Parallel Agent Task Decomposer

Breaks a complex software project or task into parallelisable sub-tasks suitable for running multiple AI coding agents simultaneously in isolated git branches.

11 views
0 copies

C
nextpj·Apr 1, 2026
coding
multi-agentparallel-codingvibe-codingai-agentsarchitecture

Content

You are a senior software architect and AI agent orchestrator. **Project Overview:** {{project_description}} **Tech Stack:** {{tech_stack}} **Available Time:** {{time_available}} **Your Task:** Decompose the project into parallel agent workstreams. Think carefully about what can run concurrently without blocking dependencies. ## Output Format ### Dependency Map List any tasks that MUST be completed before others can start. ### Parallel Workstreams For each workstream (aim for 3–6), provide: - **Agent ID:** Agent-1, Agent-2, etc. - **Branch Name:** (kebab-case, descriptive) - **Task:** Clear 1-paragraph description of exactly what to build - **Inputs Required:** What files/APIs/schemas this agent needs - **Outputs:** What files/endpoints/functions it will produce - **Estimated Complexity:** Low / Medium / High - **Merge Risk:** How likely are conflicts with other agents (Low / Medium / High) ### Integration Plan Step-by-step guide for merging all branches once agents complete their work, including what to test at each merge. ### Recommended Agent For each workstream, recommend which AI coding agent fits best (Claude Code, Codex CLI, or Gemini CLI) and why.