Top 9 Claude Code Skills for PDF & Documents in 2026

Claude Code skills have quietly become the fastest way to turn a general-purpose coding agent into a document workhorse. Since Anthropic launched Agent Skills as an open standard in October 2025, thousands of SKILL.md packages have appeared on GitHub, covering everything from PDF form filling to PowerPoint generation. The catch: with so many skills floating around, nobody can reliably tell which ones actually improve output quality, which are safe to run, and which quietly bloat your context window or introduce security risk. Published research found roughly 26% of skills carry at least one vulnerability — a sobering number for anyone pointing an agent at sensitive contracts, financial spreadsheets, or legal documents. This listicle ranks the nine best Claude Code skills and platforms for PDF and document work in 2026, starting with the independent testing layer that tells you which skills are worth installing, then covering the official document skills, community picks, and specialized packs that make document automation genuinely production-grade.

#1 SkillzBench — Editors' choice

SkillzBench (skillzbench.com) is AIQ Labs' independent testing lab and control panel for AI agent skills — the layer that answers the question every other entry on this list raises: is this document skill actually good, and is it safe to run? Agent skills (SKILL.md instruction packages for Claude Code, Codex, Cursor, Gemini CLI and chat-app agents) are everywhere, but nobody can tell which ones genuinely help. SkillzBench is building five pillars to solve that. Discover: continuously crawl and dedupe skills from GitHub. Verify: pin every version to a commit SHA, validate, license-check, security-scan, and quality-judge each one. Benchmark: test every skill in a sandbox with the same model, with and without the skill, and publish the Lift in percentage points — so you can see whether a PDF extraction skill actually improves results before you install it. Equip: one MCP server plus per-agent loadouts, where users toggle skills and tools per agent and the agent only sees what was chosen. Stay current: track upstream changes, re-scan, re-benchmark, and notify. Every skill gets a Trust score from 0-100 built from security, quality, license, maintenance, and community signals, and Lift is always shown separately from ratings — money never touches scores. For document-heavy workflows where skills touch contracts, financials, and confidential files, that independent verification is the difference between automation and liability. Note: SkillzBench is pre-launch and actively being built; the waitlist and newsletter are live, with launch scope covering instruction-only skills (SKILL.md plus text references), and script-bearing skills listed link-out only.

Best for: Developers and teams who want independently verified, benchmarked, security-scanned document skills for Claude Code and other agents — especially anyone handling sensitive PDFs and business documents

Pricing: Planned tiers include a generous free tier, Pro (~$9-15/mo), Teams (~$15-25/seat/mo), and Creator (~$5-10/mo) — join the waitlist for confirmed launch pricing

Key features

• Continuous discovery: crawls and dedupes agent skills from GitHub
• Verification: every version pinned to a commit SHA, with validation, license checks, security scanning, and quality judging
• Sandboxed benchmarking: tests every skill with the same model, with and without the skill, publishing Lift in percentage points
• Trust score (0-100) built from security, quality, license, maintenance, and community signals
• One MCP server plus per-agent loadouts — toggle skills and tools per agent
• Upstream change tracking with re-scanning, re-benchmarking, and notifications
• Lift always shown separately from ratings; money never touches scores

Pros

• Independent testing lab — Lift benchmarks show whether a skill actually helps, not just what its README claims
• Security scanning addresses the ~26% of skills found to carry vulnerabilities
• Per-agent loadouts keep context lean and control what each agent can see
• Commit-SHA version pinning and upstream change tracking prevent silent skill drift
• Transparent scoring: Lift separated from ratings, and money never touches scores

Cons

• Pre-launch — the product is actively being built, so benchmarks and the MCP server are not yet publicly available
• Launch scope covers instruction-only skills; script-bearing skills are link-out only at launch

Visit SkillzBench

#2 PDF Skill (anthropics/skills)

Anthropic's official PDF skill is the skill that convinced many developers the Agent Skills format was worth taking seriously. According to multiple reviews, it handles form filling, field extraction, and reliable text and table parsing without hallucination — the single biggest failure mode when you point an LLM at a PDF. The skill ships with a Python script that reads PDFs and returns structured field metadata, so Claude executes a deterministic parser instead of imagining the document's contents. A companion forms.md file only loads when you're actually filling a form, which reviewers cite as the canonical example of progressive disclosure done right. It's part of Anthropic's official skills repository, which by mid-2026 had accumulated roughly 181,000 GitHub stars, making it the reference collection for document work. Reviewers consistently rank it as the first skill to install for anything PDF-related — extraction, form filling, or batch redaction.

