skills.rest
DreamMacer
DreamMacerCOMMUNITY
Explore all skills by this author →
GitHub

openJiuwen-DeepSearch

Generates multi-agent deep research reports in Markdown, HTML, and DOCX formats.

Updated Aug 31, 2026
Search Tags:#markdown#mermaid#web search#report generation#multi-agent#docx#deep research
One-click install
npx skills add https://github.com/DreamMacer/jiuwenswarm --skill openjiuwen-deepsearch-dreammacer
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: openJiuwen-DeepSearch
Source: https://github.com/DreamMacer/jiuwenswarm/tree/main/resources/agent/workspace/skills/openJiuwen-DeepSearch
Command: npx skills add https://github.com/DreamMacer/jiuwenswarm --skill openjiuwen-deepsearch-dreammacer

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires openjiuwen-deepsearch, python-dotenv, pypandoc, markdown, requests, pyyaml, pillow, and includes scripts (resource) components.

What problem does it solve? Conducting in-depth research on complex topics requires manually planning queries, searching multiple sources, synthesizing findings, and formatting reports. This Skill automates the entire pipeline through multi-agent collaboration, producing citation-backed research reports in about 15 minutes. ## Core Features & Use Cases - Multi-Agent Deep Research: Coordinates query planning, web information gathering, comprehension, reflection, and report writing agents via the openjiuwen-deepsearch engine. - Multi-Format Report Output: Generates the final report as Markdown, HTML (with Mermaid diagram rendering and citation links), and DOCX (via Pandoc with rendered diagram images). - Background Execution with Process Control: Runs research as a detached background process tracked via PID.info, with verified process-tree termination when the user requests a stop. - Use Case: A financial analyst asks for a report on "the impact of 2025 Fed rate cuts on A-share tech stocks" and receives a structured, cited report in three formats. ## Quick Start Ask the assistant to run the openJiuwen-DeepSearch skill with your research topic, for example: generate a deep research report on 2025 electric vehicle industry trends.

Frequently Asked Questions about openJiuwen-DeepSearch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a deep research report with openJiuwen-DeepSearch?▼

Create a Python 3.11 virtual environment, install openjiuwen-deepsearch with uv, configure your .env file, then run uv run scripts/main.py --mode query --query "your topic". The task runs in the background and outputs Markdown, HTML, and DOCX reports in about 15 minutes.

What API keys does the deep research workflow require?▼

You need an LLM API key (OpenAI or SiliconFlow) and a web search API key (Tavily or Google Custom Search). Both are configured in a .env file that you create from .env.example; the skill never asks for keys in conversation.

What output formats does the research report support?▼

Reports are generated in three formats: Markdown, HTML with rendered Mermaid diagrams and clickable citations, and DOCX converted via Pandoc with Mermaid diagrams rendered as enhanced PNG images.

How do I stop a running deep research task?▼

Ask the assistant to stop the task. It reads the PID from PID.info, verifies the process identity, terminates the entire process tree (taskkill /T on Windows or kill on the process group on Linux/macOS), and confirms exit before reporting success.

Why does the deep search fail with missing environment variable errors?▼

The script validates that LLM_MODEL_NAME, LLM_API_KEY, WEB_SEARCH_API_KEY, and other required variables are set in .env before running. If any are missing or still placeholder values, execution stops and you must edit .env yourself.

Related Skills

Explore More Education & Research Skills→
Umbral-BotUmbral-Bot

research

Search the web and synthesize findings into structured research reports.

Official
Intermediate
DaveSteadmanDaveSteadman
1

WebResearchReport

Render Markdown research analysis into styled, standalone HTML reports.

Community
Intermediate
sakamoto-family-smilesakamoto-family-smile

research-ops

Coordinate ECC research workflows to produce citation-backed recommendations.

Community
Advanced
Supporter09Supporter09

llamaindex

Build retrieval-augmented generation applications over private documents and data sources.

Community
Intermediate