Roo Code Memory Bank Changelog
{v0.4.0} - Switch to Manual Installation
- Manual installation is now recommended, in order to avoid having a .roo/ directory in the workspace root.
- Test mode has been eliminated.
- Custom mode interaction rules are now eliminated, letting Roo Code handle that.
{v0.3.6} - Bug Fix
Revised wording in memory bank section of custom instructions so that Roo doesn't display its tool calls in the reply.
{v0.3.5} - .roorules
- The custom instructions files for Roo Code are now named .roorules-[mode]
- The
curl
commands to download the install scripts for Roo Code Memory Bank now also run the script.
{v0.3.4} - Follow- Up after Initializing Memory Bank
- I added the
ask_followup_question
tool after memory bank initialization, so that several clickable, suggested task options are presented.
{v0.3.2] - Install Scripts!
New installation script for Windows and Linux/macOS
[v0.3.0] - Improvements to custom instructions
- I've applied what I've learned from RooFlow to the .clinerules-[mode] files!
- This might be the final update to Roo Code Memory Bank, it's been a wild ride!
[v0.2.7] - New Test Mode
Key Changes:
- Added Test mode configuration (.clinerules-test)
- Added Test mode to .roomodes
- Configured Test mode permissions and capabilities
- Set up Test mode collaboration workflows
- Defined Test mode triggers for mode switching
[v0.2.5] - Bug Fixes
Key Changes:
- Restored "Update Memory Bank/UMB" functionality to be available along with real-time memory bank updates.
[v0.2.1] - Autonomous Task Management and Builder Mode Integration
Key Changes:
- Integrated Builder Mode Concepts:
- Incorporated elements of the proposed "Builder" mode into the existing Architect and Code modes.
- Enhanced Architect mode's project initialization process with
projectBrief.md
andprojectConfig.md
for structured project setup. - Added specific guidance on task breakdown within the Memory Bank (using
progress.md
or a dedicated task file), including task name, status, dependencies, and detailed scope. - Enabled Code mode to autonomously execute tasks defined in the Memory Bank, including pre- and post-task actions (reading relevant files, updating task status).
- Reinforced the principle of strict adherence to task scope in Code mode.
- Added
switch_mode
capability to both modes.
Benefits:
- More structured and autonomous project development workflow.
- Improved task management and tracking within the Memory Bank.
- Clearer separation of responsibilities between Architect (planning) and Code (implementation).
- Enhanced ability for the AI to work independently on well-defined tasks.
- Greater flexibility while maintaining a structured approach.
Credits:
- Inspired by concepts from the proposed "Builder" mode, with contributions from GitHub user shariqriazz
[v0.2.0] - Memory Bank and Mode Enhancements
Key Changes:
-
Refined Memory Bank System:
- Streamlined Memory Bank initialization and loading across all modes.
- Removed redundant Memory Bank update logic from individual modes, centralizing it in the system prompt.
- Clarified the roles of each mode in relation to Memory Bank management (creation, updating, usage).
- Improved real-time update triggers and targets for better responsiveness.
-
Enhanced Mode Interactions:
- Implemented consistent and comprehensive
mode_triggers
in all modes (Architect, Code, Ask, Debug). - Added specific conditions for mode switching, making transitions more context-aware and accurate.
- Clarified mode collaboration responsibilities, especially regarding delegation of tasks (e.g., Architect to Code, Debug to Code).
- Revised Ask mode to be a general-purpose knowledge assistant, using the Memory Bank when active but not managing it.
- Implemented consistent and comprehensive
-
Improved System Robustness:
- Removed redundant instructions and configurations, reducing token usage and potential conflicts.
- Standardized on
apply_diff
for Memory Bank updates (where appropriate), improving consistency. - Clarified tool usage within each mode, emphasizing read-only tools for Debug mode.
- Added explicit file access restrictions for Architect mode (limited to
.md
files).
-
Revised Custom Instructions:
- Updated all 4
.clinerules-[mode].txt
files to reflect these changes
- Updated all 4
Benefits:
- More reliable and predictable Memory Bank behavior.
- Smoother and more accurate transitions between modes.
- Clearer separation of concerns between modes.
- Reduced risk of accidental data corruption or inconsistencies.
- Improved overall system efficiency and maintainability.
[v0.1.6]
Key Changes:
-
Added comprehensive Memory Bank integration and handling
-
Implemented real-time update triggers for debugging events
-
Added structured mode-specific instructions and rules
-
Defined file authority and cross-mode collaboration guidelines
-
Integrated error handling and diagnostic procedures
-
Standardized system prompt format to match other modes
-
The enhanced Debug Mode now provides:
-
Systematic problem analysis and diagnosis
-
Root cause identification with evidence-based verification
-
Actionable solution proposals with risk assessments
-
Memory Bank integration for debugging context persistence
-
Clear mode transition triggers for implementation handoff
-
Thanks @TheRealAlexV !
[v1.0.5]
Changes
-
Added Debug Mode to developer-primer.md:
-
Updated Core Components flowchart to include Debug Mode
-
Added Debug Mode to Mode Types section with detailed capabilities
-
Updated mode switching diagram with Debug Mode relationships
-
Added Debug Mode intent triggers for automatic switching
-
Added .clinerules-debug file reference and configuration details
-
Updated File Organization section to include Debug Mode files
-
Created .roomodes configuration file:
-
Implemented Debug Mode with read and command access permissions
-
Added comprehensive role definition for debugging capabilities
-
Configured proper tool access restrictions for safe system analysis
-
Updated README.md:
-
Added Debug Mode to Key Components diagram
-
Updated Quick Start guide to include Debug Mode
-
Added Debug Mode to file organization section
-
Created a new Debug Mode section explaining its implementation
-
Updated Smart Workflows diagram to show Debug Mode integration
-
Benefits
-
Provides specialized troubleshooting capabilities with surgical precision
-
Maintains system integrity through read-only access during investigation
-
Enables systematic problem analysis with command execution for diagnostics
-
Completes the mode ecosystem with a dedicated debugging environment
-
Improves user experience by separating investigation from implementation
-
Thanks @TheRealAlexV !