research
npx machina-cli add skill bigph00t/claude-research-team/research --openclawResearch Skill
Research any topic using multi-agent web search and AI synthesis.
When to Use
USE THIS FOR ALL EXTERNAL LOOKUPS - not WebFetch or WebSearch.
- GitHub repos, libraries, or tools (e.g., "cocoindex data framework")
- Looking up unfamiliar APIs or libraries
- Need documentation or best practices
- Comparing technologies or approaches
- Encountering errors you don't recognize
- Validating technical decisions
- ANY URL you want to understand (research the topic, not just fetch)
Usage
The research service must be running. Check with:
curl http://localhost:3200/api/health
Quick Research (~10s)
For simple facts, definitions, quick lookups:
curl -X POST http://localhost:3200/api/research \
-H "Content-Type: application/json" \
-d '{"query": "What is HTMX?", "depth": "quick", "trigger": "manual"}'
Medium Research (~30s) - Default
For how-to questions, documentation lookup:
curl -X POST http://localhost:3200/api/research \
-H "Content-Type: application/json" \
-d '{"query": "How to implement rate limiting in FastAPI", "trigger": "manual"}'
Deep Research (~60s)
For comprehensive comparisons, complex topics:
curl -X POST http://localhost:3200/api/research \
-H "Content-Type: application/json" \
-d '{"query": "Rust vs Go for CLI tools", "depth": "deep", "trigger": "manual"}'
Parameters
| Parameter | Required | Description |
|---|---|---|
| query | Yes | The research topic or question |
| depth | No | quick, medium (default), or deep |
| context | No | Additional context to focus the research |
Examples
- GitHub Repo:
{"query": "cocoindex data transformation framework - features and use cases"} - API Lookup:
{"query": "Bun SQLite named parameters syntax"} - Error Research:
{"query": "ECONNREFUSED database connection error"} - Comparison:
{"query": "PostgreSQL vs SQLite for local-first apps", "depth": "deep"} - With Context:
{"query": "audio transcription APIs", "context": "comparing quality and latency for real-time use"}
Dashboard
View research activity and findings at: http://localhost:3200
Source
git clone https://github.com/bigph00t/claude-research-team/blob/main/skills/research/SKILL.mdView on GitHub Overview
Research uses multi-agent web search and AI synthesis to gather external information, documentation, and technical insights. It helps validate decisions, compare technologies, and understand unfamiliar APIs or errors by returning structured findings.
How This Skill Works
The research service runs locally and is accessed via HTTP. You submit a JSON payload to /api/research with a query, optional depth (quick, medium, or deep), and optional context. The system performs autonomous web searches and AI synthesis to produce a concise, structured answer.
When to Use It
- External lookups for GitHub repos, libraries, or tools
- Looking up unfamiliar APIs or libraries
- Documentation lookup or best practices
- Comparing technologies or approaches
- Validating technical decisions or diagnosing errors
Quick Start
- Step 1: Ensure the research service is running by checking curl http://localhost:3200/api/health
- Step 2: Send a query to the API, e.g., curl -X POST http://localhost:3200/api/research -H 'Content-Type: application/json' -d '{\"query\": \"What is HTMX?\", \"depth\": \"quick\"}'
- Step 3: Review the results and, if needed, open http://localhost:3200 to view the dashboard or refine your query
Best Practices
- Check the research service health with curl http://localhost:3200/api/health before querying
- Choose depth thoughtfully (quick for fast facts, deep for comparisons) to balance thoroughness and speed
- Provide context to narrow results (e.g., API names, language, or usage scenario)
- Cross-check findings against primary sources and cite URLs when possible
- Review results in the dashboard at http://localhost:3200 to assess findings
Example Use Cases
- GitHub Repo: {\"query\": \"cocoindex data transformation framework - features and use cases\"}
- API Lookup: {\"query\": \"Bun SQLite named parameters syntax\"}
- Error Research: {\"query\": \"ECONNREFUSED database connection error\"}
- Comparison: {\"query\": \"PostgreSQL vs SQLite for local-first apps\", \"depth\": \"deep\"}
- With Context: {\"query\": \"audio transcription APIs\", \"context\": \"comparing quality and latency for real-time use\"}
