aris-research-refine

Turn a vague research direction into a problem-anchored, elegant, frontier-aware, implementation-oriented method plan via iterative GPT-5.4 review. Use when the user says "refine my approach", "帮我细化方案", "decompose this problem", "打磨idea", "refine research plan", "细化研究方案", or wants a concrete research method that stays simple, focused, and top-venue ready instead of a vague or overbuilt idea.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Critical

Do not install without reviewing

Quality

Content

77%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable and the multi-phase workflow is sequenced with strong validation checkpoints, but it is long and monolithic with everything inline rather than progressively disclosed into reference files.

Suggestions

Move the full verbatim reviewer prompts (Phase 2 and Phase 4) and the output-file markdown templates into separate reference files under references/, leaving short summaries and links in SKILL.md to cut hundreds of lines.

Consolidate the repeated score-evolution tables and the per-phase checkpoint JSON examples into one canonical reference instead of restating them in each phase.

Trim restated guidance — the Key Rules section reiterates principles already stated in Overview and in the per-phase steps; keep only what adds new information.

Dimension	Reasoning	Score
Conciseness	The ~730-line body is mostly operational scaffolding rather than padded concept explanation, but it is verbose: full verbatim review prompts, complete output-file templates, and repeated score tables appear inline and could be tightened significantly.	2 / 3
Actionability	It provides copy-paste-ready Codex MCP call blocks with exact config ('model_reasoning_effort: xhigh'), full reviewer prompts, concrete constants (MAX_ROUNDS=5, SCORE_THRESHOLD=9), and explicit JSON checkpoint schemas.	3 / 3
Workflow Clarity	Phases 0–5 are clearly sequenced with a checkpoint after each phase, explicit stop conditions (overall score >= 9), resume/recovery logic, and validation steps (Anchor Check, Simplicity Check).	3 / 3
Progressive Disclosure	No bundle files exist and the entire skill is a single monolithic file with all templates, prompts, and phase detail inline; sections are well organized, but content that could live in separate reference files is not split out.	2 / 3
	Total	10 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, fully answers what and when with explicit bilingual triggers, and carves out a clear niche. It uses third person throughout, so no voice penalty applies.

Dimension	Reasoning	Score
Specificity	It lists concrete actions — 'Turn a vague research direction into a problem-anchored, elegant, frontier-aware, implementation-oriented method plan' and 'via iterative GPT-5.4 review' — describing the transformation and mechanism specifically rather than in the abstract.	3 / 3
Completeness	It answers both what (turn a vague direction into a concrete method plan via GPT-5.4 review) and when, with an explicit 'Use when the user says...' clause enumerating triggers.	3 / 3
Trigger Term Quality	It gives strong coverage of natural phrasings a user would actually say, including 'refine my approach', 'decompose this problem', 'refine research plan', and bilingual variants like '帮我细化方案' and '细化研究方案'.	3 / 3
Distinctiveness Conflict Risk	The niche — iterative GPT-5.4 review of a research method plan toward an elegant, top-venue-ready proposal — is specific and paired with distinctive triggers, making conflict with other skills unlikely.	3 / 3
	Total	12 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (740 lines); consider splitting into references/ and linking	Warning
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning

	Total	14 / 16 Passed

Repository: OpenLAIR/dr-claw
Path: skills/aris-research-refine/SKILL.md
Commit: 6da7f7c

Reviewed: about 7 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.