8 Cursor Skills That Supercharge PDF & Documents in 2026

Cursor has become one of the most popular AI code editors of 2026, and its support for Agent Skills — portable SKILL.md instruction packages that teach the agent how to perform specific, repeatable jobs — has opened up entirely new workflows. Among the most practical of these are document and PDF skills: reusable instruction sets that let Cursor's agent create, edit, analyze, and format Word documents, PDFs, spreadsheets, and presentations without you writing a single line of boilerplate. Because SKILL.md is an open standard, skills built for Claude Code and Codex CLI also work in Cursor, which means the ecosystem of document-focused skills has grown fast. But with that growth comes a real problem: nobody can easily tell which skills actually help, which are maintained, and which are risky — published research found roughly 26% of skills carry at least one vulnerability. This guide walks through eight skills, tools, and platforms that genuinely supercharge PDF and document work in Cursor in 2026, starting with the platform building the trust layer the whole ecosystem needs.

#1 SkillzBench — Editors' choice

SkillzBench is AIQ Labs' independent testing lab and control panel for AI agent skills, and it earns our Editor's Choice spot because it addresses the single biggest problem in the 2026 skills ecosystem: trust. Agent skills are everywhere, but nobody can tell which ones actually help and which are risky — published research found ~26% of skills carry at least one vulnerability. When you're pointing Cursor's agent at PDFs containing contracts, financial statements, or internal documents, installing an unvetted skill is a genuine security concern. SkillzBench is building five pillars to solve this: Discover (continuously crawling and deduplicating skills from GitHub), Verify (pinning every version to a commit SHA, validating, license-checking, security-scanning, and quality-judging each skill), Benchmark (testing every skill in a sandbox with the same model, with and without the skill, and publishing the Lift in percentage points), Equip (one MCP server plus per-agent loadouts, where users toggle skills and tools per agent and the agent only sees what was chosen), and Stay current (tracking upstream changes, re-scanning, re-benchmarking, and notifying users). Every skill gets a trust score from 0-100 built from security, quality, license, maintenance, and community signals — and critically, Lift is always shown separately from ratings, and money never touches scores. For document-heavy workflows, this means that when you equip a PDF or DOCX skill in Cursor, you'll know it's been security-scanned, license-checked, and benchmarked for actual performance improvement. Note that SkillzBench is pre-launch and actively being built: the waitlist and newsletter are live, and launch scope covers instruction-only skills (SKILL.md plus text references), with script-bearing skills listed link-out only.

Best for: Developers and teams who want verified, benchmarked, security-scanned document and PDF skills for Cursor and other SKILL.md-compatible agents

Pricing: Planned pricing includes a generous free tier, Pro (~$9-15/mo) for more agents and unlimited enabled skills, Teams (~$15-25/seat/mo) for org loadouts, allowlists and audit, and Creator (~$5-10/mo) for verified authors. Currently pre-launch — join the waitlist for details.

Key features

• Discover: continuously crawl and dedupe agent skills from GitHub
• Verify: pin every skill version to a commit SHA, validate, license-check, security-scan, and quality-judge
• Benchmark: test every skill in a sandbox with the same model, with and without the skill, publishing Lift in percentage points
• Equip: one MCP server plus per-agent loadouts — toggle skills and tools per agent so the agent only sees what was chosen
• Stay current: track upstream changes, re-scan, re-benchmark, and notify
• Trust score 0-100 from security, quality, license, maintenance, and community signals
• Lift always shown separately from ratings; money never touches scores

Pros

• Independent testing lab — trust scores and Lift are never influenced by money
• Security scanning directly addresses the ~26% vulnerability rate found in published research
• Per-agent loadouts mean your document-handling agent only sees the skills and tools you choose
• Commit-SHA pinning and change tracking keep skills current and auditable
• Sandboxed benchmarking shows measurable Lift, not just star ratings

Cons

• Pre-launch — the product is actively being built and not yet publicly available
• Launch scope covers instruction-only skills; script-bearing skills are link-out only
• Benchmark results are not yet published

Visit SkillzBench

#2 Anthropic Document Skills (DOCX, PDF, PPTX, XLSX)

Anthropic's official skills repository includes a document-skills collection that, according to the research, provides comprehensive DOCX, PDF, PPTX, and XLSX manipulation. This is the closest thing to a canonical set of document skills for the SKILL.md ecosystem — the same format Cursor reads from .cursor/skills/. Because SKILL.md is an open standard, these skills work in Cursor out of the box, and the chrisboden cursor-skills starter repo documents importing them directly from GitHub with a single command. The DOCX skill is listed in the AgenticSkills directory as enabling agents to create, edit, and analyze Word documents with tracked changes and formatting — a genuinely useful capability for teams generating reports, contracts, or specifications from within Cursor. As the originator of the Agent Skills standard, Anthropic's document skills are well-maintained reference implementations, and they're a natural starting point for any document workflow in Cursor before you layer on community-built alternatives.

