Exécutez n'importe quel Skill dans Manus
en un clic

Exécutez n'importe quel Skill dans Manus en un clic

markitdown

Étoiles2

Forks0

Mis à jour10 juin 2026 à 07:18

Convert files to Markdown via Microsoft's markitdown (uvx). Use for PDFs, DOCX, PPTX, XLSX, images, audio, HTML, CSV, JSON, XML, ZIP, YouTube URLs, EPubs. Triggers: /markitdown, "convert this file to markdown", "extract text from this PDF", "read this DOCX", "transcribe this audio". Skip for PDF manipulation (merge, split, rotate, fill forms, create) — use pypdf/reportlab; web page scraping — use firecrawl; DOCX editing or form-filling; files Claude can read natively.

Installation

Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.

Exécuter dans Manus

Source

vadirn

vadirn/nix

Ouvrir le dépôt GitHub Voir les dépôts du créateur

Téléchargement

Exécuter dans Manus

Métiers associésSOC

Basé sur la classification professionnelle SOC

Développeurs de logicielsProfessions informatiques et mathématiques·SOC 15-1252

SKILL.md

readonly

name

markitdown

description

markitdown

Converts files to Markdown via uvx.

Usage

uvx --with 'markitdown[all]' markitdown <file> -o output.md

Reads from stdin too:

cat file.docx | uvx --with 'markitdown[all]' markitdown

Supported formats

PDF, DOCX, PPTX, XLSX, XLS, images (EXIF + OCR), audio (EXIF + transcription), HTML, CSV, JSON, XML, ZIP (iterates contents), YouTube URLs, EPubs.

When to use

User wants to read content from a binary file (PDF, DOCX, PPTX, XLSX, images, audio, etc.)
User provides a YouTube URL and wants a transcript
User wants to extract text/tables from a file for analysis

Route elsewhere for

PDF manipulation (merge, split, rotate, fill forms, create) — use pypdf/reportlab with uv run
Web page scraping — use firecrawl
Files Claude can read natively (plain text, code, Markdown) — read directly

Plus depuis ce dépôt

même dépôt

changes

vadirn/nix

Summarize the current diff per file as markdown: a `## <filepath>` heading per changed file followed by a prose summary of its changes. Triggers: /changes, "summarize the diff", "what changed per file", "per-file summary of changes", "суммаризируй дифф", "что изменилось по файлам". Skip for reviewing quality or bugs (use /code-review) and for raw diff output (plain `git diff`).

2026-07-102

distill

vadirn/nix

Compress an expository note into readable prose backed by a certified Glossary via a standalone CLI — abstractive idea-compression, not extractive trimming. Re-expresses the note as flowing prose (carrying the thesis and the relations among terms) above an optional `## Workflow` checklist of the note's actionable directives and a definitions-only `## Glossary` table, collapsing restatements to one entry each and keeping only operational tokens verbatim; a different model fidelity-grades the glossary definitions and the workflow steps against the source and surfaces anything that did not translate. Replaces /cut. Use on /distill, "distill this", "compress this note", "make a glossary of this note", "tighten this into its ideas", "summarize as a glossary", "compress this guide/procedure into steps", "this note is too long/verbose", «дистиллируй», «сократи в глоссарий», «выжимка», «сделай глоссарий», «сократи гайд в шаги»; route whole-repo glossary maintenance to /glossary.

2026-07-102

opinion

vadirn/nix

Route the current context to one expert persona and return that lens's candid second opinion. Auto-classifies to the best-fit persona, or name one explicitly. Runs the persona as a spawned subagent so its reasoning stays out of the main context. Roster spans engineering, product, go-to-market, and craft. Out of scope: generating new designs (use /design), arguing both sides (use /debate), stress-testing one plan (use /probe), scoring confidence in a claim (use /grade). Invoke explicitly with /opinion.

2026-07-052

vault

vadirn/nix

Personal knowledge management in an Obsidian vault. Use whenever the user wants to save, find, review, or organize knowledge, even without saying "vault". Triggers: saving links/articles as references, distilling concepts into cards, writing original notes, searching or quizzing saved content, listing active projects ("what am I working on"), explicit /vault commands, weekly-log operations (backlog/бэклог, planning, completing tasks, sleep). Excludes direct file edits not routed through /vault (editing a .md file in a code repo that is not a vault artifact), Obsidian app features (kanban, canvas, plugins, .base), web search. Skip for session save/resume ("wrapping up", "where did we leave off") — use /track.

2026-07-052

work

vadirn/nix

Orchestrate a multi-step task by delegating to fresh subagents, keeping the orchestrator session small. Triggers: /work <task>, "orchestrate this", "delegate this work". Skip one-shot questions, single-file edits, and the ambient "let's work on X" / "start working on X" phrasing.

2026-07-042

experiment

vadirn/nix

Test an existing thing's behavior against a falsifiable claim and capture the result as a standalone Markdown record in the current working directory. Self-contained — no vault, no external templates, no project linking. Triggers: /experiment, "run an experiment", "test this claim", "verify whether", "does X actually", "check if X works", "is X true", "falsifiable claim".

2026-06-282

name

markitdown

description

markitdown

Converts files to Markdown via uvx.

Usage

uvx --with 'markitdown[all]' markitdown <file> -o output.md

Reads from stdin too:

cat file.docx | uvx --with 'markitdown[all]' markitdown

Supported formats

PDF, DOCX, PPTX, XLSX, XLS, images (EXIF + OCR), audio (EXIF + transcription), HTML, CSV, JSON, XML, ZIP (iterates contents), YouTube URLs, EPubs.

When to use

User wants to read content from a binary file (PDF, DOCX, PPTX, XLSX, images, audio, etc.)
User provides a YouTube URL and wants a transcript
User wants to extract text/tables from a file for analysis

Route elsewhere for

PDF manipulation (merge, split, rotate, fill forms, create) — use pypdf/reportlab with uv run
Web page scraping — use firecrawl
Files Claude can read natively (plain text, code, Markdown) — read directly