This document outlines a four-part implementation workflow for research projects. Here are the key components:
Core Process
The research plan requires completing a "Novix Plan Agent" mechanism that transforms survey findings into actionable implementation steps. The workflow mandates: "Don't ask permission. Just do it."
Four Required Sections
Dataset Plan — Specifies data source, preprocessing steps, and DataLoader configuration
Model Plan — Details architecture with component-to-formula mapping and reference code paths
Training Plan — Defines loss functions, optimizer settings, and monitoring metrics
Testing Plan — Establishes evaluation metrics, baselines, and ablation studies
Critical Requirements
Each model component must "have corresponding formula" and reference code from analyzed repositories
Plans must include concrete parameter values, not generic recommendations
Prerequisites include task.json, survey_res.md, and repository analysis
Missing survey_res.md halts execution with message: "需要先运行 /research-survey 完成深度分析"
Validation Checklist
The self-check ensures: components map to formulas, reference code is documented, datasets have acquisition methods, loss functions have mathematical definitions, and evaluation metrics are clearly specified.
1---2name: research-plan3description: Research Plan Summary4---5# Research Plan Summary67This document outlines a four-part implementation workflow for research projects. Here are the key components:89## Core Process1011The research plan requires completing a "Novix Plan Agent" mechanism that transforms survey findings into actionable implementation steps. The workflow mandates: "Don't ask permission. Just do it."1213## Four Required Sections14151. **Dataset Plan** — Specifies data source, preprocessing steps, and DataLoader configuration162. **Model Plan** — Details architecture with component-to-formula mapping and reference code paths173. **Training Plan** — Defines loss functions, optimizer settings, and monitoring metrics184. **Testing Plan** — Establishes evaluation metrics, baselines, and ablation studies1920## Critical Requirements2122- Each model component must "have corresponding formula" and reference code from analyzed repositories23- Plans must include concrete parameter values, not generic recommendations24- Prerequisites include `task.json`, `survey_res.md`, and repository analysis25- Missing `survey_res.md` halts execution with message: "需要先运行 /research-survey 完成深度分析"2627## Validation Checklist2829The self-check ensures: components map to formulas, reference code is documented, datasets have acquisition methods, loss functions have mathematical definitions, and evaluation metrics are clearly specified.
Run npx skillmds@latest add lamm-mit/research-plan in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Research Plan Summary It is listed under Research & Search on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
lamm-mit (@lamm-mit) published this skill. Their other Agent Skills are listed on their SkillMD profile.
lamm-mit