메뉴
Use when running an end-to-end academic paper project or upgrading a research idea into a submission-grade workflow with literature review, experiments, writing, review, and export.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when starting a paper project and needing to inventory local code, docs, datasets, constraints, and source materials before literature review or experiment planning.
Use when interpreting experiment outputs, checking whether data supports the claims, and deciding whether to proceed, refine, or pivot.
Use when outlining, drafting, or revising a conference paper with explicit evidence links, section word budgets, and figure planning.
Use when preparing the final paper package, checking citations and archives, and exporting a submission-ready LaTeX or markdown deliverable.
Use when turning literature synthesis into testable hypotheses, research questions, contribution framing, or pivot decisions for a paper.
Use when designing experiment plans, baselines, ablations, compute budgets, and project structure for a research-paper project.
| name | autoacad |
| description | Use when running an end-to-end academic paper project or upgrading a research idea into a submission-grade workflow with literature review, experiments, writing, review, and export. |
AutoAcad is a layered research-paper package. It keeps the lightweight repo-and-experiment workflow from academic and adds the heavier paper pipeline: stage gates, evidence checks, paper drafting, review, and export.
Use the narrowest subskill that matches the current stage. Use the package root only when the task spans multiple stages or the user wants an end-to-end workflow.
| Need | Use |
|---|---|
| Inventory the project, docs, data, codebase, and constraints | autoacad-prepare |
| Collect papers, benchmark families, datasets, and theory notes | autoacad-survey |
| Turn synthesis into hypotheses, claims, and contribution framing | autoacad-ideate |
| Design experiments, baselines, ablations, and project structure | autoacad-plan |
| Implement and run experiments with pilot timing and self-repair | autoacad-run |
| Interpret metrics, decide proceed/refine/pivot, and surface evidence gaps | autoacad-analyze |
| Produce outline, section drafts, figure plan, and paper wording | autoacad-draft |
| Stress-test methodology-evidence consistency and reviewer objections | autoacad-review |
| Prepare final package, citations, archive, and export checks | autoacad-export |
references/restricts.md as the hard floor. Do not present environment issues, setup failures, or debugging artifacts as research contributions.references/pipeline.md to map the current request onto the 25-stage loop. Do not pretend the workflow is linear when evidence requires a refine or pivot.references/experiment-rules.md before generating or modifying experiment code.references/paper-structure.md before drafting or revising paper text.scripts/init_paper_project.py rather than recreating folders ad hoc.PROGRESS.md and per-stage plans in plans/. If the project does not have them yet, initialize them first.references/pipeline.md: stage groups, loop points, and gate semantics.references/restricts.md: topic lock, evidence discipline, and anti-fabrication rules.references/experiment-rules.md: pilot timing, time-guard, convergence, and numerical-stability requirements.references/paper-structure.md: section goals, word budgets, and paper-quality checks.references/source-blueprint.md: normalized mapping from the original long-form mega-prompt into AutoAcad layers.references/update-playbook.md: maintainer guide for updating AutoAcad when the upstream long-form prompt changes.references/upstreams.md: generated summary of current AutoResearchClaw and AI-Researcher upstream state.references/upstream-snapshots/: local copies of tracked upstream source files for review.references/upstream-review.md: optional LLM-generated recommendations based on upstream drift.templates/PROGRESS.template.md: reproducibility and iteration log template.templates/stage-plan.template.md: per-stage planning template.templates/paper-tree.template.md: canonical paper project tree.scripts/init_paper_project.py: scaffold a paper project directory.scripts/check_stage_outputs.py: verify required folders and stage artifacts exist.scripts/check_results_vs_claims.py: compare results JSON against draft text to catch unsupported numeric claims.scripts/update_from_upstreams.py: refresh upstream snapshots and generate AutoAcad update reports.references/experiment-rules.md).