# Tachibot-MCP Modes Documentation ## Overview Tachibot-MCP provides intelligent orchestration of multiple AI models through specialized workflows. Each mode is designed for specific types of reasoning tasks and can be invoked using primary names or aliases. ## Available Modes ### 🎨 Brainstorm/Creative/Ideate Mode **Purpose**: Generate innovative ideas and explore creative possibilities **Tool Flow**: ``` 1. gemini_brainstorm - Initial creative exploration 2. openai_brainstorm - Alternative perspectives 3. perplexity_research - Real-world validation (optional) 4. think - Reflection and pattern recognition 5. openai_reason - Feasibility analysis ``` **Use Cases**: - Product ideation - Creative problem-solving - Feature brainstorming - Marketing campaigns - Innovation workshops **Example**: ```bash focus --mode brainstorm "Design a new mobile app for elderly users" focus --mode creative "Alternative uses for blockchain technology" focus --mode ideate "Sustainable packaging solutions" ``` ### 🔬 Research/Investigate Mode **Purpose**: Deep investigation with comprehensive evidence gathering **Tool Flow**: ``` 1. perplexity_research - Comprehensive data gathering 2. think - Pattern recognition 3. openai_reason - Systematic analysis 4. gemini_brainstorm - Creative applications (optional) ``` **Use Cases**: - Market research - Technical investigations - Academic research - Competitive analysis - Trend analysis **Example**: ```bash focus --mode research "Latest developments in quantum computing" focus --mode investigate "Impact of remote work on productivity" ``` ### 🧩 Solve/Analyze Mode **Purpose**: Systematic problem-solving and analytical reasoning **Tool Flow**: ``` 1. think - Problem decomposition 2. openai_reason - First principles analysis 3. perplexity_research - Evidence gathering 4. gemini_brainstorm - Innovative solutions ``` **Use Cases**: - Technical debugging - Business problem-solving - Process optimization - Root cause analysis - Strategic planning **Example**: ```bash focus --mode solve "Optimize database query performance" focus --mode analyze "Why is customer retention dropping?" ``` ### 🎯 Synthesis/Integrate Mode **Purpose**: Combine multiple perspectives into coherent insights **Tool Flow**: ``` 1. gemini_brainstorm - Exploratory angles 2. perplexity_research - Comprehensive data 3. openai_reason - Analytical framework 4. think - Integration reflection ``` **Use Cases**: - Strategic decision-making - Multi-stakeholder alignment - Complex report synthesis - Cross-functional planning - Holistic analysis **Example**: ```bash focus --mode synthesis "Combine user feedback with technical constraints" focus --mode integrate "Merge marketing and engineering roadmaps" ``` ### ✅ Fact-Check/Verify/Validate Mode **Purpose**: Validate claims and ideas with evidence-based analysis **Tool Flow**: ``` 1. think - Decompose claims 2. perplexity_research - Find evidence 3. openai_reason - Analyze contradictions 4. gemini_brainstorm - Alternative explanations 5. think - Final verdict synthesis ``` **Use Cases**: - Content verification - Assumption validation - Risk assessment - Due diligence - Quality assurance **Example**: ```bash focus --mode fact-check "AI will replace all programmers by 2030" focus --mode verify "This startup's growth claims" focus --mode validate "Technical feasibility of the proposed solution" ``` ### 🪞 Reflect/Review Mode **Purpose**: Synthesize results from previous orchestration runs **Tool Flow**: - No tools executed - provides guidance for manual synthesis - Reviews outputs from previous tool executions - Identifies patterns and contradictions - Creates actionable insights **Use Cases**: - Post-brainstorm synthesis - Research summary - Decision consolidation - Learning extraction - Action planning **Example**: ```bash # After running any other mode: focus --mode reflect focus --mode review "Synthesize all findings" ``` ### 🔍 Debug Mode **Purpose**: Understand available workflows and troubleshoot **Tool Flow**: - No tools executed - provides system information - Shows available workflows and tools - Displays token efficiency options **Use Cases**: - Learning the system - Troubleshooting - Configuration testing **Example**: ```bash focus --mode debug "Show me what's available" ``` ## Mode Selection Guide Choose your mode based on your primary goal: | If you want to... | Use mode | |-------------------|----------| | Generate new ideas | `brainstorm` / `creative` / `ideate` | | Find information and evidence | `research` / `investigate` | | Solve a specific problem | `solve` / `analyze` | | Combine multiple viewpoints | `synthesis` / `integrate` | | Verify claims or assumptions | `fact-check` / `verify` / `validate` | | Summarize previous results | `reflect` / `review` | | Learn about the system | `debug` | ## Token Efficiency All modes support token-efficient operation: ```bash focus --mode brainstorm --tokenEfficient true "Your query" ``` This reduces output by ~30% while maintaining core functionality. ## Workflow Customization Each workflow includes: - **Required steps**: Core tools that always run - **Optional steps**: Additional tools that may be skipped based on context - **Adaptation checks**: Dynamic adjustments based on output length ## Best Practices 1. **Start with the right mode**: Choose based on your primary goal 2. **Provide context**: Include relevant background in your query 3. **Use reflect mode**: Always synthesize after complex workflows 4. **Iterate**: Run multiple modes for comprehensive analysis 5. **Token efficiency**: Use for faster responses when full output isn't needed ## Example Multi-Mode Workflow ```bash # 1. Generate ideas focus --mode brainstorm "New features for our app" # 2. Validate feasibility focus --mode fact-check "Can we implement real-time collaboration?" # 3. Deep dive on specifics focus --mode research "WebRTC implementation best practices" # 4. Synthesize everything focus --mode reflect ``` ## Mode Aliases Reference | Primary | Aliases | |---------|---------| | brainstorm | creative, ideate | | research | investigate | | solve | analyze | | synthesis | integrate | | fact-check | verify, validate | | reflect | review | Use whichever name feels most natural for your use case!