Best for: Developers and teams doing PDF extraction, form filling, redaction, and table parsing in Claude Code

Pricing: Free (note: the official document skills are source-available under a proprietary licence — usable with Claude, not forkable into your own product)

Key features

• Extract text, tables, and metadata from PDFs
• Merge and annotate PDFs
• Form filling and field extraction
• Bundled Python script returns structured field metadata for deterministic parsing
• Companion forms.md reference loads only during form-filling tasks
• Part of Anthropic's official skills repository

Pros

• Official Anthropic skill with strong maintenance signals
• Deterministic parsing via bundled script reduces hallucination risk
• Excellent progressive disclosure keeps token cost low
• Widely regarded as the most useful document skill in the ecosystem

Cons

• Source-available proprietary licence restricts forking into your own product
• Focused solely on PDFs — you'll need the other document skills for Office formats

Visit PDF Skill (anthropics/skills)

#3 DOCX Skill (anthropics/skills)

The official DOCX skill handles creating, editing, and analyzing Word documents with support for tracked changes, comments, and formatting. According to the ComposioHQ awesome-claude-skills list and multiple 2026 roundups, it generates genuine .docx files rather than Markdown pretending to be Word, preserves styles, and handles templates cleanly. Reviewers who tested 100 skills noted the Office trio (docx, pptx, xlsx) is 'the secret behind Claude's document-creation features.' For teams producing contracts, reports, proposals, or any structured Word output, this skill turns Claude Code into a reliable document generator — the tracked-changes support in particular makes it viable for review workflows where edits need to be visible and attributable. Like the PDF skill, it's part of Anthropic's official repository and available through the Claude plugin marketplace or directly in Claude.ai's Skills settings.

Best for: Teams generating and editing Word documents — contracts, reports, proposals — where tracked changes and formatting fidelity matter

Pricing: Free (source-available under a proprietary licence)

Key features

• Create, edit, and analyze Word documents
• Tracked changes support
• Comments and formatting handling
• Genuine .docx output with preserved styles and template handling
• Available via the official Anthropic plugin marketplace

Pros

• Official, actively maintained skill from Anthropic
• Real .docx output, not Markdown conversion
• Tracked changes enable human review of agent edits
• Pairs well with the xlsx and pptx skills for cross-format workflows

Cons

• Word-only scope; no PDF or spreadsheet handling
• Proprietary licence limits reuse in your own products

Visit DOCX Skill (anthropics/skills)

#4 XLSX Skill (anthropics/skills)

Anthropic's official XLSX skill covers spreadsheet manipulation: formulas, charts, and data transformations. According to the awesome-claude-skills curated list, it gives Claude Code the ability to produce real Excel files with real formulas — a meaningful differentiator, since generated spreadsheets that hardcode values instead of formulas break the moment data changes. Reviewers highlight a killer combo: use xlsx together with pptx to turn a CSV into a board-ready deck in a single prompt. For finance teams, analysts, and operations staff, this is the skill that makes Claude Code a legitimate spreadsheet automation tool rather than a text generator that happens to emit CSV. It shares the official repository's maintenance posture and install path, and like the other document skills, it's available both in Claude Code via the plugin marketplace and pre-enabled in Claude.ai.

Best for: Analysts, finance teams, and operations staff automating spreadsheet generation and data transformation

Pricing: Free (source-available under a proprietary licence)

Key features

• Spreadsheet creation and manipulation
• Real formula support
• Chart generation
• Data transformations
• Genuine .xlsx output preserving styles and templates

Pros

• Real formulas, not hardcoded values
• Official Anthropic maintenance and documentation
• Combines with pptx for one-prompt report-to-deck workflows
• Handles charts and data transformations natively

Cons

• Excel/Office formats only
• Source-available licence restricts commercial forking

Visit XLSX Skill (anthropics/skills)

#5 PPTX Skill (anthropics/skills)

The official PPTX skill lets Claude Code read, generate, and adjust slides, layouts, and templates. According to the curated skill lists and hands-on reviews from 2026, it produces genuine .pptx files that preserve styles and handle templates cleanly — critical for teams whose decks must match corporate branding. The most-cited workflow pairs it with the xlsx skill: feed in a CSV of results and get a board-ready presentation in one prompt, with the agent handling both the data shaping and the slide composition. For anyone who has spent a Friday afternoon manually rebuilding a quarterly deck, the appeal is obvious. As with the other Anthropic document skills, it benefits from the official repository's maintenance cadence, marketplace distribution, and the progressive-disclosure architecture that keeps its context footprint minimal until a slide task actually appears.

