메뉴
Defines the checkpoint protocol for the elegant-researcher-skill pipeline. Use at major research transition points to summarize artifacts, show options, apply mode-specific defaults, and continue without blocking routine decisions.
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
Orchestrate a complete AI/ML research project from a user's idea to a finished research artifact by coordinating bundled research skills: autoresearch, research-lookup, brainstorming-research-ideas, scientific-critical-thinking, scholar-evaluation, ml-training-recipes, academic-plotting, scientific-visualization, scientific-schematics, imagegen, scientific-writing, ml-paper-writing, systems-paper-writing, and user-interaction. Supports explicit modes, evidence levels, checkpointed user steering, micro-validation, full-experiment handoff, and final response contracts.
Orchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment iterations with clear optimization targets. The outer loop synthesizes results, identifies patterns, and steers research direction. Routes to domain-specific skills for execution, supports continuous agent operation via Claude Code /loop and OpenClaw heartbeat, and produces research presentations and papers. Use when starting a research project, running autonomous experiments, or managing a multi-hypothesis research effort.
Guides researchers through structured ideation frameworks to discover high-impact research directions. Use when exploring new problem spaces, pivoting between projects, or seeking novel angles on existing work.
AI image generation using Google Gemini (Gemini) and OpenAI GPT-Image. Generate, edit, iterate, and create assets.
Look up current research information using saved source searches. Prefer direct parallel-cli search commands for fast web search when available; use research_lookup.py for API-routed Parallel Chat deep research or Perplexity academic searches. Use for finding papers, gathering research data, and verifying scientific information. Note: API-routed queries are transmitted to api.parallel.ai (PARALLEL_API_KEY) or openrouter.ai (OPENROUTER_API_KEY).
Evaluate scientific claims and evidence quality. Use for assessing experimental design validity, identifying biases and confounders, applying evidence grading frameworks (GRADE, Cochrane Risk of Bias), or teaching critical analysis. Best for understanding evidence quality, identifying flaws. For formal peer review writing use peer-review.
| name | user-interaction |
| description | Defines the checkpoint protocol for the elegant-researcher-skill pipeline. Use at major research transition points to summarize artifacts, show options, apply mode-specific defaults, and continue without blocking routine decisions. |
| license | MIT |
This skill controls how the elegant-researcher-skill pipeline involves the user.
| Mode | User Interaction Behavior |
|---|---|
quick-cycle | Minimal checkpoints; use defaults quickly |
deep-cycle | Present checkpoints, but continue with defaults when user gives no direction |
paper-ready | Ask only when full results, venue, or citation status affects claims |
theory-only | Skip experiment handoff; focus on theory and proposed experiments |
human-in-loop | Wait for user decisions at checkpoints |
Each checkpoint response should include:
## [Checkpoint Title]
Status: [one sentence]
Mode: [mode]
Evidence level: [highest relevant level]
Key artifacts:
- [path] - [what it contains]
Summary:
- [key point 1]
- [key point 2]
- [key point 3]
Options:
A. [default option] - [route]
B. [alternative] - [route]
C. [alternative] - [route]
D. [alternative] - [route]
Default: A unless mode-specific rules say otherwise.
Do not include long raw file contents.
Trigger: literature search and literature/gap-analysis.md are complete.
Present:
Options:
A. Continue with the recommended gap. B. Deepen a specific gap. C. Adjust literature direction. D. Switch to theory-only framing.
Default:
human-in-loop: wait for user selection.Trigger: candidate hypotheses and hypothesis-tree.md are complete.
Present:
Options:
A. Start with the recommended hypothesis. B. Select a different hypothesis or multiple hypotheses. C. Add a user hypothesis. D. Brainstorm again.
Default:
human-in-loop: wait for user selection.Trigger: protocols, code, micro-validation, and handoff package are ready.
Present:
Options:
A. Use micro-validation as proof-of-concept and proceed to writing. B. User will run full experiments; wait for results. C. Modify or fix experiments, then re-run micro-validation. D. Switch to theory-only or proposed-experiments path.
Default by mode:
quick-cycle: A.deep-cycle: A.paper-ready: B if user has compute and wants empirical claims; otherwise A with full claims marked pending.theory-only: D.human-in-loop: B.This checkpoint must not silently stall autonomous runs.
Trigger: first draft or research note is complete.
Present:
Options:
A. Polish and finalize. B. Strengthen a section. C. Integrate new user results. D. Rethink contribution.
Default:
human-in-loop: wait for user review.Trigger: final artifact is ready.
Present:
Options:
A. Mark cycle complete. B. Deepen a specific aspect. C. Convert to LaTeX or venue-specific submission. D. Start a new research question.
Default:
human-in-loop: wait for user decision.When full experiments are needed, produce HANDOFF.md with:
At the end of a cycle, the agent must include: