Research Plan

Core Features

This Skill develops personalized and actionable scientific research analysis strategies for users through systematic research of top journal/conference papers.

Core Values

Evidence-driven: Based on methodologies from top journal papers, not designed out of thin air
Systematic research: Automated literature retrieval, screening, and downloading processes
Deep learning: Extract analytical methods and technical details from PDFs
Actionability: Output clear goals/steps that can be directly implemented

Usage Scenarios

Formulating experimental design plans
Planning data analysis workflows
Designing technical roadmaps
Determining statistical method selection
Developing visualization strategies

Input Specifications

Required Inputs

Requirement Description: Clearly describe the problem to be solved and the objectives

Optional Inputs

Working directory: Defaults to current directory
Literature year range: Defaults to the past 5 years
Target journals/conferences: Specify specific sources
Language preference: Defaults to English
Maximum number of documents: Defaults to 30 papers

Workflow

Input

Provide necessary inputs according to user requests and configuration files; clearly list missing information and stop steps that depend on that input.

Execution Steps

For bugs caused by design flaws in this skill, first record them in
```
~/.bensz-skills/bugs/
```
in accordance with the
```
bensz-collect-bugs
```
specification, do not directly modify the skill source code installed locally by the user; if there is a workaround, first record the bug, then continue to complete the task.
Only when the user explicitly requests "report bensz skills bugs" or other public reporting, use the local
```
gh
```
to upload new bugs to
```
huangwb8/bensz-bugs
```
; do not pull / clone the entire repository.

AI-powered Scientific Research Analysis Strategy Planning Assistant

The legacy name

make-research-plan

is only retained as a prompt compatibility alias; historical

.make-research-plan/

directories are only for explicit compatibility reading, migration or cleanup, and new runs uniformly use task-level workspaces.

Phase 0: Initialization

Input:

User requirement description
Working directory path (optional, defaults to current directory)

Actions:

Verify that the working directory exists and is writable

Create a task-level hidden workspace

.bensz-api/task-{yyyymmdd-hhmm}-{short-description}/research-plan/

Clearly inform the user of the location and purpose of the created directory

Output:

Directory structure of

.bensz-api/task-{yyyymmdd-hhmm}-{short-description}/research-plan/

Directory Structure:

.bensz-api/task-{yyyymmdd-hhmm}-{short-description}/research-plan/
├── input/papers/        # Downloaded PDF literatures
├── input/metadata/      # Research metadata
│   ├── theme.json      # Themes and keywords
│   └── search_history.json # Search history
├── output/extracted/    # Extracted literature information
│   └── papers_info.json # Structured paper information
├── output/analysis-framework.md # Analysis framework summary
├── output/plan.md             # Draft within task
└── log/                       # Command, verification and error logs

Phase 1: Topic Extraction and Literature Research

Step 1.1: Topic Extraction

Objective: Structurally extract research topics and keywords from user requirements

Methods:

Analyze user requirement text
Identify core research areas
Extract key conceptual terms
Generate multiple related topics (if applicable)
Generate 3-5 keywords for each topic

Output:

metadata/theme.json

json

{
  "primary_topic": {
    "name": "Topic Name",
    "description": "Topic Description",
    "keywords": ["Keyword1", "Keyword2", "Keyword3"]
  },
  "secondary_topics": [
    {
      "name": "Related Topic 1",
      "keywords": ["Keyword1", "Keyword2"]
    }
  ],
  "extracted_at": "2026-01-19T09:30:00Z"
}

Step 1.2: Literature Retrieval

Retrieval Strategy:

Multi-source retrieval:
- PubMed (Biomedicine)
- Google Scholar (Comprehensive)
- IEEE Xplore (Engineering and Technology)
- arXiv (Preprints)
- Semantic Scholar (AI-driven)
Search query construction:
- Primary topic keywords AND analysis method-related terms
- Prioritize retrieving literatures from the past 5 years
- Limit to high-impact factor journals/top conferences
Quality filtering:
- Prioritize peer-reviewed articles
- Exclude non-English literatures (unless specified by user)
- Prioritize highly cited papers

Output:

metadata/search_history.json

json

{
  "searches": [
    {
      "topic": "Topic Name",
      "query": "Search Query",
      "source": "Data Source",
      "results_count": 150,
      "timestamp": "2026-01-19T09:35:00Z"
    }
  ]
}

Step 1.3: Literature Screening and Scoring

AI Article-by-Article Evaluation:

