dsh-longtask-orchestrator

Handling long, complex tasks in AI workflows often requires coordinating multiple models, which can be cumbersome and error-prone. The dsh-longtask-orchestrator plugin addresses this by automating orchestration, with Codex handling high-level planning, scoring, and review, Kimi providing visual verification, summarization, and cross-validation, and DSH acting as the execution layer to form a seamless collaborative chain.

Its core capabilities lie in its intelligent tool suite: from task decomposition (codex_plan) to automatic scoring (codex_score), final review (codex_review), result compression (kimi_summarize), and validation (kimi_review), all integrated through state management (orchestrate_state). Users simply define a goal, and the plugin automatically initializes the task, iterates on subtasks, and optimizes until final approval, greatly simplifying complex workflows.

This plugin is ideal for content creators, developers, or anyone dealing with multi-stage AI tasks. Whether generating long-form content, analyzing datasets, or executing coding projects, it boosts efficiency, reduces manual intervention, and makes AI collaboration smarter and more reliable.