Best for: Developers who want official, well-maintained document creation and editing skills that work in Cursor out of the box

Pricing: Free and open source

Key features

• Comprehensive DOCX, PDF, PPTX, and XLSX manipulation (per the repository's documented collection)
• DOCX skill for creating, editing, and analyzing Word documents with tracked changes and formatting
• Open SKILL.md format that works in Cursor without modification
• Importable directly from GitHub into a Cursor skills setup
• Maintained by Anthropic, the originator of the Agent Skills standard

Pros

• Official reference implementations from the creator of the skills standard
• Covers the full document spectrum: Word, PDF, PowerPoint, and Excel
• Free and open source, importable in one step
• Tracked-changes support in Word documents

Cons

• General-purpose by design — less specialized than niche community skills
• No built-in security scanning or benchmarking of the skills themselves

Visit Anthropic Document Skills (DOCX, PDF, PPTX, XLSX)

#3 chrisboden/cursor-skills (Skills MCP Server)

The chrisboden/cursor-skills repository is a starter template that transforms Cursor into a general-purpose AI agent through a role-based rules system and a Skills MCP server. According to the repository, the MCP server lets the agent discover available skills via list_skills, invoke specialized skills via invoke_skill, find community skills via find_skill, and import new skills from a GitHub repo URL via import_skill. For document work specifically, the repo's example structure shows a document-skills directory with separate docx, pdf, and xlsx skills, and the import tool can pull Anthropic's full document-skills collection into your Cursor setup automatically. The Orchestrator role is the key innovation: it delegates tasks to specialized skills via MCP tools rather than solving everything directly, meaning when you ask Cursor to generate or edit a PDF, it checks for a relevant skill first. The design is token-efficient — the agent initially only reads short metadata from each skill, loading the full SKILL.md only when a task requires it.

Best for: Developers who want a structured MCP-based framework in Cursor for discovering, invoking, and importing document skills

Pricing: Free and open source

Key features

• Skills MCP server with list_skills, invoke_skill, find_skill, and import_skill tools
• Orchestrator role that delegates to specialized skills, including document-skills for docx, pdf, and xlsx
• One-command import of skills from any GitHub repository, including Anthropic's document-skills collection
• Token-efficient progressive disclosure — full skill content loads only when needed
• Nested directory organization for skills, e.g., document-skills/pdf
• skill-creator skill for building new custom document skills

Pros

• Turns Cursor into a skill-delegating agent rather than a plain code editor
• Direct support for importing Anthropic's DOCX, PDF, and XLSX skills
• Automatic validation that each imported skill has a proper SKILL.md
• Free, with no separate dependency installation (uv handles everything)

Cons

• Requires Python 3.10+ and manual MCP configuration
• A starter template — you assemble your own skill collection rather than getting a curated one
• No security scanning or benchmarking of imported skills

Visit chrisboden/cursor-skills (Skills MCP Server)

#4 Agensi Cursor Skills Marketplace

Agensi is a skills marketplace that, according to its site, filters its catalogue down to skills Cursor loads without modification — every skill uses the open SKILL.md format. For document-adjacent work, Agensi's most relevant offerings include readme-generator (117 installs), which scans project structure, dependencies, and code to generate a complete README.md, and skill-polisher (13 installs), which transforms raw documentation into readable, well-structured markdown and works on any technical documentation. For teams that treat documents as part of the development workflow — onboarding docs, release notes, technical specs — these skills slot naturally into Cursor. Agensi's delivery model is notable: you connect Cursor through MCP and ask it to use the skill, and it loads the latest instructions on demand without saving skill files to disk. The marketplace reports install counts for its skills (code-reviewer leads with 447 installs), which gives a popularity signal, though Agensi notes local .cursor/skills/ folders are for skills you author yourself or obtain under a license permitting local files.

Best for: Teams that want a curated marketplace of Cursor-compatible documentation and workflow skills delivered via MCP

Pricing: The documented document and workflow skills (readme-generator, skill-polisher, changelog-generator) are listed as Free; paid skills require ownership

Key features

• Cursor skills marketplace filtered to SKILL.md skills that work in Cursor without modification
• readme-generator: scans project structure, dependencies, and code to generate a complete README.md
• skill-polisher: transforms raw documentation into readable, well-structured markdown for any technical docs
• changelog-generator: turns developer commit messages into user-facing release notes
• MCP-based delivery — agent loads the latest approved instructions on demand
• Install counts shown per skill for popularity signals

Pros

• Every listed skill is verified Cursor-compatible
• Strong documentation-focused skills for READMEs, changelogs, and technical docs
• MCP delivery keeps skills up to date without local file management
• Clear install counts and free entry-level skills

Cons

• Agensi notes you should never write purchased skills to disk, limiting offline use
• Smaller catalogue than open GitHub ecosystems
• No independent security benchmarking of listed skills

Visit Agensi Cursor Skills Marketplace

#5 Open Design (nexu-io)

Open Design, listed on OpenAgentSkill with 88,117 installs, is described as a local-first, open-source Claude Design alternative with a native desktop app. For document work specifically, the listing highlights HTML/PDF/PPTX/MP4 export — meaning you can produce actual PDF and PowerPoint files from prototypes and designs built with the agent. According to the research, it bundles 259+ skills and 142+ design systems, supports web, desktop, and mobile prototypes, slides, images, and videos, and works with Claude Code, Codex, Cursor, OpenCode, Qwen, Copilot, and 17+ CLIs. For developers and teams whose document needs lean toward visual deliverables — pitch decks, slide-based reports, PDF exports of designed artifacts — Open Design is one of the most capable multi-format options in the Cursor ecosystem. Its sandboxed preview and local-first architecture also mean your document content stays on your machine during the design process.

Best for: Teams producing visually rich PDFs, slide decks, and design-driven documents from within Cursor

Pricing: Contact for pricing

Key features

• HTML/PDF/PPTX/MP4 export for document deliverables
• 259+ skills and 142+ design systems bundled
• Local-first, open-source with a native desktop app
• Sandboxed preview for prototypes
• Web, desktop, and mobile prototypes, slides, images, and videos
• Works with Cursor plus 17+ other CLIs and agent platforms

Pros

• Direct PDF and PPTX export covers real document output formats
• