Primary Workflows Complete Academic Research Lifecycle -7 stages from idea to publication. 1. Research Ideation Systematic research initiation with idea generation and literature review: - Tools: research-ideation skill + literature-reviewer agent - Process: - 5W1H brainstorming: What, Why, Who, When, Where, How → structured thinking framework - Literature review: arXiv + Semantic Scholar integration → automated paper search and classification - Gap analysis: 5 types (literature, methodology, application, interdisciplinary, temporal) → identify research opportunities - Research questions: SMART principles → formulate specific, measurable questions - Command: /research-init “topic” → launch complete research initiation workflow 2. ML Project Development Maintainable ML project structure for experimental code: - Tools: architecture-design skill + code-reviewer agent + git-workflow skill - Process: - Structure: Factory & Registry patterns → config-driven models (cfg-only parameters) → enforced by rules/coding-style.md - Code style: 200–400 line files → type hints required → configs use @dataclass(frozen=True) → max 3-level nesting - Debugging (bug-detective): Error pattern matching for Python/Bash/JS → stack trace analysis → anti-pattern recognition - Git: Conventional Commits (feat/scope: message) → branching strategy (master/develop/feature) → merge with — no-ff - Commands: /plan, /commit, /code-review, /tdd 3. Experiment Analysis Statistical analysis and visualization of experimental results: - Tools: results-analysis skill + data-analyst agent - Process: - Data processing: automated cleaning and preprocessing of experiment logs - Statistical tests: t-test, ANOVA, Wilcoxon signed-rank → validate significance - Visualization: matplotlib/seaborn integration → publication-quality charts (line plots, bar charts, heatmaps) - Ablation studies: systematic component analysis → understand each part’s contribution - Command: /analyze-results → generate analysis report with charts and statistics 4. Paper Writing Systematic paper writing from template to final draft: - Tools: ml-paper-writing skill + paper-miner agent + latex-conference-template-organizer skill - Process: - Template preparation: download conference .zip → extract main files → remove example content → output clean Overleaf-ready structure - Citation verification (citation-verification): multi-layer validation (format → API → information → content) → prevent hallucinated citations - Systematic writing: narrative framework → 5-sentence abstract formula → section-by-section drafting with feedback loops - De-AI processing (writing-anti-ai): remove inflated symbolism, promotional language, vague attributions → add human voice and rhythm → bilingual support (Chinese/English) - Venues: NeurIPS, ICML, ICLR, ACL, AAAI, Nature, Science, Cell, PNAS 5. Paper Self-Review Quality assurance before submission: - Tools: paper-self-review skill - Process: - Structure check: logical flow, section balance, narrative coherence - Logic validation: argument soundness, claim-evidence alignment, assumption clarity - Citation audit: citation accuracy, proper attribution, reference completeness - Figure quality: visual clarity, caption completeness, color accessibility- Writing polish: grammar, clarity, conciseness, academic tone - Compliance: page limits, formatting requirements, ethics disclosure - 6-item checklist → systematic quality assessment 6. Paper Submission & Rebuttal Paper submission and review response: - Tools: review-response skill + rebuttal-writer agent - Submission process: - Pre-submission check: conference-specific checklists (NeurIPS 16items, ICML broader impact, ICLR LLM disclosure) - Format check: page limits, anonymization, supplementary materials - Final review: proofreading, citation check, figure verification - Rebuttal process: - Review analysis: parse and categorize comments (major/minor/typo/misunderstanding) - Response strategy: accept/defend/clarify/experiment → tailored approach for each comment type - Rebuttal writing: structured responses with evidence and reasoning - Tone management: professional, respectful, evidence-based language - Command: /rebuttal → generate complete rebuttal document and experiment plan 7. Post-Acceptance Conference preparation and research promotion: - Tools: post-acceptance skill - Process: - Presentation: slide creation guidance (15/20/30-minute formats) → visual design principles → narrative structure - Poster: academic poster templates (A0/A1 sizes) → layout optimization → visual hierarchy - Promotion: social media content (Twitter/X, LinkedIn) → blog posts → press releases → research summaries - Commands: /presentation, /poster, /promote → automated content generation

Claude Scholar

AI-Powered Full-Cycle Academic Research Assistant

AI-Powered Full-Cycle Academic Research Assistant