Automate browser based workflows with AI
Python aiapiautomationbrowserbrowser-automationcomputergptllmplaywrightpowerautomatepuppeteerpythonrpaseleniumvisionworkflow
chore(observability): trim high-volume production log noise (#6406)
Shuchang Zheng85a3885ab85c
diff --git a/skyvern/forge/agent.py b/skyvern/forge/agent.pyindex dd5f8a7bc..457d972fb 100644--- a/skyvern/forge/agent.py+++ b/skyvern/forge/agent.py@@ -3730,7 +3730,7 @@ class ForgeAgent:cache_enabled = prompt_caching_settings.get(EXTRACT_ACTION_PROMPT_NAME) or prompt_caching_settings.get(EXTRACT_ACTION_TEMPLATE)- LOG.info(+ LOG.debug("Extract-action prompt caching evaluation",template=template,cache_enabled=cache_enabled,@@ -4812,12 +4812,6 @@ class ForgeAgent:Note: This should only be called when verification is needed (i.e., whenthe standard flow would have called check_user_goal_complete in agent_step)."""- LOG.info(- "Starting parallel user goal verification with speculative extract-actions",- step_id=step.step_id,- task_id=task.task_id,- )-verification_task = asyncio.create_task(self.check_user_goal_complete(page=page,diff --git a/skyvern/forge/sdk/api/llm/vertex_cache_manager.py b/skyvern/forge/sdk/api/llm/vertex_cache_manager.pyindex a85019498..8d0bc90fe 100644--- a/skyvern/forge/sdk/api/llm/vertex_cache_manager.py+++ b/skyvern/forge/sdk/api/llm/vertex_cache_manager.py@@ -106,10 +106,8 @@ class VertexCacheManager:# Check if still validexpire_time = datetime.fromisoformat(cache_data["expireTime"].replace("Z", "+00:00"))if expire_time > datetime.now(expire_time.tzinfo):- LOG.info("Reusing existing cache", cache_key=cache_key, cache_name=cache_data["name"])return cache_dataelse:- LOG.info("Cache expired, creating new one", cache_key=cache_key)# Clean up expired cachetry:self.delete_cache(cache_key)diff --git a/skyvern/forge/sdk/experimentation/llm_prompt_config.py b/skyvern/forge/sdk/experimentation/llm_prompt_config.pyindex fde1f071b..ff2d5702f 100644--- a/skyvern/forge/sdk/experimentation/llm_prompt_config.py+++ b/skyvern/forge/sdk/experimentation/llm_prompt_config.py@@ -73,13 +73,6 @@ async def get_llm_handler_for_prompt_type(llm_config_name = config[prompt_type]try:handler = LLMAPIHandlerFactory.get_llm_api_handler(llm_config_name)- LOG.info(- "Using LLM handler for prompt type from experiment",- prompt_type=prompt_type,- llm_config_name=llm_config_name,- distinct_id=distinct_id,- organization_id=organization_id,- )return handlerexcept Exception:LOG.error(