research-assistant

Deep research and synthesis assistant. Use when you need to investigate a topic across multiple sources, compare findings, produce structured research reports, or do multi-step web research with citation tracking. NOT for simple single-query lookups (use web_search directly).

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "research-assistant" with this command: npx skills add terrycarter1985/research-assistant-deep

Research Assistant

Structured multi-source research with citation tracking and synthesis.

When to Use

  • Multi-step investigation across 3+ sources
  • Comparative analysis requiring side-by-side findings
  • Reports needing citations and source attribution
  • Topic exploration where depth matters more than speed

Workflow

1. Scope

Clarify the research question. Break broad topics into 2-5 sub-questions. Define the output format upfront (brief, detailed, table, report).

2. Search

Run 2-3 parallel web_search calls with varied query phrasing per sub-question. Use freshness and country filters when recency or region matters.

3. Fetch & Extract

For each promising result, use web_fetch with extractMode: "markdown" and maxChars: 8000. Skip paywalled or low-value pages quickly.

4. Synthesize

Cross-reference findings across sources. Flag contradictions explicitly. Assign confidence levels:

LevelMeaning
High3+ independent sources agree
Medium2 sources or 1 authoritative source
LowSingle unverified source

5. Cite & Deliver

Every factual claim gets a source line: [N] Source Title — URL

Output format:

## [Topic]

### Key Findings
- Finding 1 [1]
- Finding 2 [2,3]

### Contradictions
- Source A says X [1], Source B says Y [2]

### Sources
[1] Title — https://...
[2] Title — https://...

Tips

  • Vary queries: rephrase, use synonyms, add site: filters
  • Go deep on one source before moving on: don't skim 20 pages
  • Track dead ends: note queries that returned nothing to avoid repeating
  • Set maxChars conservatively: 4000-8000 per fetch; increase only if needed
  • Parallelize searches, serialize fetches (respect rate limits)

Advanced Patterns

For multi-phase research, see references/deep-dive.md.

For automated source credibility scoring, see scripts/credibility.py.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Research

Web Research

Conduct structured web research by searching, fetching, and synthesizing information into reports with citations and source verification.

Registry SourceRecently Updated
1910Profile unavailable
Research

Academic Citation Manager

Add real references and standardize citations for research papers and theses (为科研论文和毕业论文添加真实参考文献并规范引用标注)

Registry SourceRecently Updated
1.2K2Profile unavailable
Research

LaunchFast Full Research Loop

Complete Amazon FBA product research pipeline using the LaunchFast MCP. Runs product research, IP checks, supplier sourcing, and PPC keyword research in sequ...

Registry SourceRecently Updated
6300Profile unavailable
Research

Initiation of Coverage Agent

依托东方财富数据库,面向沪深京港美上市公司(A 股、港股、美股、北交所),生成首次覆盖报告或深度研究报告。 当用户提及「首次覆盖」「首次覆盖报告」「初始覆盖」「深度研究」「深度报告」「个股深度」「公司深度」「标的深度分析」,或要求对某只股票/上市公司撰写全面、体系化的研究报告时,必须触发本 Skill。 即使用户...

Registry SourceRecently Updated
7.8K2Profile unavailable