Report:OpenAI instrumentation does not record gen_ai.request.seed, gen_ai.request.stop_sequences or gen_ai.output.type

  traceloop/openllmetry#4538 · Filter by author armanunix opened on Oct 1, 2026
  1

• 🐛 Bug Report: penAI instrumentation consumes generator/iterator messages and tools, so the model receives an empty conversation (default settings)

  traceloop/openllmetry#4537 · Filter by author armanunix opened on Oct 1, 2026
  3

• 🐛 Bug Report: Anthropic instrumentation: streaming spans are never ended on early exit (with-block, messages.stream() helper, close)

  traceloop/openllmetry#4536 · Filter by author armanunix opened on Oct 1, 2026
  1

• 🐛 Bug Report: OpenAI instrumentation: ChatStream does not end the span on close() (sync or async)

  traceloop/openllmetry#4532 · Filter by author armanunix opened on Sep 30, 2026
  2

• 🐛 Bug Report: mistralai event mode loses prompt events when a ToolMessage is present

  traceloop/openllmetry#4523 · Filter by author chrikrah opened on Sep 28, 2026

• mcp: nested async with Client leaks the session span context

  traceloop/openllmetry#4519 · Filter by author doronkopit5 opened on Sep 27, 2026

• mcp: @dont_throw swallows Client.__aenter__/__aexit__ errors (async with gives None)

  traceloop/openllmetry#4518 · Filter by author doronkopit5 opened on Sep 27, 2026

• Is openai the right name for every host that speaks the same API?

  traceloop/openllmetry#4513 · Filter by author TheCheetah11 opened on Sep 25, 2026

• fix(openai): @dont_throw is a no-op on Responses API wrappers, so tracing bugs crash host apps

  bugpython
  traceloop/openllmetry#4505 · Filter by author doronkopit5 opened on Sep 24, 2026
  4

• feat(openai): trace responses.with_raw_response.create(stream=True) instead of skipping it

  enhancementpython
  traceloop/openllmetry#4504 · Filter by author doronkopit5 opened on Sep 24, 2026

• 🐛 Bug Report: Bedrock gen_ai.client.operation.duration measures time since the client was created, not the call

  traceloop/openllmetry#4502 · Filter by author kimnamu opened on Sep 24, 2026
  1

• 🐛 Bug Report: Bedrock profile IDs with a dot in the model name (us.openai.gpt-5.6-sol) are recorded as model 'gpt-5'

  traceloop/openllmetry#4501 · Filter by author kimnamu opened on Sep 24, 2026

• 🐛 Bug Report: Bedrock converse reasoning blocks are recorded as JSON-dumped text (GPT-6 Sol/Luna/Astra redactedContent shows up as bytes)

  traceloop/openllmetry#4499 · Filter by author kimnamu opened on Sep 24, 2026

• 🐛 Bug Report: Bedrock converse token and duration metrics are labeled with no model or the previous invoke_model's model

  traceloop/openllmetry#4497 · Filter by author kimnamu opened on Sep 24, 2026

• 🐛 Bug Report: Bedrock invoke_model spans for OpenAI models (GPT-6 Sol/Luna/Astra, gpt-oss) have no token usage, finish reason or messages

  traceloop/openllmetry#4495 · Filter by author kimnamu opened on Sep 24, 2026

• 🐛 Bug Report: gen_ai.bedrock.guardrail.activation counts every Bedrock call

  traceloop/openllmetry#4471 · Filter by author AnoshanJ opened on Sep 15, 2026