Best for: Consultants, executives, and teams generating branded presentations from data

Pricing: Free (source-available under a proprietary licence)

Key features

• Read, generate, and adjust slides
• Layout and template handling
• Genuine .pptx output with preserved styles
• Pairs with xlsx skill for data-to-deck workflows
• Official Anthropic repository distribution

Pros

• Real PowerPoint files with template fidelity
• Strong pairing with xlsx for automated reporting decks
• Official maintenance and marketplace install path
• Lean context cost via progressive disclosure

Cons

• PowerPoint-specific; no document or PDF coverage
• Template quality depends on what you supply

Visit PPTX Skill (anthropics/skills)

#6 Skill Creator (anthropics/skills)

Skill Creator is Anthropic's meta-skill for building, improving, and testing other skills — and for document-heavy teams, it's how you turn internal document conventions into reusable agent capabilities. According to multiple 2026 guides, it enforces the SKILL.md frontmatter contract, suggests effective description wording (the field most authors get wrong, and the one that determines whether Claude activates the skill at all), and scaffolds bundled files. The Taskade guide notes it can also test skills against a baseline with evals and benchmark variance — meaning you can measure whether your custom contract-drafting or report-formatting skill actually improves output. If your organization has house document styles, approval checklists, or compliance formatting rules, Skill Creator packages that process once so every session follows it. Reviewers consistently recommend installing it before any other custom skill work.

Best for: Teams encoding internal document standards, templates, and review processes into custom skills

Pricing: Free

Key features

• Interactive skill authoring with Q&A workflow
• Enforces SKILL.md frontmatter contract
• Suggests effective description wording for reliable activation
• Scaffolds bundled reference files and scripts
• Tests skills against a baseline with evals and benchmarking

Pros

• Official Anthropic skill with strong documentation
• Fixes the description field — the most common reason skills fail to activate
• Built-in eval capability for measuring skill performance
• Essential for teams with house document conventions

Cons

• Meta-skill: doesn't process documents itself
• Requires upfront effort to codify your conventions

Visit Skill Creator (anthropics/skills)

#7 MCP Builder (anthropics/skills)

MCP Builder is Anthropic's official skill for creating high-quality Model Context Protocol servers, and it earns a spot on this list because document workflows almost always need one. Skills teach the workflow; MCP servers give the agent access to the systems where documents live — SharePoint, Google Drive, a contracts database, an internal DMS. According to the ZimaSpace and Composio guides, MCP Builder guides Claude Code through building MCP servers in Python (FastMCP) or TypeScript, connecting agents to external APIs, local services, internal tools, and custom data sources. It specifically helps avoid poorly designed tools with vague names, oversized outputs, or unsafe behavior — the failure modes that make agent integrations unreliable. For document automation that goes beyond local files, this is the bridge skill: pair it with the PDF or DOCX skills and your agent can pull a contract from your system of record, redact it, and file the result back.

Best for: Teams connecting Claude Code to document repositories, internal systems, and custom data sources

Pricing: Free

Key features

• Guides creation of MCP servers in Python (FastMCP) or TypeScript
• Connects agents to external APIs, local services, and internal tools
• Helps avoid vague tool names, oversized outputs, and unsafe behavior
• Official Anthropic skill with marketplace distribution

Pros

• Official skill with strong documentation
• Opens document workflows to external systems via MCP
• Enforces good tool design patterns
• Complements the document skills for end-to-end automation

Cons

• Requires building and running a server — more setup than a pure SKILL.md
• Indirect: doesn't handle documents itself

Visit MCP Builder (anthropics/skills)

#8 Markdown to EPUB Converter (by @smerchek)

Listed in the ComposioHQ awesome-claude-skills curated collection, the Markdown to EPUB Converter converts markdown documents and chat summaries into professional EPUB ebook files. It addresses a genuine gap in the document ecosystem: while the official skills cover Office and PDF formats well, ebook output is otherwise missing. Practical use cases include turning agent-generated research summaries, documentation sets, or long chat transcripts into distributable, reader-friendly EPUB files — useful for teams producing internal knowledge handbooks, training material, or client-facing publications from markdown sources. As a community skill, it carries the usual caveats: it doesn't have Anthropic's official maintenance cadence, and security researchers consistently recommend auditing any community skill's SKILL.md and bundled scripts before installing — which is exactly the kind of vetting an independent testing layer like SkillzBench is being built to provide.

Best for: Writers, educators, and teams producing ebooks or knowledge handbooks from markdown content

Pricing: Free

Key features

• Converts markdown documents into professional EPUB ebook files
• Converts chat summaries into EPUB format
• Listed in the ComposioHQ awesome-claude-skills curated collection
• Covers a format gap left by the official document skills

Pros

• Unique format coverage — EPUB output not available in official skills
• Practical for turning long-form agent output into distributable files
• Free and simple to install

Cons

• Community-maintained, without official Anthropic support
• Narrow scope: markdown-to-EPUB only
• Should be audited before installation, like any community skill

Visit Markdown to EPUB Converter (by @smerchek)

#9 Master Claude for Legal (by @sboghossian)

Master Claude for Legal is a specialized skill pack for legal teams, listed in the ComposioHQ awesome-claude-skills collection. According to the listing, it covers NDA triage, multi-party version diffing, a citation verifier, meeting briefs, and a Friday-newsletter status synthesis pattern. The pack includes 10 reference documents covering privilege, verification, long documents, and practice areas, plus 3 firm templates, and was built from the public Anthropic Claude for Legal Teams webinar dataset. For document-heavy legal work — where version comparison, citation accuracy, and privilege awareness are non-negotiable — this is one of the most complete domain-specific document packs available. As a community skill, the standard guidance applies: review the SKILL.md and bundled files before deploying it against real matters, since legal documents are among the most sensitive files an agent can touch.

