Task Planning Framework
Transform high-level goals into executable workflows. The Task Planning Framework understands your intent, formulates intelligent plans using cognitive modes, and orchestrates specialized tasks to deliver results autonomously.
Core Concepts
Cognitive Modes
Nine distinct "thinking styles" from linear Waterfall planning to autonomous Adaptive agents and collaborative Council deliberation.
Task Library
50+ specialized tasks for file operations, code execution, web scraping, reasoning, and content generation.
Orchestration
Intelligent dependency management, parallel execution, and automatic context passing between workflow steps.
Sub-Planning
Recursive task decomposition allows complex goals to spawn independent sub-plans with their own cognitive modes.
Cognitive Modes
Select the right thinking style for your problem. Each mode offers a unique approach to planning and execution.
The Architect. Analyzes the entire request upfront and generates a comprehensive, sequential plan with visual diagrams. Review and modify before execution begins. Ideal for well-defined tasks where the path to the solution is clear.
The Partner. Operates interactively, selecting and executing one optimal task at a time based on conversation context. Maintains full session history for informed decisions. Perfect for exploratory work and debugging.
The Strategist. Autonomous agent using a cyclical "Think, Act, Reflect" loop. Maintains dynamic internal state of goals, facts, and hypotheses. Adapts strategy based on results until the goal is achieved.
The Manager. Divide-and-conquer strategy that recursively decomposes massive goals into a tree of sub-goals and tasks. Manages dependencies and executes in parallel where possible.
The Factory. High-throughput batch operations using templates and variable sets. Generates all combinations and executes simultaneously with configurable concurrency. Ideal for codebase-wide changes.
The Auditor. Enforces strict methodologies (like TDD) using a state machine with validation gates. A 'Referee' agent verifies each step before proceeding, ensuring high-quality output.
The Operator. Assigns a dedicated AI operator to drive a specific tool continuously. Maintains deep session context through repeated interaction until the goal is achieved.
The Board. Multi-agent deliberation where a "Council" of personas (CEO, CTO, QA) collaborates. They nominate tasks, vote on actions, and maintain separate perspectives for balanced decisions.
The Executor. Loads predefined plan templates, substitutes variables from user input, and executes the workflow. Ideal for repeatable processes and sharing proven strategies.
Task Library
A comprehensive toolkit of specialized tasks that cognitive modes orchestrate to accomplish goals.
File Operations
FileModificationTask- Create, update, delete filesFileSearchTask- Search files by content or patternAnalysisTask- Analyze file contents and structure
Code & Execution
RunCodeTask- Execute code snippetsRunShellCommandTask- Run terminal commandsSelfHealingTask- Auto-fix failing scripts
Reasoning
ChainOfThoughtTask- Step-by-step reasoningDecompositionSynthesisTask- Break down & synthesizeMultiPerspectiveAnalysisTask- Multiple viewpointsSystemsThinkingTask- Holistic analysis
Online & Web
CrawlerAgentTask- Web scrapingGitHubSearchTask- Search GitHub codeSeleniumSessionTask- Browser automation
Knowledge Base
KnowledgeIndexingTask- Build knowledge baseVectorSearchTask- Semantic search
Content Generation
WriteHtmlTask- Generate web pagesGeneratePresentationTask- Create slidesNarrativeGenerationTask- Write stories
Interactive Simulator
See how different cognitive modes approach the same problem. Select a mode and generate a plan.
Configuration
{}