{
  "id": "page:docs-reference-repos-orchestra-research-ai-research-skills-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/orchestra-research/ai-research-skills/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/Orchestra-Research/AI-Research-SKILLs/research.md",
    "sourceKind": "repo-docs",
    "title": "Orchestra-Research/AI-Research-SKILLs",
    "displayName": "Orchestra-Research/AI-Research-SKILLs",
    "slug": "docs/reference-repos/orchestra-research/ai-research-skills/research",
    "articlePath": "wiki/docs/reference-repos/Orchestra-Research/AI-Research-SKILLs/research.md",
    "article": "\n# Orchestra-Research/AI-Research-SKILLs\n\n## Metadata\n- **Stars**: 6,637\n- **License**: MIT\n- **Last pushed**: 2026-04-10\n- **Topics**: ai, ai-research, claude, claude-code, claude-skills, codex, gemini, skills, vllm\n- **Fork**: No\n\n## Overview\nComprehensive open-source library of 87 AI research skills organized into 22 categories covering the full AI research lifecycle — from idea to paper. Includes an `autoresearch` orchestration skill that uses a two-loop architecture (inner experiment loop + outer synthesis loop) to run autonomous research projects.\n\n## Architecture\n- 22 numbered skill categories (model-architecture, tokenization, fine-tuning, interpretability, data-processing, post-training, safety, distributed-training, infrastructure, optimization, evaluation, inference, mlops, agents, rag, prompt-engineering, observability, multimodal, emerging-techniques, paper-writing, ideation)\n- Each skill follows SKILL.md format with YAML frontmatter, references/, scripts/, templates/\n- Claude Code plugin marketplace with `marketplace.json`\n- NPM installer: `npx @orchestra-research/ai-research-skills`\n\n## Extractable Value\n\n### Process: specializations/science/ai-research-orchestration\nThe `autoresearch` SKILL.md contains a complete two-loop research orchestration methodology:\n1. Bootstrap: scope question, literature survey, form hypotheses\n2. Inner loop: pick hypothesis, experiment, measure, record, learn\n3. Outer loop: review results, find patterns, update findings, new hypotheses\n4. Finalize: write paper, presentation, archive\n\nIncludes workspace structure template (research-state.yaml, experiments/, paper/), agent continuity patterns, and routing to domain skills. This is a full generic research methodology suitable for `specializations/science/autonomous-ai-research`.\n\n### Process: specializations/science/ml-paper-writing\nThe `research-paper-writing` skill (also found in hermes-agent, attributed to Orchestra Research) provides a complete 7-phase pipeline for ML paper writing targeting NeurIPS/ICML/ICLR/ACL/AAAI/COLM. Includes experiment design, monitoring, statistical analysis, iterative drafting, self-review, and submission prep.\n\n### Plugin idea: ai-research-skills\nThe marketplace already has a `marketplace.json` with plugin structure. Could be assimilated as a babysitter marketplace plugin providing AI research domain skills. However, these are tool/framework-specific skills (vLLM, DeepSpeed, etc.), not babysitter processes — they'd need to be wrapped as process definitions.\n\n## Processes\n\n### 1. Autonomous AI Research Orchestration\n- **Source**: autoresearch SKILL.md two-loop architecture (inner experiment loop + outer synthesis loop)\n- **Placement**: `specializations/science/autonomous-ai-research.js`\n- **Description**: Complete research methodology: bootstrap (scope question, literature survey, form hypotheses) → inner loop (pick hypothesis, experiment, measure, record, learn) → outer loop (review results, find patterns, update findings, new hypotheses) → finalize (write paper, presentation, archive).\n\n### 2. ML Paper Writing Pipeline\n- **Source**: research-paper-writing skill 7-phase pipeline for ML conferences\n- **Placement**: `specializations/science/ml-paper-writing.js`\n- **Description**: End-to-end ML paper writing for NeurIPS/ICML/ICLR/ACL/AAAI/COLM: experiment design → monitoring → statistical analysis → iterative drafting → self-review → submission prep.\n\n## Plugin Ideas\n\n- **AI Research Skills Suite**: Babysitter marketplace plugin providing AI research domain skills with 87 skills across 22 categories covering full AI research lifecycle.\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| Autonomous AI Research Orchestration | NEW | Two-loop research methodology with experiment and synthesis cycles | - | specializations/science/autonomous-ai-research.js |\n| ML Paper Writing Pipeline | NEW | 7-phase ML paper writing process for major conferences with statistical analysis | - | specializations/science/ml-paper-writing.js |\n| Research Workspace Organization | NEW | Structured research workspace with research-state.yaml and experiment tracking | - | specializations/science/research-workspace-organization.js |\n| Agent Continuity Patterns | NEW | Maintaining research state and context across long-running autonomous research projects | - | specializations/shared/agent-continuity-patterns.js |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| AI Research Skills Suite | NEW | Comprehensive AI research domain skills across 22 categories with orchestration | - | plugins/a5c/marketplace/blueprints/ai-research-skills-suite/ |\n\n## Classification\n- **Archetype**: Domain skill library with orchestration layer\n- **Primary value**: Two processes extractable — autonomous research orchestration and ML paper writing pipeline\n- **Process placement**: `specializations/science/autonomous-ai-research` and `specializations/science/ml-paper-writing`\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-orchestra-research-ai-research-skills-research",
      "kind": "contains_page"
    }
  ]
}