- Completed the implementation of the P3 Run phase, integrating task execution and validation mechanisms.
- Introduced a new `RunConfig` struct to manage execution parameters such as retries and validation thresholds.
- Developed a two-layer validation system using the new `yao/assert` package, supporting both natural language and structured JSON rules.
- Enhanced the `RunExecution` method to execute tasks sequentially with progress tracking and a retry mechanism for validation failures.
- Updated task structures to include comprehensive validation rules and expected outputs, ensuring robust task management.
- Added unit tests for the new execution and validation features, achieving high test coverage across the implementation.
- Revised documentation to reflect changes in the architecture and functionality of the P3 phase.
- Introduced a structured Goals type with delivery metadata for improved task management and output handling.
- Updated the DeliveryTarget type to include additional fields for recipients, format, and options, enhancing flexibility in result delivery.
- Enhanced the Executor's RunGoals method to parse and handle delivery information from agent responses.
- Revised input formatting to include robot identity context, improving clarity in generated goals.
- Updated tests to validate the new structure and ensure comprehensive coverage of delivery functionalities.
- Introduced multiple executor modes (Standard, DryRun, Sandbox) to accommodate various use cases, enhancing flexibility in execution strategies.
- Updated DESIGN.md to reflect the new executor modes and their respective use cases, including detailed descriptions and configuration examples.
- Revised TECHNICAL.md to outline the new executor package structure, emphasizing the modular design for future enhancements.
- Enhanced the TODO.md to track the progress of executor mode implementations and related tasks.
- Removed outdated executor stub files and tests, streamlining the codebase for improved maintainability.
- Updated integration tests to utilize the new DryRun executor, ensuring comprehensive coverage of execution scenarios without real agent calls.
- Integrated trigger handling directly into the Manager, consolidating the logic for clock, human intervention, and event triggers.
- Updated the Manager to include methods for processing human interventions and event triggers, ensuring robust execution control.
- Refactored the trigger package to provide validation and utility functions, enhancing the overall structure and clarity of trigger-related logic.
- Improved documentation and comments throughout the Manager and trigger implementations for better understanding and maintainability.
- Updated tests to cover new functionalities, ensuring comprehensive validation of the Manager's behavior with various trigger types.
- Updated the Task struct in DESIGN.md to replace the Input field with Messages, allowing for rich content input (text, images, files, audio).
- Revised the TriggerRequest struct in TECHNICAL.md to utilize Messages instead of Input, enhancing the flexibility of user interactions.
- Improved related code examples to demonstrate the new Messages structure, ensuring consistency and clarity in task management and execution processes.
- Updated the TriggerRequest and InterveneRequest structs to replace the Description field with an Input field, enhancing clarity on user input for task actions.
- Revised related code examples to reflect the new Input field, ensuring consistency across the implementation.
- Improved documentation comments for better understanding of the changes and their impact on task management.
- Revised the Task struct in DESIGN.md to include an Input field for natural language descriptions and a Source field to indicate task origin (auto, human, event).
- Updated the Execution struct in TECHNICAL.md to add an Input field for storing original trigger input, improving traceability of task execution.
- Enhanced documentation for both structs to clarify their roles and improve overall understanding of the agent's task management and execution processes.
- Updated DESIGN.md to improve the structure of the InspirationReport and Goals, clarifying their roles in the agent's execution process.
- Revised the Task struct to include a reference to the goal in markdown, enhancing task management.
- Enhanced TECHNICAL.md by introducing new types for RobotStatus, ClockMode, and EventSource, improving type safety and clarity.
- Updated various structs and interfaces to reflect new types and improve documentation consistency across the codebase.
- Updated the DESIGN.md and TECHNICAL.md files to consolidate the knowledge base (KB) and database (DB) configurations into a unified structure, enhancing clarity on shared resources.
- Removed the PrivateKB and SharedKB fields, replacing them with a single KB field that includes collections and options for better organization.
- Introduced a new DB struct to define shared database models, streamlining the configuration process.
- Enhanced the Learn struct to clarify its purpose for the robot's private knowledge base, including comments for better understanding.
- Removed the Monitor and related types from the configuration, simplifying the overall structure and focusing on essential components.
- Updated the DESIGN.md to clarify the relationship between robots and executions, emphasizing that each trigger creates a new execution mapped to a job.Job for monitoring.
- Revised the RobotState struct to include fields for tracking multiple running executions and their IDs, improving concurrency management.
- Enhanced the TECHNICAL.md to reflect the global robot object for static methods, streamlining the API usage for robot management.
- Improved TypeScript interfaces to align with the new execution tracking structure, ensuring consistency across documentation.
- Added detailed comments and examples for job creation and execution handling, clarifying the integration of the job system within the robot's functionality.
- Updated the TriggerRequest struct to include new fields: InsertAt and AtIndex for improved task insertion control.
- Revised action descriptions for clarity, changing terms to better reflect their functionality.
- Added detailed comments and examples for the robot.Trigger method to clarify usage for human and event triggers.
- Introduced new types and constants for task insertion positions and sources, enhancing the overall structure and readability of the documentation.
- Deleted the DESIGN.md and TECHNICAL.md files to streamline the documentation structure and eliminate redundancy.
- This change reflects a shift towards a more concise documentation approach, focusing on essential information and reducing clutter for better usability.
2026-01-14 10:46:48 +08:00
Renamed from agent/autonomous/TECHNICAL.md (Browse further)
