Agentic Task Decomposer

Breaks down any complex goal into an ordered sequence of atomic, executable sub-tasks optimized for AI agent execution — perfect for Manus, Claude Code, Codex, or any autonomous AI workflow.

28 views
0 copies

C
nextpj·Mar 22, 2026
productivity
agentic-aitask-planningautomationai-agentsworkflowmanus

Content

You are an expert AI orchestration architect who specializes in designing agentic workflows for autonomous AI systems in 2026. Decompose the following complex goal into an executable agentic task plan: **High-Level Goal:** {{goal}} **Available Tools/Capabilities:** {{available_tools}} **Constraints:** {{constraints}} **Success Criteria:** {{success_criteria}} **Preferred Agent:** {{preferred_agent}} Create a structured agentic execution plan: ## Phase 1: Planning & Reconnaissance List 2-3 information-gathering sub-tasks the agent should do first. ## Phase 2: Core Execution Tasks Break the main work into atomic, independently executable steps. For each step: - **Task ID:** T-001, T-002... - **Action:** Specific verb + object (e.g., "Search web for X", "Write file Y", "Call API Z") - **Input:** What data/context this task needs - **Output:** Expected result - **Dependencies:** Which prior tasks must complete first - **Error handling:** What to do if this task fails ## Phase 3: Verification Tasks Checks to verify the goal was achieved correctly. ## Phase 4: Cleanup & Reporting Final steps to report results and clean up. Also provide: - Estimated token cost per phase - Critical path (minimum steps to reach success) - Parallelization opportunities (which tasks can run simultaneously)