Skip to content

chase6666/DeepResearch-Skill

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 

Repository files navigation

deep-research

An agent skill for generating deep research reports — multi-section, citation-backed, with interactive charts. Compatible with any agent platform that supports the Skills convention (Cursor, Claude Code, etc.).

Install

npx add-skill https://github.com/chase6666/DeepResearch-Skill --skill deep-research

Or manually copy the deep-research/ folder into your agent's skills directory (e.g. .cursor/skills/).

Usage

Tell your agent what you want in natural language:

Write a deep research report on the future trends of AI Agents

The skill handles the full pipeline automatically: outline → parallel section writing → review loop → chart rendering → report assembly.

Output

Format File Charts Best for
HTML (recommended) output/{topic}/report.html Interactive (ECharts / Mermaid inline) Local viewing, sharing
Markdown output/{topic}/report.md Static PNGs Version control, portability

Prerequisites

  • Python 3 — stdlib only, no pip installs needed
  • Node.js ≥ 18 — for chart rendering

Markdown path only: requires playwright + Chromium to screenshot charts as PNGs.

  • Default: auto-installed on first render (~130 MB, cached in ~/.cache/ms-playwright)
  • Manual:
    cd deep-research/scripts && npm install && npx playwright install chromium
    Set NO_AUTO_INSTALL=1 to disable auto-installs after that.

The HTML path has no extra dependencies.

How it Works

User prompt
    │
    ▼
Planner ──── outline.md
    │
    ▼
Writers (parallel) ──── sections/01_*.md … 07_*.md
    │                    (inline citations, AVR chart blocks)
    ▼
Reviewer ──── CLEF evaluation → revision loop
    │
    ▼
Renderer ──── AVR blocks → images/html/*.html  [→ images/png/*.png]
    │
    ▼
Assembler ── report.html  /  report.md

Sub-agents run in parallel where possible. The Reviewer uses a 5-dimension CLEF framework (Organization, Depth, Relevance, Alignment, Synergy) to evaluate and request targeted revisions before assembly.

License

MIT

About

A general deep research skill for guiding models to automatically generate high-quality multimodal reports.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors