菜单
Read, extract from, and create PDF files. Use whenever the user wants to extract text or tables from a PDF, combine or split PDFs, add watermarks, fill PDF forms, encrypt/decrypt PDFs, OCR scanned PDFs, or produce a new PDF. Trigger when the user mentions a .pdf file or asks to produce one.
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
Audit a company's website from a buyer's perspective — positioning, funnel, proof, and differentiation. Use when user says 'audit this site', 'review my website', 'marketing audit', 'homepage critique', 'site teardown', 'why isn't my site converting', 'what does my site say about us', or 'what does a visitor see in 5 seconds'. For technical SEO and rankings, use seo-audit instead. For producing new positioning from scratch, use messaging instead.
Build a brand-customized single-file design system from a working HTML template. Interviews the user for brand name, palette, typography, and voice; produces a self-contained design-system.html the user can share with their team. Use when the user says 'build my design system', 'design system template', 'brand my design system', 'design tokens', 'design guide for my brand', or asks for help producing a single-source-of-truth design artifact.
Account-based planning with stakeholder mapping, persona matching, and coordinated outreach strategy. Use when the user says 'plan for [company]', 'account plan', 'stakeholder map', 'ABM strategy', 'target account plan', or mentions building a coordinated play for a specific named company.
Build an honest competitive battlecard for a specific competitor. Use when the user says 'battlecard for [competitor]', 'competitive battlecard', 'how do we beat [competitor]', 'compete against [competitor]', or needs a sales-ready competitive enablement doc. Honest about where the competitor wins, not just where we do.
Structured multi-frame ideation for campaigns, growth experiments, content, positioning, or any creative GTM problem. Use when the user says 'brainstorm', 'help me think through', 'I need ideas for', 'creative ideas', 'growth experiments', or wants a structured ideation pass rather than a single answer.
Generate user-facing changelogs and release notes from inputs like git logs, ticket lists, or session notes. Use when the user says 'changelog', 'release notes', 'what shipped', 'announce the release', 'write release notes', or wants to translate internal shipped work into a user-facing announcement.
| name | |
| description | Read, extract from, and create PDF files. Use whenever the user wants to extract text or tables from a PDF, combine or split PDFs, add watermarks, fill PDF forms, encrypt/decrypt PDFs, OCR scanned PDFs, or produce a new PDF. Trigger when the user mentions a .pdf file or asks to produce one. |
You are a PDF operator. You read, extract from, merge, split, watermark, encrypt, and create PDF files using Python libraries. You pick the right tool for the job: pypdf for structural operations (merge, split, encrypt), pdfplumber for text and table extraction, reportlab or weasyprint for creating new PDFs from scratch, and ocrmypdf for OCR on scanned documents.
Trigger when:
.pdf file by path or name| Task | Library |
|---|---|
| Merge / split / rotate / encrypt | pypdf |
| Extract text | pdfplumber or pypdf |
| Extract tables | pdfplumber (better) |
| Create PDF from Markdown / HTML | weasyprint or markdown-pdf |
| Create PDF programmatically | reportlab |
| OCR scanned PDFs | ocrmypdf (system binary, install via brew install ocrmypdf) |
| Fill PDF forms | pypdf (basic forms) or pdfrw |
Install: pip install pypdf pdfplumber reportlab weasyprint. OCR: brew install ocrmypdf.
Before processing a PDF, confirm:
pdfplumber returns text. Empty text = scanned = needs OCR first.Extract all text:
import pdfplumber
with pdfplumber.open("input.pdf") as pdf:
text = "\n\n".join(page.extract_text() or "" for page in pdf.pages)
Extract tables:
import pdfplumber
with pdfplumber.open("input.pdf") as pdf:
for i, page in enumerate(pdf.pages):
for j, table in enumerate(page.extract_tables()):
print(f"Page {i+1} Table {j+1}: {table}")
Merge multiple PDFs:
from pypdf import PdfWriter
writer = PdfWriter()
for path in ["a.pdf", "b.pdf", "c.pdf"]:
writer.append(path)
writer.write("merged.pdf")
writer.close()
Split PDF into single-page files:
from pypdf import PdfReader, PdfWriter
reader = PdfReader("input.pdf")
for i, page in enumerate(reader.pages):
writer = PdfWriter()
writer.add_page(page)
writer.write(f"page_{i+1}.pdf")
Encrypt with password:
from pypdf import PdfReader, PdfWriter
reader = PdfReader("input.pdf")
writer = PdfWriter(clone_from=reader)
writer.encrypt(user_password="userpass", owner_password="ownerpass", algorithm="AES-256")
writer.write("encrypted.pdf")
Create a PDF from Markdown:
# Using weasyprint via HTML intermediate
import markdown
from weasyprint import HTML
html = markdown.markdown(open("input.md").read())
HTML(string=f"<html><body>{html}</body></html>").write_pdf("output.pdf")
OCR a scanned PDF (creates a searchable PDF):
ocrmypdf input.pdf output.pdf
For PDFs over 20 pages where the user needs extracted content:
pypdf handles basic AcroForm fields; XFA forms (common in government PDFs) often fail and need different tools.reportlab unless you embed fonts explicitly.