• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested adding a Pull Request to address the issue. The change allows for passing an instance of an LLM class to execute the evaluations.

• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested that the issue may stem from the use of \n characters. To resolve this, they recommend replacing \n with commas or utilizing triple quotation marks with """ instead of single quotation marks "". They believe that either solution should resolve the problem.

• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested setting the CREWAI_STORAGE_DIR environment variable to specify the directory where crew instances and memories should be stored. This allows users to easily change the storage location for their crew-related data.

• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested that there might be an error in the way arguments are being passed to the tool. To troubleshoot this, they have requested the entire code to be shared for further analysis.

• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested that the complete code for the project should be shared here for easy reference and to provide a comprehensive understanding of the project. This will allow users to quickly grasp the technical details and functionality of the project without having to search for the code in different

• Commented in [crewAIInc/crewAI] on 2025-02-09.

  AI Summary: @Vidit-Ostwal has suggested exploring alternative options to Serper.dev. They recommend considering using SerpAPI or Google Search API. SerpAPI provides a user-friendly interface and quick response times, while Google Search API offers more advanced features and customization options. They also mention that the choice between the two depends on the specific

• Commented in [crewAIInc/crewAI] on 2025-02-08.

  AI Summary: @Vidit-Ostwal has suggested closing the issue if the above PR resolves it, which @heckfy88 can verify.

• Commented in [crewAIInc/crewAI] on 2025-02-08.

  AI Summary: @Vidit-Ostwal suggests that when yielding a value, it is converted to a generator object. To use the actual value, it needs to be iterated over. @Vidit-Ostwal demonstrates this using an example code that uses yield to pass a random city name to a subsequent step in a flow. The use

• Commented in [crewAIInc/crewAI] on 2025-02-07.

  AI Summary: @Vidit-Ostwal has suggested that the issue may have been resolved. They recommend updating the crew version and referenced PR #2055 for further details.

• Commented in [crewAIInc/crewAI] on 2025-02-07.

  AI Summary: @Vidit-Ostwal suggests incorporating built-in evaluations for CrewAI to assess tool outputs, agent performance, and task results against specific criteria like factual accuracy. They cite the work done by RAGAS in structured evaluations for retrieval-augmented generation as an example and express curiosity about the possibility of implementing a similar approach within

• Raised an issue in [langfuse/langfuse]: bug: Running Langfuse Locally Dcoker (2025-02-08).

  AI Summary: @Vidit-Ostwal has suggested that the issue with setting up Langfuse via Docker Compose is due to a conflict with an existing process using port 9000. While the documentation suggests changing the port, @Vidit-Ostwal believes there may be additional internal references to port 9000 that need to be updated. They are

• Raised an issue in [crewAIInc/crewAI]: Documentation Error in memory docs (2025-02-04).

  AI Summary: @Vidit-Ostwal has suggested that in the code examples provided in the CrewAI documentation for using embeddings with different providers (Azure OpenAI, Google AI, Vertex AI, Cohere), the model_name parameter in the embedder function should be replaced with model. This is because the embedder function currently takes model as input instead

• Raised an issue in [crewAIInc/crewAI]: output_log_file should also suppport JSON format. (2025-01-27).

  AI Summary: @Vidit-Ostwal has suggested a feature to add JSON format support to the output_log_file, which is currently in .txt format. This will enhance the file's parsability and aid in further analysis. The suggestion is partially related to issue #1970. @Vidit-Ostwal has expressed willingness to contribute a pull request to implement this

• Raised an issue in [browser-use/browser-use]: Use of use_vision while defining agent (2025-01-26).

  AI Summary: @Vidit-Ostwal has suggested clarifying the impact of setting use_vision to True when defining an agent. They question whether the entire process falls back to scraping elements or if the screenshot is parsed by the multimodal LLM. They have observed that the only noticeable difference is that a .gif is saved

• Opened a PR in [crewAIInc/crewAI]: Added functionality to have any llm run test functionality (2025-02-09).

  AI Summary: @Vidit-Ostwal has suggested a fix for an issue detailed in #2067. Although the specific technical details of the fix have not been provided in the comment, it is implied that this change addresses the problem outlined in the referenced issue.

• Opened a PR in [crewAIInc/crewAI]: Added reset memories function inside crew class (2025-02-06).

  *AI Summary: @Vidit-Ostwal has suggested fixing the reset memories issue mentioned in #2023. Other proposed changes include:

• Removing the CLI command to reset memories

• Making necessary documentation changes

@Vidit-Ostwal can implement these changes if the proposed solution is in line with the project's goals.*

• Opened a PR in [crewAIInc/crewAI]: Fixed the memory documentation (2025-02-04).

  AI Summary: @Vidit-Ostwal has suggested a fix for the documentation issue raised in issue #2030. The comment does not provide any specific details about the changes made to the documentation, but it implies that the changes address the issue reported in the linked issue.

• Opened a PR in [crewAIInc/crewAI]: Added functionality to have json format as well for the logs (2025-01-27).

  AI Summary: @Vidit-Ostwal has suggested adding support to save logs as a JSON file when initializing the crew. Users can now set output_log_file to True or provide a file name and enable save_as_json to True. The resulting JSON file will contain an array of JSON events, simplifying parsing and further analysis of

• Opened a PR in [explodinggradients/ragas]: Changed the parse_run_traces to include last 4 letters of run_id (2025-01-25).

  AI Summary: @Vidit-Ostwal has suggested changes to the parse_run_traces function. They have modified it to include the last four characters of the run_id, resulting in a unique key for every call. This change aims to improve the identification and tracking of trace calls, ensuring reliable and distinct identification of each call instance.

• Starred sentient-engineering/sentient on 2025-02-08.
• Starred langflow-ai/langflow on 2025-02-07.
• Starred langfuse/langfuse on 2025-02-04.
• Starred microsoft/autogen on 2025-01-31.
• Starred browser-use/browser-use on 2025-01-26.

• Forked crewAIInc/crewAI on 2025-01-27.
• Forked explodinggradients/ragas on 2025-01-24.