Refine vague research directions into implementation-ready method plans.
npx skills add https://github.com/Mang30/myskills --skill research-refine-mang30
Please help me install this Agent Skill. Skill: research-refine Source: https://github.com/Mang30/myskills/tree/main/skills/research-refine Command: npx skills add https://github.com/Mang30/myskills --skill research-refine-mang30
Helps convert a vague research direction into a concrete, implementation-oriented research plan anchored to the original problem, with minimal validation experiments and frontier-aware method choices.
Use cases: refining an approach you already know is on the right track but still “fuzzy,” decomposing a research idea into a paper-ready method, and producing a top-venue style plan without turning it into an overbuilt benchmark campaign.
Use the research-refine skill by telling it your bottom-line problem and the vague approach you currently have, for example: refine my approach to solve [problem] using [high-level idea] so that the result is implementation-oriented, elegant, and frontier-aware.
High-intent search queries and answers about installing and using this skill.
To turn a vague research idea into an implementation-ready method plan, you need iterative problem framing and method-first proposal writing. This process anchors your original problem and refines the technical route into concrete system, thesis, training, and inference specifications.
The best way to refine experiment design for foundation-model-era research is through a frontier-aware review loop. This method evaluates your proposal for fidelity, specificity, and contribution sharpness while ensuring appropriate use of modern foundation-model primitives.
To prevent problem drift during iterative research refinement, use problem anchor preservation. This technique freezes an immutable problem statement and carries it verbatim through each review round to ensure the technical route remains anchored to the original core issue.
Yes, you can use WebSearch and WebFetch for literature grounding during research method planning. Integrating web search allows you to validate your problem framing against existing literature and incorporate frontier-aware machine learning techniques into your refined plan.
Yes, the iterative review process supports checkpoint recovery by persisting state across rounds. This allows you to recover from interruptions and maintains round-by-round artifact logs for full auditability of your method-first proposal writing.
You should not use method-first proposal writing when your core problem is completely invisible or when you need an overbuilt benchmark campaign instead of an elegant plan. It specifically applies when the problem is visible but the technical route remains unclear.
Chanw-researchTurn vague research directions into anchored, implementable problem formulations.
Shallow-WRefine vague research directions into implementation-ready method plans.
dogekikiRefine vague research directions into implementation-oriented method plans.
raja21068Refine vague ML/AI research directions into implementable method plans.
