* patching for non-gpt model
* removal of json_object tool name assignment
* fixed issue for smaller models due to instructions prompt
* fixing for ollama llama3 models
* WIP: generated summary from documents split, could also create memgpt approach
* WIP: need tests but user inputted summarization strategy implemented - handling context window exceeding errors
* rm extra line
* removed type ignores
* added tests
* handling n to summarize prompt
* code cleanup, using click for cli asker
* rm not used class
* better refactor
* reverted poetry lock
* reverted poetry.locl
* improved context window exceeding exception class
* Performed spell check across the entire documentation
Thank you once again!
* Performed spell check across the most of code base
Folders been checked:
- agents
- cli
- memory
- project
- tasks
- telemetry
- tools
- translations
* Trying to add a max_token for the agents, so they limited by number of tokens.
* Performed spell check across the rest of code base, and enahnced the yaml paraser code a little
* Small change in the main agent doc
* Improve _save_file method to handle both dict and str inputs
- Add check for dict type input
- Use json.dump for dict serialization
- Convert non-dict inputs to string
- Remove type ignore comments
---------
Co-authored-by: João Moura <joaomdmoura@gmail.com>
* feat: add max retry limit to agent execution
* feat: add test to max retry limit feature
* feat: add code execution docstring
---------
Co-authored-by: João Moura <joaomdmoura@gmail.com>
* Cleaned up task execution to now have separate paths for async and sync execution. Updating all kickoff functions to return CrewOutput. WIP. Waiting for Joao feedback on async task execution with task_output
* Consistently storing async and sync output for context
* outline tests I need to create going forward
* Major rehaul of TaskOutput and CrewOutput. Updated all tests to work with new change. Need to add in a few final tricky async tests and add a few more to verify output types on TaskOutput and CrewOutput.
* Encountering issues with callback. Need to test on main. WIP
* working on tests. WIP
* WIP. Figuring out disconnect issue.
* Cleaned up logs now that I've isolated the issue to the LLM
* more wip.
* WIP. It looks like usage metrics has always been broken for async
* Update parent crew who is managing for_each loop
* Merge in main to bugfix/kickoff-for-each-usage-metrics
* Clean up code for review
* Add new tests
* Final cleanup. Ready for review.
* Moving copy functionality from Agent to BaseAgent
* Fix renaming issue
* Fix linting errors
* use BaseAgent instead of Agent where applicable
* Fixing missing function. Working on tests.
* WIP. Needing team to review change
* Fixing issues brought about by merge
* WIP
* Implement major fixes from yesterdays group conversation. Now working on tests.
* The majority of tasks are working now. Need to fix converter class
* Fix final failing test
* Fix linting and type-checker issues
* Add more tests to fully test CrewOutput and TaskOutput changes
* Add in validation for async cannot depend on other async tasks.
* Update validators and tests
* fix: call asserts
* fix: test_increment_tool_errors
* fix: test_increment_delegations_for_sequential_process
* fix: test_increment_delegations_for_hierarchical_process
* fix: test_code_execution_flag_adds_code_tool_upon_kickoff
* fix: test_tool_usage_information_is_appended_to_agent
* fix: try to fix test_crew_full_output
* fix: try to fix test_crew_full_output
* fix: test remove vcr to test crew_test test
* fix: comment test to see if ci passes
* fix: comment test to see if ci passes
* fix: test changing prompt tokens to get error on CI
* fix: test changing prompt tokens to get error on CI
* fix: test changing prompt tokens to get error on CI
* fix: test changing prompt tokens to get error on CI
* fix: test new approach
* fix: comment funciont not working in CI
* fix: github python version
* fix: remove need of vcr
* fix: fix and add comments for all type checking errors
* Adding support to force a tool return to be the final answer.
This will at the end of the execution return the tool output.
It will return the output of the latest tool with the flag
* Update src/crewai/agent.py
Co-authored-by: Gui Vieira <guilherme_vieira@me.com>
* Update tests/agent_test.py
Co-authored-by: Gui Vieira <guilherme_vieira@me.com>
---------
Co-authored-by: Gui Vieira <guilherme_vieira@me.com>
* WIP. Figuring out disconnect issue.
* Cleaned up logs now that I've isolated the issue to the LLM
* more wip.
* WIP. It looks like usage metrics has always been broken for async
* Update parent crew who is managing for_each loop
* Merge in main to bugfix/kickoff-for-each-usage-metrics
* Clean up code for review
* Add new tests
* Final cleanup. Ready for review.
