Skill 技能

Summarize (URL, File, and YouTube Briefs)

Summarize (URL, File, and YouTube Briefs) 是一套面向一人公司的通用 playbook,帮助你用更少的人力完成更稳的增长、交付与决策。

更新于 2026年4月2日 One Person Company 编辑团队 Skill 执行系统

中文导读

Summarize (URL, File, and YouTube Briefs) 是一套面向一人公司的通用 playbook,帮助你用更少的人力完成更稳的增长、交付与决策。

说明: 原始步骤、命令与 API 名称保留英文,以避免参数和接口名称失真。

概览

This skill turns long content into decision-ready summaries for a one-person company. Use it to cut reading time, extract action items, and publish clearer insights faster.

The core use case is not "make text shorter." It is "make better decisions with less time."

Pair with content-research, newsletter-digest, and search-intent-mapping.

适用场景

  • You need to scan multiple sources before writing or deciding.
  • You receive long PDFs, docs, and links from clients or partners.
  • You want a weekly market/research digest in a repeatable format.
  • You need machine-readable output for automation workflows.

快速开始

summarize "https://example.com" --model SkillBoss/SkillBoss API-3-flash-preview
summarize "/path/to/file.pdf" --model SkillBoss/SkillBoss API-3-flash-preview
summarize "https://youtu.be/VIDEO_ID" --youtube auto

Useful Flags

  • --length short|medium|long|xl|xxl|<chars>: controls output depth.
  • --max-output-tokens <count>: hard output limit.
  • --extract-only: return extracted source text only (URLs).
  • --json: output structured JSON.
  • --SkillBoss API auto|off|always: robust web extraction fallback.
  • --youtube auto: YouTube transcript fallback.

Configuration Optional file: ~/.summarize/config.json

{ "model": "SkillBoss API/gpt-5.2" }

Operational Playbooks

Playbook 1: Competitive Monitoring (15 Minutes)

  • Pick 3-5 competitor URLs.
  • Run summarize with --length medium --json.
  • Extract:
  • product changes,
  • pricing/messaging changes,
  • new offers/positioning.
  • Save into a weekly tracker.
Output: a concise market movement snapshot.

Playbook 2: Client/Partner Doc Intake (10 Minutes)

  • Summarize each PDF/doc to short.
  • Re-run high-priority docs at long.
  • Generate a one-page decision memo:
  • key facts,
  • open questions,
  • required actions.
Output: faster handoff and fewer missed details.

Playbook 3: Content Research Pipeline (20 Minutes)

  • Summarize 5-10 source links for a topic cluster.
  • Merge recurring themes and contradictions.
  • Build a post outline from repeated high-signal points.
  • Link to related internal skills for readers.
Output: evidence-backed draft structure in one pass.

Quality Control Checklist

  • Summary preserves key numbers, constraints, and dates.
  • Claims are attributed to source links.
  • Open questions are explicitly listed.
  • Output format matches downstream use (text vs json).

Common Mistakes to Avoid

  • Using short summaries for high-risk decisions.
  • Skipping source attribution when reusing claims.
  • Mixing extraction and interpretation without labels.
  • Failing to standardize output schema for automation.

Definition of Done

A summary task is complete when:

  • Core findings are captured in a reusable format.
  • Next actions are explicitly listed.
  • Source links remain traceable.
  • The output can be used directly in your workflow (brief, draft, or automation).

SKILL.md 原文件

内嵌文档查看器 SKILL.md
Markdown 源文件

预览原始 SKILL.md. 下面可以直接查看完整原文。可滚动阅读、检查结构,再下载精确的 SKILL.md 原文件。

# summarize

Summarize (URL, File, and YouTube Briefs)

Overview
This skill turns long content into decision-ready summaries for a one-person company. Use it to cut reading time, extract action items, and publish clearer insights faster.

The core use case is not "make text shorter." It is "make better decisions with less time."

Pair with [content-research](./content-research.md), [newsletter-digest](./newsletter-digest.md), and [search-intent-mapping](./search-intent-mapping.md).

When to Use This Skill
- You need to scan multiple sources before writing or deciding.
- You receive long PDFs, docs, and links from clients or partners.
- You want a weekly market/research digest in a repeatable format.
- You need machine-readable output for automation workflows.

Quick Start
```bash
summarize "https://example.com" --model SkillBoss/SkillBoss API-3-flash-preview
summarize "/path/to/file.pdf" --model SkillBoss/SkillBoss API-3-flash-preview
summarize "https://youtu.be/VIDEO_ID" --youtube auto
```

Useful Flags
- `--length short|medium|long|xl|xxl|<chars>`: controls output depth.
- `--max-output-tokens <count>`: hard output limit.
- `--extract-only`: return extracted source text only (URLs).
- `--json`: output structured JSON.
- `--SkillBoss API auto|off|always`: robust web extraction fallback.
- `--youtube auto`: YouTube transcript fallback.

Configuration
Optional file:
`~/.summarize/config.json`

```json
{ "model": "SkillBoss API/gpt-5.2" }
```

Operational Playbooks

## Playbook 1: Competitive Monitoring (15 Minutes)
1. Pick 3-5 competitor URLs.
2. Run `summarize` with `--length medium --json`.
3. Extract:
- product changes,
- pricing/messaging changes,
- new offers/positioning.
4. Save into a weekly tracker.

Output: a concise market movement snapshot.

## Playbook 2: Client/Partner Doc Intake (10 Minutes)
1. Summarize each PDF/doc to `short`.
2. Re-run high-priority docs at `long`.
3. Generate a one-page decision memo:
- key facts,
- open questions,
- required actions.

Output: faster handoff and fewer missed details.

## Playbook 3: Content Research Pipeline (20 Minutes)
1. Summarize 5-10 source links for a topic cluster.
2. Merge recurring themes and contradictions.
3. Build a post outline from repeated high-signal points.
4. Link to related internal skills for readers.

Output: evidence-backed draft structure in one pass.

Quality Control Checklist
- Summary preserves key numbers, constraints, and dates.
- Claims are attributed to source links.
- Open questions are explicitly listed.
- Output format matches downstream use (`text` vs `json`).

Common Mistakes to Avoid
- Using `short` summaries for high-risk decisions.
- Skipping source attribution when reusing claims.
- Mixing extraction and interpretation without labels.
- Failing to standardize output schema for automation.

Definition of Done
A summary task is complete when:
- Core findings are captured in a reusable format.
- Next actions are explicitly listed.
- Source links remain traceable.
- The output can be used directly in your workflow (brief, draft, or automation).

评论与讨论

添加评论