Commit Graph

15 Commits

Author SHA1 Message Date
SashaXser
f6de0928c4 Refractoring (#88)
Co-authored-by: João Moura <joaomdmoura@gmail.com>
2024-01-10 00:04:13 -03:00
João Moura
4b802a15c3 Improving agent delegation prompt 2024-01-07 21:35:27 -03:00
João Moura
2ef682edf3 Reliability improvements (#77)
* fixing identation for AgentTools
* updating gitignore to exclude quick test script
* startingprompt translation
* supporting individual task output
* adding agent to task output
* cutting new version
* Updating README example
2024-01-07 12:43:23 -03:00
João Moura
ca8c7266ed Tools cache and delegation improvements (#68)
* Fixing repeated tool usage treatment
* Improving agent delegation prompt
2024-01-06 11:46:34 -03:00
João Moura
13bb3abf57 Better agent execution error handling (#54)
A few quality of life improvements around cache handling and repeated tool usage
2024-01-05 11:04:59 -03:00
João Moura
5602160caf Refactoring task cache to be a tool (#50)
* 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.
2024-01-04 21:29:42 -03:00
Greyson LaLonde
ed61f467b1 Update to use absolute imports (#17)
Update to use absolute imports
2023-12-29 22:39:59 -03:00
João Moura
d3ecd1d490 Adding tool caching a loop execution prevention. (#25)
* 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.
2023-12-29 22:35:23 -03:00
Greyson LaLonde
d214100f0a Refactor Codebase to Use Pydantic v2 and Enhance Type Hints, Documentation (#24)
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.
2023-12-29 21:24:30 -03:00
Greyson Lalonde
1547986b69 Make tools a subpackage 2023-12-27 15:13:42 -05:00
Greyson Lalonde
73716f35fc Run pre-commit hooks
In the title !
2023-12-27 15:13:42 -05:00
João Moura
9e03b5d50e adding more specific guidelines to agent delegation tools 2023-12-25 11:13:46 -03:00
Joao Moura
cd30e2a5ba Making config optional with default value as it's WIP and Adding new treatment for wrong agent tool calls 2023-12-04 23:58:48 -08:00
Joao Moura
492f361634 adding extra detail on delegtion errors so LLM can recover 2023-11-15 00:41:17 -03:00
Joao Moura
6ff18128a7 Add support to delegate work 2023-11-10 18:15:45 -03:00