# Bilibili Favorites

> B站收藏夹智能分类与同步。触发词：B站收藏、收藏夹整理、bilibili收藏分类、收藏夹分组、视频收藏整理

- **Type:** Skill
- **Install:** `agentstack add skill-sunrisever-bilibili-favorites-bilibili-favorites`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [sunrisever](https://agentstack.voostack.com/s/sunrisever)
- **Installs:** 0
- **Category:** [Content & Media](https://agentstack.voostack.com/c/content-and-media)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [sunrisever](https://github.com/sunrisever)
- **Source:** https://github.com/sunrisever/bilibili-favorites

## Install

```sh
agentstack add skill-sunrisever-bilibili-favorites-bilibili-favorites
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# B站收藏夹分类助手

AI 辅助整理 B站收藏夹，完成采集、规则生成、分类审核与收藏夹同步。

## 触发关键词

B站收藏、收藏夹整理、bilibili收藏分类、收藏夹分组、视频收藏整理、整理收藏夹

## 适用场景

- 收藏的视频已经很多，想按主题重组
- 希望先自动分类，再人工复核低置信度项
- 想把本地分类结果同步回 B站收藏夹结构

## 运行前提

- 在 `data/config.json` 中配置 B站 Cookie
- 在 `data/config.json` 中配置用于分析/审核的 LLM API
- 运行环境默认使用 Anaconda `base`
- 安装依赖：`bilibili-api-python>=17.4.1`、`anthropic>=0.18.0`、`httpx>=0.24.0`

## 数据与隐私

- 账号 Cookie、API Key、原始收藏数据都应留在本地
- 建议把敏感配置放在 `local/` 或未纳入版本控制的 `data/config.json`
- 发布前按 [RELEASE_CHECKLIST.md](RELEASE_CHECKLIST.md) 检查隐私与示例数据

## 标准工作流

### 1. 采集收藏数据

```bash
python fetch.py all
python fetch.py resume
python fetch.py stats
```

输出：

- `data/收藏视频数据.json`
- `data/fetch_checkpoint.json`

### 2. 生成分类规则

```bash
python analyze.py
python analyze.py summary
```

输出：

- `data/classify_rules.json`

### 3. 导入 UP 主分类映射

```bash
python import_up_map.py
python import_up_map.py "path/to/bilibili-follow-project"
```

输出：

- `data/up_classify_map.json`

### 4. 三阶段分类

```bash
python classify.py
python classify.py algo
python classify.py ai
python classify.py review
```

输出：

- `data/分类结果.json`

### 5. 生成可读文档

```bash
python generate_info.py
```

输出：

- `data/视频信息汇总.txt`
- `data/分类结果.md`

### 6. 预览并同步到 B站

```bash
python sync.py --dry-run
python sync.py
```

## 辅助脚本

### 增量处理新收藏

```bash
python add_new.py
python add_new.py --days 30
```

### 恢复丢失视频

```bash
python recover.py --dry-run
python recover.py
```

## 风险提醒

- `sync.py` 会重建非默认收藏夹，必须先 `--dry-run`
- AI 审核会消耗模型额度
- Cookie 过期或 B站风控会导致同步失败
- 这个项目与 `bilibili-follow` 配合使用效果更好：一个整理关注，一个整理收藏

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [sunrisever](https://github.com/sunrisever)
- **Source:** [sunrisever/bilibili-favorites](https://github.com/sunrisever/bilibili-favorites)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** yes
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-sunrisever-bilibili-favorites-bilibili-favorites
- Seller: https://agentstack.voostack.com/s/sunrisever
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.