NoiseMeld

Custom skills for document extraction and writing

Structured-critique skill that runs ideas past a panel of named personas to push back on decisions, plans, and validation-seeking instead of finding angles to rescue weak ideas. Triggers on "Panel," "Team," a persona name, or any subgroup keyword. Auto-fires when the user is seeking validation. Customize personas.md with three to five real people whose judgment you trust. No Python deps.

Write and edit prose that is clear, forceful, and unrecognizable as machine-generated. Ten editing passes combining Strunk's composition rules with Wikipedia AI-tell detection.

Convert PDF books into structured Markdown with chapter detection, metadata extraction (author, publisher, ISBN, copyright), and image-page rendering. Requires pdfplumber.

Convert research papers (IMRaD format) into structured Markdown with DOI/PMID extraction, section detection, and preserved references. Requires pdfplumber.

Convert video and podcast transcripts into structured Markdown summaries. Accepts YouTube URLs or pasted text. Requires youtube_transcript_api.

Extract web pages into structured Markdown with boilerplate removal. Static HTML and JavaScript-rendered SPAs (React, Vue, Angular). Single pages or full-site crawls. Requires trafilatura and playwright.

Extract web pages into structured Markdown using the obscura headless browser binary with stealth on by default. For sites behind Cloudflare/bot walls where extract-webpage falls down. Single pages or full-site crawls. Requires the obscura binary on PATH plus trafilatura.

Bundle: extract-book + extract-study + extract-transcript + extract-webpage

Structured-critique skill that runs ideas past a panel of named personas to push back on decisions, plans, and validation-seeking instead of finding angles to rescue weak ideas. Triggers on "Panel," "Team," a persona name, or any subgroup keyword. Auto-fires when the user is seeking validation. Customize personas.md with three to five real people whose judgment you trust. No Python deps.

Write and edit prose that is clear, forceful, and unrecognizable as machine-generated. Ten editing passes combining Strunk's composition rules with Wikipedia AI-tell detection.

Convert PDF books into structured Markdown with chapter detection, metadata extraction (author, publisher, ISBN, copyright), and image-page rendering. Requires pdfplumber.

Convert research papers (IMRaD format) into structured Markdown with DOI/PMID extraction, section detection, and preserved references. Requires pdfplumber.

Convert video and podcast transcripts into structured Markdown summaries. Accepts YouTube URLs or pasted text. Requires youtube_transcript_api.

Extract web pages into structured Markdown with boilerplate removal. Static HTML and JavaScript-rendered SPAs (React, Vue, Angular). Single pages or full-site crawls. Requires trafilatura and playwright.

Extract web pages into structured Markdown using the obscura headless browser binary with stealth on by default. For sites behind Cloudflare/bot walls where extract-webpage falls down. Single pages or full-site crawls. Requires the obscura binary on PATH plus trafilatura.

Bundle: extract-book + extract-study + extract-transcript + extract-webpage

Bundle: extract-book + extract-study + extract-transcript + extract-webpage

Bundle: extract-book + extract-study + extract-transcript + extract-webpage

Bundle: extract-book + extract-study + extract-transcript + extract-webpage