{
  "id": "page:docs-reference-repos-clawhub-parags-deep-research-pro-research",
  "_kind": "Page",
  "_file": "wiki/docs/reference-repos/clawhub/parags/deep-research-pro/research.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/reference-repos/clawhub/parags/deep-research-pro/research.md",
    "sourceKind": "repo-docs",
    "title": "parags/deep-research-pro",
    "displayName": "parags/deep-research-pro",
    "slug": "docs/reference-repos/clawhub/parags/deep-research-pro/research",
    "articlePath": "wiki/docs/reference-repos/clawhub/parags/deep-research-pro/research.md",
    "article": "\n# parags/deep-research-pro\n\n- **Archetype**: methodology-repo\n- **Stars**: 1\n- **Last pushed**: 2026-02-03\n- **License**: MIT\n- **Discovered**: 2026-04-12\n- **Source**: ClawHub skills\n- **Skills found**: 1\n\n## Summary\n\nA multi-source deep research agent skill for OpenClaw/Clawdbot. Implements a 6-step research methodology: understand goal, plan sub-questions, execute multi-source search (DuckDuckGo web + news), deep-read key sources, synthesize cited report, save and deliver. No API keys required. Includes Python DDG search script and curl-based page reading.\n\nThe skill defines a clear, replayable research workflow with quality rules (every claim needs a source, cross-reference, recency preference, acknowledge gaps, no hallucination). Reports follow a structured template with executive summary, themed sections with inline citations, key takeaways, and methodology notes.\n\n## Assessment\n\nThe research methodology is well-structured and directly maps to a babysitter process. The 6-step workflow (clarify -> plan sub-questions -> multi-source search -> deep-read -> synthesize -> deliver) is a clean multi-phase process with clear inputs/outputs at each stage. The quality rules are codifiable as verification criteria.\n\nLow star count but high methodology quality. The DDG search dependency is ClawHub-specific but the methodology itself is tool-agnostic.\n\n## Extraction Priority\n- High\n- Rationale: Clean multi-step research methodology that maps directly to a babysitter process. The plan-search-synthesize pattern is domain-agnostic and reusable. Good candidate for specializations/shared/ or specializations/research/.\n\n# Extractable Value: parags/deep-research-pro\n\n## Processes\n\n- **deep-research**: Multi-source research with cited report synthesis\n  - Source: SKILL.md 6-step workflow\n  - Placement: specializations/shared/deep-research\n  - Complexity: moderate\n  - Steps: (1) Clarify goal with 1-2 questions, (2) Break topic into 3-5 sub-questions, (3) Execute multi-source search with keyword variations, (4) Deep-read 3-5 key sources, (5) Synthesize structured report with inline citations, (6) Save and deliver\n  - Quality gates: Every claim sourced, cross-referencing required, recency preference, gap acknowledgment, no hallucination\n  - Output template: Executive summary, themed sections with citations, key takeaways, sources list, methodology notes\n\n## Plugin Ideas\n\nNone directly. The DDG search dependency is ClawHub-specific infrastructure, not a babysitter plugin pattern.\n\n## Library Mapping\n\n| Extractable Process | Library Status | Action | Existing Path | Target Placement |\n|-------------------|----------------|--------|---------------|------------------|\n| Deep Research Methodology | NEW | 6-step research workflow: clarify → plan sub-questions → multi-source search → deep-read → synthesize → deliver | - | specializations/shared/deep-research-methodology.js |\n| Sub-Question Decomposition | NEW | Breaking research topics into 3-5 orthogonal sub-questions for systematic coverage | - | specializations/shared/sub-question-decomposition.js |\n| Multi-Keyword Search Strategy | NEW | 2-3 keyword variations per sub-question with web/news source mixing | - | specializations/shared/multi-keyword-search-strategy.js |\n| Confidence-Scored Report Generation | NEW | Research report synthesis with High/Medium/Low confidence ratings based on source quality | - | specializations/shared/confidence-scored-report-generation.js |\n\n## Plugin Marketplace Mapping\n\n| Plugin Idea | Marketplace Status | Action | Existing Plugin | Target Placement |\n|-------------|-------------------|--------|-----------------|------------------|\n| N/A | N/A | No plugin ideas identified - DDG search dependency is ClawHub-specific | - | N/A |\n\n## Implicit Procedural Knowledge\n\n- **Sub-question decomposition pattern**: Breaking a research topic into 3-5 orthogonal sub-questions before searching is a reusable strategy for any information-gathering process.\n- **Multi-keyword search strategy**: Using 2-3 keyword variations per sub-question, mixing web + news sources, and aiming for 15-30 unique sources provides a template for thorough coverage.\n- **Confidence scoring**: The report includes a High/Medium/Low confidence rating based on source quality and cross-referencing, which maps well to babysitter breakpoint decisions.\n- **Sub-agent delegation pattern**: The SKILL.md includes explicit sub-agent spawn instructions with task description, model selection, and wake-on-completion -- directly relevant to babysitter orchestrator task patterns.\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-reference-repos",
      "to": "page:docs-reference-repos-clawhub-parags-deep-research-pro-research",
      "kind": "contains_page"
    }
  ]
}