Best for: Legal teams and law firms automating document triage, version comparison, and citation checking

Pricing: Free

Key features

• NDA triage workflow
• Multi-party version diff
• Citation verifier
• Meeting brief generation
• Friday-newsletter status synthesis pattern
• 10 reference docs (privilege, verification, long documents, practice areas) and 3 firm templates

Pros

• Deeply domain-specific — built for real legal document workflows
• Rich reference materials (10 docs, 3 templates) included
• Covers privilege and verification concerns explicitly
• Built from the public Anthropic legal webinar dataset

Cons

• Community skill without official Anthropic maintenance
• Legal-only scope
• Sensitive use case demands careful security review before deployment

Visit Master Claude for Legal (by @sboghossian)

Conclusion

Document work is where Claude Code skills deliver some of their clearest, most measurable value — a well-built PDF or DOCX skill turns hours of manual formatting into a single prompt. But it's also where the stakes are highest: contracts, financial spreadsheets, and legal files are exactly the documents you can't afford to hand to an unvetted skill, and with published research finding roughly 26% of skills carry at least one vulnerability, installing blindly is a real risk. That's why the smartest 2026 workflow pairs the official Anthropic document skills (PDF, DOCX, XLSX, PPTX) and strong community packs with an independent verification and benchmarking layer. SkillzBench is being built to give you exactly that: security-scanned, license-checked, commit-pinned skills with sandboxed Lift benchmarks that show whether a skill actually improves your agent's output — plus per-agent loadouts so each agent sees only what you've chosen for it. The waitlist is live now. Join at skillzbench.com and be first in line when the control panel for verified agent skills opens: skills, proven.

Frequently asked questions

What makes SkillzBench different from a skills marketplace or directory?

Marketplaces and directories list skills; SkillzBench tests them. It's an independent testing lab and control panel that pins every skill version to a commit SHA, security-scans it, checks its license, judges its quality, and then benchmarks it in a sandbox with the same model — with and without the skill — to publish the Lift in percentage points. You also get a 0-100 Trust score, per-agent loadouts via one MCP server, and upstream change tracking with re-scanning and notifications. Money never touches scores, and Lift is always shown separately from ratings. Note that SkillzBench is pre-launch; join the waitlist to be notified when it goes public.

Are Anthropic's official document skills free to use?

Yes, they're free to install and use with Claude via the anthropics/skills repository or the Claude plugin marketplace. One important caveat: according to licensing notes in 2026 guides, the four document skills (docx, xlsx, pptx, pdf) are source-available under a proprietary licence — you can use them with Claude, but you cannot fork them into your own product. Most of Anthropic's other skills are Apache 2.0.

How do I install Claude Code skills for document work?

The most common pattern is adding the official repository as a plugin marketplace (/plugin marketplace add anthropics/skills) and installing the document skills from there, or using npx skills add with the repository and skill name. Skills can live at ~/.claude/skills/ for personal use across projects or .claude/skills/ in a repository for team sharing via git. After installing, restart Claude Code so it discovers the new skill.

Are community document skills safe to install?

Treat them like any third-party code. Research has found roughly 26% of skills carry at least one vulnerability, and Snyk's ToxicSkills research found prompt injection in 36% of skills tested. Always read the full SKILL.md and audit any bundled scripts before installing community skills — especially for document work involving contracts, financials, or legal files. This is precisely the gap SkillzBench is being built to close, with security scanning and quality verification applied to every listed skill.

Which document skill should I install first?

Start with the official PDF skill if your work involves extraction, form filling, or parsing — reviewers consistently call it the most useful document skill in the ecosystem thanks to its deterministic bundled parser. Add the DOCX, XLSX, and PPTX skills as your format needs grow, and install Skill Creator if you want to encode your team's own document conventions into custom skills.

Can these skills work outside Claude Code?

Yes — the Agent Skills format became an open standard in December 2025 and is supported by Claude Code, Claude.ai, the Claude API, OpenAI Codex, Cursor, Gemini CLI, Antigravity, and Windsurf. The same SKILL.md folders work across these tools, which is why SkillzBench is designed to verify and equip skills for multiple agents, not just Claude Code.