菜单
Handles PDF operations (reading, creating, modifying). Use when extracting text from PDFs, converting markdown to PDF, merging or splitting PDFs, compressing, rotating pages, handling metadata, or when Claude Code's native PDF reading fails.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Delegates a task to OpenAI Codex running as an interactive session in a herdr pane - uses the user's ChatGPT subscription, visible in herdr, steerable mid-session. Use when delegating work to codex, offloading a task to a GPT model (sol, terra, luna), or running a second opinion from a non-Claude frontier model. Not for Claude Code's own Agent tool or local-LLM delegation (delegating-to-local-llm).
Delegates a task to a local LLM running as a Pi coding-agent session in a herdr pane - the subagent is visible in herdr, can be steered mid-session, and costs no cloud tokens. Use when delegating work to a local model, spawning a Pi subagent, offloading a task to omlx (qwen, gemma), or running agent work offline. Not for Claude Code's own Agent tool or subagents inside Pi.
Loads the rageatc-code-oss operating rules directly from the plugin when the SessionStart hook has not delivered them. Use when the rageatc-code-oss failsafe line appears in session-start hook output, when the plugin's operating rules are absent from context, or when the injected rules look truncated.
Loads the rageatc-core-oss operating rules directly from the plugin when the SessionStart hook has not delivered them. Use when the rageatc-core-oss failsafe line appears in session-start hook output, when the plugin's operating rules are absent from context, or when the injected rules look truncated.
Raises an artefact's signal-to-noise ratio by removing text that earns nothing for its reader — filler, repetition, over-explanation, ornamental phrasing — and relocating conditional detail to references. Use to slim a skill, agent, or document, cut bloat, tighten prose, or shrink a file that has grown too long. Tuned for SKILL.md files.
Creates a design system for software with a UI. Use when a project has a user interface and architecture is confirmed — whether creating from scratch or extracting patterns from existing code.
| name | working-with-pdfs |
| description | Handles PDF operations (reading, creating, modifying). Use when extracting text from PDFs, converting markdown to PDF, merging or splitting PDFs, compressing, rotating pages, handling metadata, or when Claude Code's native PDF reading fails. |
Select the tool by operation category, detect what's installed before recommending anything, and keep a fallback ready. Covers macOS, Linux (Debian/Ubuntu), and WSL2; native Windows is out of scope.
# Python tools
python3 -c "import pypdf" 2>/dev/null && echo "pypdf: installed" || echo "pypdf: not found"
python3 -c "import fitz" 2>/dev/null && echo "PyMuPDF: installed" || echo "PyMuPDF: not found"
python3 -c "import pdfplumber" 2>/dev/null && echo "pdfplumber: installed" || echo "pdfplumber: not found"
python3 -c "import weasyprint" 2>/dev/null && echo "WeasyPrint: installed" || echo "WeasyPrint: not found"
python3 -c "import pikepdf" 2>/dev/null && echo "pikepdf: installed" || echo "pikepdf: not found"
# CLI tools
command -v qpdf >/dev/null 2>&1 && echo "qpdf: installed" || echo "qpdf: not found"
command -v pdftotext >/dev/null 2>&1 && echo "poppler: installed" || echo "poppler: not found"
command -v pandoc >/dev/null 2>&1 && echo "Pandoc: installed" || echo "Pandoc: not found"
command -v md-to-pdf >/dev/null 2>&1 && echo "md-to-pdf: installed" || echo "md-to-pdf: not found"
Table extraction from clean, machine-generated PDFs → pdfplumber (pip install pdfplumber). Best table accuracy, visual debugging tool. No OCR — useless on scans.
Python, commercial use anticipated → pypdf (pip install pypdf) — the default recommendation. Pure Python, BSD licence, actively maintained. 10–20× slower than PyMuPDF at scale.
Highest performance, AGPL-compatible project → PyMuPDF (pip install PyMuPDF). Fastest extraction (~0.1s/document), full-featured. AGPL v3.0 — commercial use requires AGPL compliance or a commercial licence from Artifex; always flag this before recommending.
Simple text extraction in shell scripts → poppler/pdftotext (brew install poppler / sudo apt install poppler-utils). Fast; does not preserve layout.
Simple PDF (under ~20 pages, plain text), not a production workflow → try Claude Code's native Read tool first. See Known Claude Code issues below if it errors.
Scanned documents needing OCR → out of scope; investigate Tesseract or OCRmyPDF separately.
Fallbacks: pypdf ↔ PyMuPDF (licence permitting) cover each other; pdfplumber unavailable → PyMuPDF/pypdf for text and parse tables manually; no Python tools at all → poppler/pdftotext.
Node.js ecosystem or minimal dependencies → md-to-pdf (npm install -g md-to-pdf). Lightest option, no LaTeX; renders via Puppeteer (~170MB Chromium). Single file input only — cannot merge files.
CSS-based styling, comfortable with system dependencies → WeasyPrint. Python, BSD licence, strong CSS control, no JavaScript execution. Requires system libraries (Pango, GObject) that cause installation friction — see references/tool-installation.md.
Advanced Markdown features or publication-quality typesetting → Pandoc + BasicTeX. Tables, footnotes, citations, maths; merges multiple files. Requires LaTeX (~1GB BasicTeX, ~6GB full TeX Live).
Frequent compilation, control over output format → Pandoc + Typst (experimental). 27× faster than XeLaTeX, cleaner syntax. Good for internal documents; not for journal submissions.
Avoid wkhtmltopdf — unmaintained, cannot compile with modern toolchains.
Fallbacks: no npm → WeasyPrint or Pandoc; WeasyPrint dependencies fail → md-to-pdf or Pandoc; LaTeX too large → md-to-pdf; nothing installable → Docker approach in references/tool-installation.md.
CLI automation or comprehensive manipulation → qpdf (brew install qpdf / sudo apt-get install qpdf) — the primary recommendation. Merge, split, rotate, compress, encrypt, repair; Apache licence; clean cross-platform installs.
Python workflow, basic operations → pypdf. Split, merge, crop, transform, basic metadata; pure Python, BSD. Cannot create linearised PDFs.
Python workflow, advanced features → pikepdf (pip install pikepdf). qpdf C++ bindings: linearised PDFs, XMP metadata, better validation, faster. Cannot extract text; needs C++ compilation.
Avoid original PDFtk — abandoned (GCJ-based, uncompilable since GCC 7). pdftk-java only when migrating from PDFtk; otherwise prefer qpdf.
Fallbacks: qpdf ↔ pypdf ↔ pikepdf cover each other; if none installable, flag it as an environment issue.
references/tool-installation.md has platform-specific installation commands, verification, troubleshooting, Docker alternatives, and working code examples for every tool above.
pip install pypdf. Older workarounds reference PyPDF2, which is deprecated — use pypdf.Permissive: pypdf (BSD), qpdf (Apache), WeasyPrint (BSD). poppler and Pandoc are GPL, but invoking a GPL CLI tool from a script has no licence impact on the calling code or documents. The one restrictive licence is PyMuPDF (AGPL v3.0) — commercial use requires open-sourcing under AGPL or buying a licence from Artifex.
OCR workflows (Tesseract, OCRmyPDF), PDF/A archival compliance, accessible-PDF standards, digital signatures, advanced encryption, native Windows.