research

Generates research outlines and field definitions through web-search-augmented preliminary investigation.

2.2k|185|Updated Dec 29, 2025
One-click install
npx skills add https://github.com/Weizhena/Deep-Research-skills --skill research-weizhena
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: research
Source: https://github.com/Weizhena/Deep-Research-skills/tree/main/skills/research-codex-en/research
Command: npx skills add https://github.com/Weizhena/Deep-Research-skills --skill research-weizhena

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyyaml.

What problem does it solve? Starting a structured research project requires defining what to investigate and which dimensions to cover, but doing this manually often misses important items or recent developments. This Skill combines model knowledge with live web search to produce a confirmed research outline before deep investigation begins. ## Core Features & Use Cases - Two-Source Outline Generation: Builds an initial framework from model knowledge, then launches a web-search agent to supplement missing items and fields within a user-specified time range. - Human-in-the-Loop Confirmation: Uses request_user_input at every stage to confirm items, field frameworks, time ranges, and execution parameters like batch size. - Structured YAML Output: Produces outline.yaml (items plus execution config) and fields.yaml (field definitions with detail levels) saved to a topic directory, ready for the follow-up /research-deep phase. - Use Case: A researcher comparing AI coding tools runs /research to enumerate tools and evaluation dimensions, confirms the framework interactively, and receives outline.yaml and fields.yaml that drive the subsequent deep research phase. ## Quick Start Run /research followed by your topic, for example asking the agent to research AI coding assistants and confirm each step of the generated outline.

Frequently Asked Questions about research

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

FAQPage Schema
How do I generate a research outline with an AI agent?▼

Run /research followed by your topic. The skill first drafts an item list and field framework from model knowledge, then launches a web-search agent to supplement recent items, and finally saves outline.yaml and fields.yaml after your confirmation.

How to validate JSON research results against fields.yaml?▼

Use the bundled validate_json.py script with --fields pointing to fields.yaml and --dir pointing to the results directory. It reports coverage rate, missing required and optional fields per category, and exits non-zero if any required field is absent.

Does the research skill work with Codex and OpenCode?▼

Yes, separate skill variants exist for Claude Code, OpenCode, and Codex. Codex requires installing the web-researcher agent TOML and enabling multi_agent in config.toml, while OpenCode requires OPENCODE_ENABLE_EXA=1 for web search.

What Python dependency does the research skill require?▼

The skill requires pyyaml, installed via pip install pyyaml. The validation script parses fields.yaml with yaml.safe_load and fails loudly if the file does not match the expected fields-per-category schema.

What are the limitations of the preliminary research phase?▼

The phase only produces the outline and field definitions; it does not perform deep per-item investigation, which is handled by the separate /research-deep command. Web supplementation quality also depends on the configured web-search agent being available.