ktncodes/ideaverse-os/web-clip-report

web-clip-report

Process a raw web clip saved by Obsidian Web Clipper (or written by ideaverse-capture) into a standardized vault research note. Rewrites the file in-place, preserving the original source URL.

Path resolution

Read 00-agentic-OS/paths.json for vaultRoot. All paths are relative to it.

Optionally read 00-agentic-OS/user.md if available -- it provides context for the "So What for You" section.

Step 1: Read the source file

Read the file at {vaultRoot}/40-raw/web-clips/$ARGUMENTS.

If the file does not exist, halt: "File not found: 40-raw/web-clips/{arg}. Check the filename and try again."

Preserve from the original frontmatter:

• source or source_url field (the original URL -- never lose this)
• author field if present
• published date if present
• title if present (use as fallback title if no better title is found in the body)

Step 2: Analyze the content

Read the full body. Identify:

• The main argument or thesis
• 5-7 key takeaways
• The content type (article, blog post, documentation, research paper, social post, video transcript, etc.)
• 1-3 open questions the content raises but does not answer
• Source bias or agenda (selling a product? promoting a framework? sharing direct experience? anonymous?)
• Confidence rating:
• High -- primary research, expert practitioner with direct experience, or official documentation
• Medium -- credible analysis or opinion, secondhand evidence, reputable publication
• Low -- speculative, poorly sourced, anonymous, or heavily promotional
• Relevance to the vault owner (read user.md if available for their active work and goals)
• Any external sources, tools, books, or links referenced in the body

Step 3: Rewrite the file

Overwrite the file at {vaultRoot}/40-raw/web-clips/$ARGUMENTS with exactly this format:

---
title: "[Article or page title]"
description: "[One sentence: what this source argues or covers. One sentence: why it matters or what problem it addresses.]"
topic: "[primary-topic-slug]"
source: "[Publication name, site name, or author name]"
source_url: "[original URL]"
source_type: web-clip
date: YYYY-MM-DD
tags: [research, web-clip, topic-tag-1, topic-tag-2]
confidence: Low | Medium | High
related: []
---
 
## TL;DR
 
-[Core argument in one line]
-[Most important implication]
-[Key actionable insight or warning]
 
## Key Takeaways
 
-**[Label]:** [explanation]
-**[Label]:** [explanation]
-**[Label]:** [explanation]
-**[Label]:** [explanation]
-**[Label]:** [explanation]
 
## Open Questions
 
-[Question this content raises but does not answer]
-[Follow-up question worth investigating]
 
## Bias / Agenda Notes
 
[1-2 sentences on source credibility. Note if the author is selling a product, promoting a framework, or has a known slant. If none detected, write "No obvious bias detected."]
 
## So What for You
 
[1-2 sentences connecting this content to the vault owner's active projects, goals, or technical work. Read user.md for context. If user.md is not available, write a general relevance statement based on the topic.]
 
## Sources
 
-[Any external sources, papers, tools, or links referenced in the body.]
-[If none were explicitly mentioned, write "No external sources referenced."]

Quality rules

• Frontmatter first. The description field is the most critical -- an LLM scanning only frontmatter must determine relevance without reading the body.
• ASCII only. No em dashes, curly quotes, ellipsis characters, or Unicode bullets.
• Preserve provenance. Always carry forward the original source URL. Never lose it.
• Do not hallucinate. Only include information present in the clipped content.
• Date field: Use today's date if no clear publication date exists in the original clip.
• No archiving. This skill rewrites in-place. cortex-compile handles moving files to 99-archive/ after compilation.
• "So What for You" reads the vault owner's profile from user.md -- if available, reference their active projects or goals specifically. If not available, keep it generic.