Semantic relevance scoring (1-10 points)
- Title matching degree
- Abstract relevance
- Keyword overlap
Subtopic grouping:
- Methodology papers
- Application cases
- Review articles
- Tools/software
High-priority selection:
- Retain papers with score ≥ 7
- Ensure representative papers for each subtopic
- Control total number between 15-30 papers

Output:

metadata/papers_scored.json

json

{
  "papers": [
    {
      "title": "Paper Title",
      "authors": ["Author1", "Author2"],
      "year": 2024,
      "journal": "Journal Name",
      "relevance_score": 9,
      "subtopic": "Methodology",
      "doi": "10.xxxx/xxxxx",
      "url": "https://doi.org/..."
    }
  ]
}

Step 1.4: Generate Reference List

Output:

references.bib

Format: Standard BibTeX, including all necessary fields

Step 1.5: Download PDFs

Methods:

Prioritize downloading via open access links
Use Unpaywall API to find legal free versions

Save to

papers/

directory, naming format:

{First Author's Last Name}_{Year}_{Journal Abbreviation}.pdf

Output:

papers/*.pdf

Error Handling:

Record DOIs that failed to download
Do not block subsequent processes
Specify which papers could not be obtained in the final report

Phase 2: Deep Learning Analysis Strategy

Step 2.1: PDF Information Extraction

Objective: Extract analytical methods and technical details from each PDF

Extraction Content:

Study Design: Experiment type, control settings, sample size
Data Preprocessing: Cleaning steps, normalization methods, quality control
Statistical Methods: Test methods, multiple comparison correction, effect size
Visualization Strategy: Chart types, color schemes, tools
Software Tools: R/Python packages, versions, parameter settings
Validation Methods: Cross-validation, bootstrap, sensitivity analysis

Methods:

Use PDF reading tools to extract text and tables
AI semantic analysis to identify content in the Methods section
Structurally store extracted information

Output:

extracted/papers_info.json

json

{
  "papers": [
    {
      "id": "paper_001",
      "title": "Paper Title",
      "methods": {
        "study_design": "Cohort Study",
        "preprocessing": ["Quality Control", "Normalization"],
        "statistical_tests": ["t-test", "ANOVA"],
        "correction": "FDR",
        "visualization": ["Box Plot", "Heatmap"],
        "software": {
          "R": ["limma", "ggplot2"],
          "parameters": {"adjust.method": "BH"}
        }
      },
      "strengths": ["Large sample size", "Pre-registration"],
      "limitations": ["Single-center study"]
    }
  ]
}

Step 2.2: Comprehensive Analysis Framework

Objective: Synthesize all papers and refine best practices

Analysis Dimensions:

Methodological Consensus:
- Methods shared by multiple papers
- Standard workflows in the field
- Recommended best practices
Methodological Differences:
- Applicable methods for different scenarios
- Emerging methods vs traditional methods
- Controversial points
Tool Ecosystem:
- Most commonly used software packages
- Comparison of tool advantages and disadvantages
- Version compatibility
Quality Control:
- Data validation standards
- Result robustness testing
- Reproducibility practices

Output:

analysis-framework.md

Template see:

research-plan/references/output-templates.md

Phase 3: Develop Personalized Plan

Step 3.1: Requirement Alignment

Analyze User Requirements:

What is the research question?
What are the special constraints?
What is the expected output?
What is the skill level?

Map to Analysis Framework:

Match relevant methodological sections
Identify applicable tools
Determine quality control points

Step 3.2: Plan Generation

Output:

plan.md

Template see:

research-plan/references/output-templates.md

Step 3.3: Plan Review

Present to User:

Plan overview
Explanation of key decision points
Comparison of optional solutions
Invitation for modification suggestions

Review Questions:

Does the plan cover your needs?
Are the steps clear and feasible?
Is any analysis missing?
Are there any parts that need adjustment?

For Users

Be specific with requirement description: Avoid vague expressions
Clarify constraints: Data types, sample size, skill level, etc.
Be thorough when reviewing the plan: This is your research, you have the final say
Keep the working directory: Facilitates subsequent adjustments and reproducibility

For AI Execution

Transparency: Clearly inform users what each step is doing
Traceability: All recommendations are supported by literature
Conservatism: Admit uncertainty instead of making up information when unsure
User-centric: The plan should conform to the user's actual constraints

Implementation/script status and responsibility boundaries see:

research-plan/references/implementation-notes.md

Change log see:

research-plan/CHANGELOG.md

(version number is subject to

research-plan/config.yaml:skill_info.version

)

Output

Core Outputs

plan.md
: Final analysis strategy plan (user-friendly format)
analysis-framework.md
: Analysis framework summary (detailed technical document)
references.bib
: Reference list (citable)

Auxiliary Outputs

metadata/
: Research metadata (reproducible)
extracted/papers_info.json
: Structured paper information (reusable)

Output Management

Intermediate files for new tasks of this Skill are uniformly written to

./.bensz-api/task-{yyyymmdd-hhmm}-{short-description}/{skill name}/input|output|log/

. The same task reuses one task root directory; create

shared/

only for multi-Skill collaboration. Formal deliverables are not written to this directory; historical hidden directories only allow explicit compatibility reading, migration or cleanup.

