source-extract

name	source-extract
description	Fetches a web page and extracts structured content — key facts, code examples, definitions, and best practices

Source Extract

Fetch a web page and extract the most valuable content into clean, structured markdown notes.

Inputs

• {{SOURCE_URL}} — URL of the web page to extract content from.
• {{OUTPUT_PATH}} — File path where the cleaned markdown will be written.

Process

1. Fetch the source.

   • Use WebFetch to retrieve the content at {{SOURCE_URL}}.
   • If the fetch fails (e.g., 404, paywall, bot-blocked), write a note to the output path indicating the failure and the reason.

2. Identify valuable content.

   • Locate the primary article or documentation body.
   • Identify: definitions, explanations, code examples, configuration snippets, architectural diagrams (describe them), best practices, warnings, and important notes.
   • Ignore: navigation menus, sidebars, ads, cookie banners, footers, related articles, comment sections, social sharing buttons.

3. Structure the extraction.

   • Organize content under clear headers that reflect the source structure.
   • Preserve code blocks exactly as they appear, with language annotations.
   • Convert important callouts or warnings into blockquotes.
   • Preserve tables if they contain valuable data.
   • For any quoted content, provide proper attribution.

4. Write the output.

Output Format

Write a markdown file to {{OUTPUT_PATH}} with this structure:

# [Page Title]

**Source:** [URL]
**Extraction ID Prefix:** EXT-[N] (where N is the source number from the manifest)
**Extracted:** [ISO 8601 timestamp]

## Summary

[2-3 sentence summary of what this page covers and its key takeaway.]

## Key Facts

Each fact gets a unique extraction ID for downstream traceability:

- `EXT-[N]-fact-1`: [First important discrete fact from the source]
- `EXT-[N]-fact-2`: [Second important discrete fact]
- `EXT-[N]-fact-3`: [Third important discrete fact]
- ...continue numbering sequentially

## Definitions

[Only if the source contains important term definitions]

- `EXT-[N]-def-1`: **Term** — Definition
- `EXT-[N]-def-2`: **Term** — Definition

## Code Examples

[Only if the source contains code. Preserve original code blocks.]

### `EXT-[N]-code-1`: [Description of what this code does]

\`\`\`language
code here
\`\`\`

## Patterns and Best Practices

- `EXT-[N]-pattern-1`: [Architectural pattern or recommended approach]
- `EXT-[N]-pattern-2`: [Configuration guideline]

## Important Warnings

- `EXT-[N]-warn-1`: [Caveat, deprecation notice, or security concern]

The EXT-[N]-* IDs create a provenance chain: the research skill can cite EXT-2-fact-3 to trace a claim back to its exact origin in this extraction.

Guidelines

• Be selective. Extract only content that has educational or practical value.
• Do not paraphrase code. Preserve code blocks exactly.
• If the page is primarily a reference table or API doc, structure the output to match (use tables).
• Omit sections from the template that have no relevant content from this source.
• Keep the Summary concise. The bulk of value should be in the structured sections.
• If the source is very long, focus on the most important and unique content rather than trying to capture everything.
• Always include the source URL and extraction timestamp at the top.