* Moving copy functionality from Agent to BaseAgent
* Fix renaming issue
* Fix linting errors
* use BaseAgent instead of Agent where applicable
* better spacing
* works with llama index
* works on langchain custom just need delegation to work
* cleanup for custom_agent class
* works with different argument expectations for agent_executor
* cleanup for hierarchial process, better agent_executor args handler and added to the crew agent doc page
* removed code examples for langchain + llama index, added to docs instead
* added key output if return is not a str for and added some tests
* added hinting for CustomAgent class
* removed pass as it was not needed
* closer just need to figuire ou agentTools
* running agents - llamaindex and langchain with base agent
* some cleanup on baseAgent
* minimum for agent to run for base class and ensure it works with hierarchical process
* cleanup for original agent to take on BaseAgent class
* Agent takes on langchainagent and cleanup across
* token handling working for usage_metrics to continue working
* installed llama-index, updated docs and added better name
* fixed some type errors
* base agent holds token_process
* heirarchail process uses proper tools and no longer relies on hasattr for token_processes
* removal of test_custom_agent_executions
* this fixes copying agents
* leveraging an executor class for trigger llamaindex agent
* llama index now has ask_human
* executor mixins added
* added output converter base class
* type listed
* cleanup for output conversions and tokenprocess eliminated redundancy
* properly handling tokens
* simplified token calc handling
* original agent with base agent builder structure setup
* better docs
* no more llama-index dep
* cleaner docs
* test fixes
* poetry reverts and better docs
* base_agent_tools set for third party agents
* updated task and test fix
* feat: add training logic to agent and crew
* feat: add training logic to agent executor
* feat: add input parameter to cli command
* feat: add utilities for the training logic
* feat: polish code, logic and add private variables
* feat: add docstring and type hinting to executor
* feat: add constant file, add constant to code
* feat: fix name of training handler function
* feat: remove unused var
* feat: change file handler file name
* feat: Add training handler file, class and change on the code
* feat: fix name error from file
* fix: change import to adapt to logic
* feat: add training handler test
* feat: add tests for file and training_handler
* feat: add test for task evaluator function
* feat: change text to fit in-screen
* feat: add test for train function
* feat: add test for agent training_handler function
* feat: add test for agent._use_trained_data
* fix: 'from datetime import datetime for logging' to print the timestamp
* fix: correct default model (gpt-4o), correct token counts, and correct TaskOutput attributes (added agent)
* test: verify Task callback data is an instance of TaskOutput
* fix: fix test actually running
* fix: fix test to not send request to openai
* fix: fix linting to remove cli files
* fix: exclude only files that breaks black
* fix: Fix all Ruff checkings on the code and Fix Test with repeated name
* fix: Change linter name on yml file
* feat: update pre-commit
* feat: remove need for isort on the code
* feat: remove black linter
* feat: update tests yml to try to fix the tests gh action
* feature: human input per task
* Update executor.py
* Update executor.py
* Update executor.py
* Update executor.py
* Update executor.py
* feat: change human input for unit testing
added documentation and unit test
* Create test_agent_human_input.yaml
add yaml for test
---------
Co-authored-by: João Moura <joaomdmoura@gmail.com>
This now will allow to add a max_inter option to agents while also making sure to force the agent to give it's best final answer before running out of it's max_inter.
* Refactoring task cache to be a tool
The previous implementation of the task caching system was early exiting
the agent executor due to the fact it was returning an AgentFinish object.
This now refactors it to use a cache specific tool that is dynamically
added and forced into the agent in case of a task execution that was
already executed with the same input.
* Adding tool caching a loop execution prevention.
This adds some guardrails, to both prevent the same tool to be used
consecutively and also caching tool's results across the entire crew
so it cuts down execution time and eventual LLM calls.
This plays a huge role for smaller opensource models that usually fall
into those behaviors patterns.
It also includes some smaller improvements around the tool prompt and
agent tools, all with the same intention of guiding models into
better conform with agent instructions.
Update to Pydantic v2:
Transitioned all references from pydantic.v1 to pydantic (v2), ensuring compatibility with the latest Pydantic features and improvements.
Affected components include agent tools, prompts, crew, and task modules.
Refactoring & Alignment with Pydantic Standards:
Refactored the agent module away from traditional __init__ to align more closely with Pydantic best practices.
Updated the crew module to Pydantic v2 and enhanced configurations, allowing JSON and dictionary inputs. Additionally, some (not all) exceptions have been migrated to leverage Pydantic's error-handling capabilities.
Enhancements to Validators and Typings:
Improved validators and type annotations across multiple modules, enhancing code readability and maintainability.
Streamlined the validation process in line with Pydantic v2's methodologies.
Import and Configuration Adjustments:
Updated to test-related absolute imports due to issues with Pytest finding packages through relative imports.
