name

researchcritic

description

Universal scientific literature critical analysis and survey-generation workflow. Use when a user asks to analyze academic papers, critique research claims, audit experiments, compare papers, generate literature reviews, create related-work sections, or plan early-stage research from papers. The skill enforces a background-first research frame, reviewer-grade evidence criticism, paper memory cards, and cross-paper survey synthesis.

ResearchCritic

Use ResearchCritic to perform rigorous scientific paper analysis and literature survey generation.

Required Workflow

Establish a fixed research frame before formal analysis:
- field
- core scientific question
- scenario characteristics
- analysis goal
If the user has not provided this frame, ask for it first.
If the user explicitly requests generic review, create a temporary generic frame and label the output as generic.
For a single paper, produce a critical Markdown report using the template in prompts/single_paper_template.zh.md.
After each analysis, record a paper memory card with title, problem, method, evidence, strengths, weaknesses, unresolved questions, relevance, and confidence.
For a literature survey, synthesize across all paper memory cards using the template in prompts/literature_survey_template.zh.md.

Core Standard

Do not summarize papers superficially. Audit whether the paper actually proves what it claims. Check:

problem definition
contribution novelty and necessity
experimental validity
baseline fairness
metric alignment
ablation sufficiency
statistical reliability
reproducibility
limitation severity
relevance to the user's research frame

Resources

Full implementation specification: ResearchCritic_Skill.md
Modular system prompt: prompts/system_prompt.zh.md
Single-paper template: prompts/single_paper_template.zh.md
Survey template: prompts/literature_survey_template.zh.md
Tool schemas: schemas/openai_tools.json
Evaluation tasks: evals/sample_tasks.md

Output Language

Default to Chinese reports unless the user requests another language.

ResearchCritic

Use ResearchCritic to perform rigorous scientific paper analysis and literature survey generation.

Required Workflow

Establish a fixed research frame before formal analysis:

field
core scientific question
scenario characteristics
analysis goal

If the user has not provided this frame, ask for it first.

If the user explicitly requests generic review, create a temporary generic frame and label the output as generic.

For a single paper, produce a critical Markdown report using the template in prompts/single_paper_template.zh.md.

After each analysis, record a paper memory card with title, problem, method, evidence, strengths, weaknesses, unresolved questions, relevance, and confidence.

For a literature survey, synthesize across all paper memory cards using the template in prompts/literature_survey_template.zh.md.

Core Standard

Do not summarize papers superficially. Audit whether the paper actually proves what it claims. Check:

problem definition

contribution novelty and necessity

experimental validity

baseline fairness

metric alignment

ablation sufficiency

statistical reliability

reproducibility

limitation severity

relevance to the user's research frame

Resources

Full implementation specification: ResearchCritic_Skill.md

Modular system prompt: prompts/system_prompt.zh.md

Single-paper template: prompts/single_paper_template.zh.md

Survey template: prompts/literature_survey_template.zh.md

Tool schemas: schemas/openai_tools.json

Evaluation tasks: evals/sample_tasks.md

Output Language

Default to Chinese reports unless the user requests another language.