mcpField
S
summarize

summarize

github.com/bjesuiterProductivity072 installsUpdated 1w ago
Repository

About summarize

summarize is a CLI skill for extracting or summarizing web pages and local files, including PDFs and YouTube links. It provides controls for output length, token limits, extraction-only results, and JSON formatting.

Key features

  1. 01Summarize web and local content
  2. 02Control summary length and token limits
  3. 03Extract content without summarizing
  4. 04Format output as JSON
  5. 05Report retrieval limitations

Use cases

  1. 01Create a short summary of a web page
  2. 02Extract text from a local PDF
  3. 03Retrieve page content without summarization
  4. 04Process a YouTube link

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/bjesuiter/skills --skill summarize

After installing, type/summarizein Claude Code to confirm it loaded.

Details

From the project README

Summarize

Use summarize to extract or summarize web and local content.

Workflow

  1. Confirm the input is a URL or an accessible local path.
  2. Run summarize <input> with the configured/default model.
  3. Add only the controls needed for the requested output.
  4. Return the summary or extracted content, noting any retrieval limitation.
summarize "https://example.com"
summarize "/path/to/document.pdf"
summarize "https://youtu.be/VIDEO_ID" --youtube auto

Do not select or recommend a specific model unless the user requests one. If they do, inspect the installed CLI for currently supported model syntax before passing --model.

Output controls

  • Use --length short|medium|long|xl|xxl|<chars> to control summary size.
  • Use --max-output-tokens <count> for a hard output-token limit.
  • Use --extract-only to retrieve URL content without summarizing it.
  • Use --json when structured, machine-readable output is useful.
summarize "https://example.com" --length short
summarize "/path/to/document.pdf" --max-output-tokens 1200 --json
summarize "https://example.com" --extract-only

Authentication and extraction

Ensure the API key required by the configured model provider is available in the environment. Do not print or expose keys.

For difficult or blocked pages, use the optional extraction service only when configured:

  • Set FIRECRAWL_API_KEY, then use --firecrawl auto|off|always.
  • Set APIFY_API_TOKEN when YouTube processing needs the Apify fallback.

Live discovery

Treat installed CLI help as authoritative because flags, providers, and models can change:

summarize --help
summarize --help | rg -i 'model|provider|extract|youtube|firecrawl'

Comments

Markdown supported