deep-research
Conduct enterprise-grade research with multi-source synthesis, citation tracking, and verification. Use when user needs comprehensive analysis requiring 10+ sources, verified claims, or comparison of approaches. Triggers include "deep research", "comprehensive analysis", "research report", "compare X vs Y", or "analyze trends". Do NOT use for simple lookups, debugging, or questions answerable with 1-2 searches.
npx skills add majiayu000/claude-skill-registry --skill deep-research-bauhaus-infau-infau-skill-base --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
The skill performs complex, citation-backed research intended for analyses requiring many sources and verification. It operates through an eight-phase pipeline (Scope → Plan → Retrieve → Triangulate → Synthesize → Critique → Refine → Package) to deliver a comprehensive report with source credibility scoring and progressive context management.
How it works
- It uses a decision process to determine when complex analysis is needed and selects a mode (Quick, Standard, Deep, UltraDeep) with corresponding time estimates and source counts. It announces the chosen plan and proceeds without waiting for user approval.
- Phase execution includes: SCOPE, PLAN, RETRIEVE (parallel searches and agents), TRIANGULATE, OUTLINE REFINEMENT, SYNTHESIZE, CRITIQUE (for Deep/UltraDeep), and REFINE. Phase 8 concludes with PACKAGE, generating a report from a template.
- It enforces an Anti-Hallucination Protocol: every factual claim must cite a source with explicit markers like "According to [1]..." or "[1] reports..." and avoids fabrication.
- It requires parallel retrieval: decomposing the query into 5-10 angles, launching all searches and agents in a single message, and spawning multiple parallel agents for deep investigation.
- It includes a Verify step with Python scripts to check citations and structure, and a Report step to assemble a comprehensive markdown report, HTML (McKinsey-style), and PDF formats.
When to use it
Use when the user requires enterprise-grade research with 10+ sources, verified claims, or a comparison of approaches. Trigger phrases include "deep research", "comprehensive analysis", "research report", "compare X vs Y", or "analyze trends". Do not use for simple lookups, debugging, or questions answerable with 1-2 searches.
What it can touch
- Tools: claude-code is declared for execution tasks like running scripts and assembling reports. The workflow relies on external scripts (e.g., python scripts for validation) and file systems for storing outputs.
- The process creates and manipulates documents in a structured folder: ~/Documents/[TopicName]Research[YYYYMMDD]/, and generates files named like research_report_[YYYYMMDD]_[topic_slug].md/.html/.pdf, following a specific template. It also opens the generated HTML/PDF in appropriate viewers.
Caveats
- License is MIT. The workflow emphasizes avoiding hallucinations with explicit sourcing and verification steps. It enforces a strict token output cap (32,000 tokens) and a practical recommendation to keep total output around 20,000 words. It supports unlimited length via progressive section writes but requires multi-step runs for very long reports. Citations must be verifiable; if a source cannot be confirmed, it should be marked accordingly.
# Deep Research <!-- STATIC CONTEXT BLOCK START - Optimized for prompt caching --> <!-- All static instructions, methodology, and templates below this line --> <!-- Dynamic content (user queries, results) added after this block --> ## Core System Instructions **Purpose:** Deliver citation-backed, verified research reports through 8-phase pipeline (Scope → Plan → Retrieve → Triangulate → Synthesize → Critique → Refine → Package) with source credibility scoring and progressive context management. **Context Strategy:** This skill uses 2025 context engineering best practices: - Static instructions cached (this section) - Progressive disclosure (load references only when needed) - Avoid "loss in the middle" (critical info at start/end, not buried) - Explicit section markers for context navigation --- ## Decision Tree (Execute First) ``` Request Analysis ├─ Simple lookup? → STOP: Use WebSearch, not this skill ├─ Debugging? → STOP: Use standard tools, not this skill └─ Complex analysis needed? → CONTINUE Mode Selection ├─ Initial exploration? → quick (3 phases, 2-5 min) ├─ Standard research? → standard (6 phases, 5-10 min) [DEFAULT] ├─ Critical decision? → deep (8 phases, 10-20 min) └─ Co
- Core System Instructions
- Decision Tree (Execute First)
- Workflow (Clarify → Plan → Act → Verify → Report)
- 1. Clarify (Rarely Needed - Prefer Autonomy)
- 2. Plan
- 3. Act (Phase Execution)
- 4. Verify (Always Execute)
- 5. Report
- Output Contract
- Error Handling & Stop Rules
- Quality Standards (Always Enforce)
- Inputs & Assumptions
- When to Use / NOT Use
- Scripts (Offline, Python stdlib only)
python scripts/verify_citations.py --report [path] python scripts/validate_report.py --report [path] Extract topic slug from research question Create folder: ~/Documents/[TopicName]_Research_[YYYYMMDD]/ mkdir -p ~/Documents/[folder_name] Create initial markdown file with frontmatter File path: [folder]/research_report_[YYYYMMDD]_[slug].md python ./scripts/md_to_html.py [markdown_report_path] python scripts/verify_html.py --html [html_path] --md [md_path]
What does the deep-research skill do?
Conduct enterprise-grade research with multi-source synthesis, citation tracking, and verification. Use when user needs comprehensive analysis requiring 10+ sources, verified claims, or comparison of approaches. Triggers include "deep research", "comprehensive analysis", "research report", "compare X vs Y", or "analyze trends". Do NOT use for simple lookups, debugging, or questions answerable with 1-2 searches.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill deep-research-bauhaus-infau-infau-skill-base --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.