storm-generate

Orchestrates the full STORM pipeline to generate Wikipedia-style articles from research through polish.

588|49|Updated Aug 4, 2025
One-click install
npx skills add https://github.com/FradSer/dotclaude --skill storm-generate
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: storm-generate
Source: https://github.com/FradSer/dotclaude/tree/main/storm/skills/generate
Command: npx skills add https://github.com/FradSer/dotclaude --skill storm-generate

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Writing a well-researched, long-form article requires coordinating multiple stages—research, outlining, drafting, and polishing—which is tedious to manage manually and easy to lose progress on when interrupted.

Core Features & Use Cases

  • End-to-End Pipeline Orchestration: Runs the four STORM phases (research, outline, write, polish) sequentially, invoking each phase skill in order.
  • Resumable Execution: Skips phases whose artifacts already exist, so re-running after a failure resumes from the first incomplete phase instead of starting over.
  • Flexible Grounding and Output: Supports local document grounding via --docs, retriever selection (mcp, web, or local), and explicit or persistent output directories.
  • Use Case: Ask to "write a Wikipedia-style article about quantum error correction" and receive a polished, cited article with a run-config.json tracking every phase's status.

Quick Start

Ask the assistant to generate a STORM article on your chosen topic, optionally specifying the number of perspectives, output directory, or local documents to ground the research.

Frequently Asked Questions about storm-generate

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

FAQPage Schema
How do I generate a Wikipedia-style article with STORM?

Invoke /storm:generate with a topic, and it runs research, outline, write, and polish phases in sequence. The final output is article-polished.md with cited sources, plus a summary of word count and section count.

How do I resume a failed STORM article generation run?

Re-invoke the skill without the --force flag. It checks each phase's completion artifact and skips finished phases, resuming from the first incomplete one. Use --force only to re-run everything from scratch.

Can I ground STORM article generation on local documents?

Yes, pass --docs DIR to ground research on local documents alongside web retrieval, or --docs-only to use local files exclusively instead of web sources.

What retriever options does STORM support for research?

The --retriever flag accepts mcp, web, or local, with mcp as the default and automatic fallback behavior defined by the storm-engine skill. This controls where the research phase gathers source material.

Where does the generated STORM article get saved?

By default output goes to a temporary directory whose path is surfaced at the end. Use --save to persist to docs/storm/<slug>/ or --output-dir PATH for an explicit location; these two options are mutually exclusive.