Skills Index

All skills are located in the skills/ directory and trigger on specific keywords. The most complex skills use directory-style packaging with SKILL.md and optional references/ files.

Skill List

Skill	Triggers	Description
hot-topics	今日热榜, AI热榜, 热门话题, 热点追踪, 今日话题, 热榜	Multi-source trending collector with source ranking and optional ViralKB ingestion
last30days	last30days, /last30days, 深度研究, research topic, 过去30天	Deep research across Reddit/X/YouTube/HN/Polymarket/GitHub, synthesized into one brief
research-brief	先研究一下, 帮我先梳理这个题, research brief, deep brief	Converts a promising topic into a source-backed article brief before drafting
viral-patterns	viral-patterns, 爆款模式, 查找爆款, 找标题公式, 参考爆款, check article structure	ViralKB pattern lookup by keyword
viral-mining	挖掘爆款, viral mining, 发现爆款, 爆款挖掘	Multi-source discovery + ViralKB ingestion
write-article	写文章, 生成文章, create-article, 写篇, 公众号, 小红书	Full pipeline: research -> outline -> draft -> illustrate
article-audit	检查这篇文章是否齐全, 审计这篇文章, article audit, publish ready	Audits one article slug across brief, draft, cover, and inline image artifacts
cover-image	生成封面, create cover, 文章封面, cover image	YouTube thumbnail style personal brand cover
article-illustrate	生成插图, 文章配图, insert images, illustrate article	Auto-analyze + insert illustrations via Images API
image-generation	生成图片, create image, generate image, AI画图, draw	Direct image generation across configured providers
baoyu-imagine	baoyu-imagine, 生成图片, create image, generate image	Compatibility alias for image-generation
package-neat	整理一下, 同步文档, 收尾, release audit, skill audit	Syncs package docs, manifests, and public-facing repository metadata

Skill Orchestration

user: I want to write a WeChat article about Claude Code

→ hot-topics: Research latest Claude Code trends (daily discovery)
→ last30days: Deep research on Claude Code ecosystem (deep investigation)
→ research-brief: Turn the topic into a stronger article brief
→ viral-patterns: Look up Claude-related viral title patterns
→ write-article: Generate full article (with illustrations via image-generation)
→ article-audit: Check whether the article family is publish-ready
→ cover-image: Generate cover image via image-generation

hot-topics vs last30days:

hot-topics — daily discovery: what's trending today across AI?
last30days — deep research: what's the complete picture of one topic over 30 days?

The two are complementary: hot-topics feeds the discovery layer; last30days handles the investigation layer.

Core Tools

tools/nanobanana_client.py — Image generation CLI (OpenAI official, Gemini official, OpenAI-compatible)
tools/article_illustrate.py — Auto-illustration for markdown articles (uses Images API internally)
tools/viral_kb.py — ViralKB database interface
tools/jina_reader.py — URL → markdown fetcher
data/viralkb/ — ViralKB storage (patterns.jsonl + embeddings.npy)

Layout Conventions

skills/<name>/SKILL.md - canonical skill entry point for complex skills
skills/<name>/references/ - extra guidance loaded only when needed
all package skills now use directory-style entries with SKILL.md

Provider Configuration

Set any of these in config/.env:

Provider	Env Variable
OpenAI-compatible Images API	`OPENAI_COMPATIBLE_API_KEY`, `OPENAI_COMPATIBLE_BASE_URL`
OpenAI official Images API	`OPENAI_API_KEY`, optional `OPENAI_BASE_URL`
Google Gemini	`GOOGLE_AI_API_KEY`

Default: OpenAI-compatible > OpenAI official > Google Gemini

Skill List

Skill

Triggers

Description

hot-topics

今日热榜, AI热榜, 热门话题, 热点追踪, 今日话题, 热榜

Multi-source trending collector with source ranking and optional ViralKB ingestion

last30days

last30days, /last30days, 深度研究, research topic, 过去30天

Deep research across Reddit/X/YouTube/HN/Polymarket/GitHub, synthesized into one brief

research-brief

先研究一下, 帮我先梳理这个题, research brief, deep brief

Converts a promising topic into a source-backed article brief before drafting

viral-patterns

viral-patterns, 爆款模式, 查找爆款, 找标题公式, 参考爆款, check article structure

ViralKB pattern lookup by keyword

viral-mining

挖掘爆款, viral mining, 发现爆款, 爆款挖掘

Multi-source discovery + ViralKB ingestion

write-article

写文章, 生成文章, create-article, 写篇, 公众号, 小红书

Full pipeline: research -> outline -> draft -> illustrate

article-audit

检查这篇文章是否齐全, 审计这篇文章, article audit, publish ready

Audits one article slug across brief, draft, cover, and inline image artifacts

cover-image

生成封面, create cover, 文章封面, cover image

YouTube thumbnail style personal brand cover

article-illustrate

生成插图, 文章配图, insert images, illustrate article

Auto-analyze + insert illustrations via Images API

image-generation

生成图片, create image, generate image, AI画图, draw

Direct image generation across configured providers

baoyu-imagine

baoyu-imagine, 生成图片, create image, generate image

Compatibility alias for image-generation

package-neat

整理一下, 同步文档, 收尾, release audit, skill audit

Syncs package docs, manifests, and public-facing repository metadata

Skill Orchestration

user: I want to write a WeChat article about Claude Code → hot-topics: Research latest Claude Code trends (daily discovery) → last30days: Deep research on Claude Code ecosystem (deep investigation) → research-brief: Turn the topic into a stronger article brief → viral-patterns: Look up Claude-related viral title patterns → write-article: Generate full article (with illustrations via image-generation) → article-audit: Check whether the article family is publish-ready → cover-image: Generate cover image via image-generation

hot-topics vs last30days:

hot-topics — daily discovery: what's trending today across AI?

last30days — deep research: what's the complete picture of one topic over 30 days?

The two are complementary: hot-topics feeds the discovery layer; last30days handles the investigation layer.

Core Tools

tools/nanobanana_client.py — Image generation CLI (OpenAI official, Gemini official, OpenAI-compatible)

tools/article_illustrate.py — Auto-illustration for markdown articles (uses Images API internally)

tools/viral_kb.py — ViralKB database interface

tools/jina_reader.py — URL → markdown fetcher

data/viralkb/ — ViralKB storage (patterns.jsonl + embeddings.npy)

Provider

Env Variable

OpenAI-compatible Images API

OPENAI_COMPATIBLE_API_KEY, OPENAI_COMPATIBLE_BASE_URL

OpenAI official Images API

OPENAI_API_KEY, optional OPENAI_BASE_URL

Google Gemini

GOOGLE_AI_API_KEY

Skills Index

Additional Files (1)

Skills Index

Skill List

Skill Orchestration

Core Tools

Layout Conventions

Provider Configuration

Related Skills

<h1 align="center">

Frontend Typescript Linting.mdc

2. Apply Deepthink Protocol (reason about dependencies

Additional Files (1)

Skills Index

Skill List

Skill Orchestration

Core Tools

Layout Conventions

Provider Configuration