discussion-researcher

Generates event research reports from Top-K event cards using one web-search-enabled model interaction.

5|1|Updated Jul 31, 2026
One-click install
npx skills add https://github.com/shiker1996/wechat-editroom --skill discussion-researcher-shiker1996
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: discussion-researcher
Source: https://github.com/shiker1996/wechat-editroom/tree/main/skills/discussion-researcher
Command: npx skills add https://github.com/shiker1996/wechat-editroom --skill discussion-researcher-shiker1996

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Editors working on trending topics need structured research material—anomalies, stakeholder conflicts, and cross-event relationships—before writing, but gathering and organizing this manually across sources is slow and inconsistent. ## Core Features & Use Cases - Intra-event analysis: Identifies anomalies (expectation vs. observation gaps), interest conflicts among stakeholders, and directions worth further questioning for each Top-K event. - Inter-event relationship mapping: Establishes temporal, response, comparison, trend, and counterexample relationships between events using batch event indexes and native web search. - Source-grounded output: Every judgment is bound to numbered sources (S1, S2, ...) with titles, URLs, and summaries, and unsupported sections are explicitly marked as pending verification. - Use Case: Given a batch of clustered hot events with source summaries, produce a Markdown research report per event that downstream writing and editorial review stages can consume directly. ## Quick Start Provide the Top-K event cards, source summaries, and batch event index, then ask the assistant to generate the event research report for a selected event.

Frequently Asked Questions about discussion-researcher

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

FAQPage Schema
How do I generate research material for trending news events with AI?

Provide the Top-K event cards, source summaries or URLs, and the batch event index as input. The skill runs one model interaction with native web search and returns a Markdown report covering anomalies, conflicts, and event relationships.

What types of event relationships can this analysis detect?

It detects five relationship types: temporal sequence, response, comparison, trend, and counterexample. Each requires concrete differences in time, action, or outcome—shared keywords or subjects alone are not sufficient.

Does the event research report verify facts from sources?

No. The report records only source titles, URLs, and search summaries as research material. Full-text verification is intentionally left to the downstream editorial stage, and unverified items are marked as pending.

Can the skill run multiple search rounds per event?

No. Each event receives exactly one complete model interaction with a budget of one model step and one tool call. It returns the final Markdown report directly without intermediate search plans or follow-up phases.

What happens when there is not enough evidence for a section?

The report keeps the section heading and writes that there is insufficient basis rather than inventing content. Judgments without bound sources are never stated as confirmed conclusions.