Verification

After completion, execute the Skill's existing static checks, script verification or manual review, and record the passing standards.

Failure and Recovery

Frequently Asked Questions

Q: Unable to download some PDFs A: Record the failed DOIs, which can be specified in the final report, or try to obtain them manually

Q: Extracted method information is incomplete A: Use supplementary materials or contact the authors

Q: Insufficient number of literatures A: Relax retrieval conditions (year, keyword range)

Q: Conflicting methods in different papers A: Mark controversial points in the analysis framework and provide multiple solution options

Constraints

Functional Boundaries

✅ Allowed:
- Research literatures and summarize methodologies
- Develop analysis strategies and steps
- Recommend tools and best practices
- Generate executable plan documents
❌ Prohibited:
- Perform actual data analysis
- Modify user's original data
- Write formal scientific papers
- Draw scientific conclusions

Usage Limitations

Only provide strategic suggestions, specific implementation requires user verification
Methodological recommendations are based on literatures, cannot guarantee applicability to all scenarios
PDFs depend on accessibility, some literatures may not be obtainable
AI extraction may have errors, key information is recommended to be manually reviewed

Quality Assurance

All recommendations come from peer-reviewed papers
Clearly mark evidence strength (consensus of multiple papers vs single paper)
Provide reference traceability
Mark uncertainties or controversial points

Public Hard Constraints

When tasks require disk storage, use a unique
```
./.bensz-api/task-{yyyymmdd-hhmm}-{short-description}/
```
root directory; shared materials are placed in
```
shared/
```
, and Skill-exclusive materials are placed in the Skill's
```
input/
```
,
```
output/
```
,
```
log/
```
. Formal deliverables, source code and formal plans are saved according to project agreements and not written to the task workspace; unauthorized overwriting, deletion, migration or remote writing is prohibited.
Project maintenance changes check BAC availability and record requirements, AI outputs, tool results, file changes and verification summaries; BAC only performs process audits and does not replace signature, responsibility or compliance judgments.
Do not record API Keys, access tokens, passwords, Cookies, environment/credential files, private Prompts, identity information, local usernames, hostnames or unnecessary large-volume raw data.
File paths must be normalized and limited within authorized project scope; external URLs, subprocesses and network access follow the principle of least privilege to prevent path traversal, SSRF and command injection.
Skill versions are uniquely recorded in its own
```
config.yaml:skill_info.version
```
; public API, protocol, directory or configuration changes are synchronized with documents and
```
CHANGELOG.md
```
.
Only design defects of the Skill or Bensz infrastructure itself are submitted to
```
bensz-collect-bugs
```
; first desensitize and write to
```
~/.bensz-skills/bugs/
```
, do not interrupt the current task, only report publicly when explicitly requested by the user, and prohibit direct modification of the Skill source code installed by the user.

Skill-specific Constraints

Do not exceed the scope stated in this Skill description and the above workflow; do not disguise unverified information as definitive conclusions.

research-plan

NPX Install

Tags

SKILL.md Content (Chinese)

Research Plan

Core Features

Core Values

Usage Scenarios

Input Specifications

Required Inputs

Optional Inputs

Workflow

Input

Execution Steps

Phase 0: Initialization

Phase 1: Topic Extraction and Literature Research

Step 1.1: Topic Extraction

Step 1.2: Literature Retrieval

Step 1.3: Literature Screening and Scoring

Step 1.4: Generate Reference List

Step 1.5: Download PDFs

Phase 2: Deep Learning Analysis Strategy

Step 2.1: PDF Information Extraction

Step 2.2: Comprehensive Analysis Framework

Phase 3: Develop Personalized Plan

Step 3.1: Requirement Alignment

Step 3.2: Plan Generation

Step 3.3: Plan Review

For Users

For AI Execution

Output

Core Outputs

Auxiliary Outputs

Output Management

Verification

Failure and Recovery

Frequently Asked Questions

Constraints

Functional Boundaries

Usage Limitations

Quality Assurance

Public Hard Constraints

Skill-specific Constraints