Skip to content

Latest commit

 

History

History
67 lines (56 loc) · 4.35 KB

File metadata and controls

67 lines (56 loc) · 4.35 KB

Pipeline Outputs

Dr. Claw's research pipeline produces structured artifacts across five stages. Each stage writes to a dedicated directory created when the project is initialized.

Output Artifacts

Artifact Location Description
📚 Survey reports Survey/reports/ Literature review summaries with citations, synthesized from arXiv, Semantic Scholar, and web sources
📄 Reference papers Survey/references/ Downloaded PDFs and structured notes (abstract, methodology, evaluation, knowledge graph entries)
💡 Research ideas Ideation/ideas/ Structured brainstorming outputs using creative frameworks (SCAMPER, SWOT, Mind Mapping) with multi-persona evaluation scores
📖 Ideation references Ideation/references/ Supporting materials and prior work collected during idea generation
🔬 Experiment code Experiment/core_code/ Implementation code produced by the plan → implement → judge loop
📦 Datasets Experiment/datasets/ Downloaded or generated datasets used in experiments
🧪 Code references Experiment/code_references/ Cloned GitHub repos and code survey outputs (architecture maps, dependency graphs)
📊 Analysis results Experiment/analysis/ Statistical analysis, tables, charts, and paper-ready figures from experiment runs
📝 Paper draft Publication/paper/ Academic manuscript in IEEE/ACM format with proper citations, structure, and LaTeX math
🖼️ Generated figures Publication/paper/ AI-generated figures via Gemini image models (Nano Banana Pro)
🎞️ Slide deck Promotion/slides/ Academic presentation slides with narration scripts
🔊 Audio narration Promotion/audio/ TTS-generated audio for presentation delivery
🎬 Demo video Promotion/video/ Combined slides + audio demo video
🌐 Project homepage Promotion/homepage/ Generated project landing page for dissemination

Project Directory Structure

When a project is created, Dr. Claw initializes the following workspace:

your-project/
├── instance.json                    # Project config with absolute paths
├── Survey/
│   ├── references/                  # Downloaded papers and structured notes
│   └── reports/                     # Literature review summaries
├── Ideation/
│   ├── ideas/                       # Generated and evaluated research ideas
│   └── references/                  # Supporting materials for ideation
├── Experiment/
│   ├── code_references/             # Cloned repos and code survey outputs
│   ├── datasets/                    # Experiment datasets
│   ├── core_code/                   # Implementation code
│   └── analysis/                    # Results, statistics, and figures
├── Publication/
│   └── paper/                       # Manuscript drafts and generated figures
└── Promotion/
    ├── homepage/                    # Project landing page
    ├── slides/                      # Presentation deck
    ├── audio/                       # TTS narration
    └── video/                       # Demo video

Pipeline Flow

Survey → Ideation → Experiment → Publication → Promotion

Each stage is powered by one or more research skills. The agent reads and follows the corresponding SKILL.md to produce the artifacts above. Skills can be run independently (e.g., only paper writing) or as a full end-to-end pipeline.

Stage → Skill Mapping

Stage Skills Used Key Outputs
Survey inno-prepare-resources, inno-code-survey, inno-deep-research, paper-analyzer, paper-finder Literature reports, reference notes, code survey maps
Ideation inno-idea-generation, inno-idea-eval Ranked research ideas with multi-dimension scores
Experiment inno-experiment-dev, inno-experiment-analysis Runnable code, results tables, statistical analysis
Publication inno-paper-writing, inno-figure-gen, inno-paper-reviewer, inno-humanizer, inno-reference-audit, inno-rclone-to-overleaf Manuscript, figures, review feedback, Overleaf sync
Promotion making-academic-presentations Slides, narration audio, demo video, homepage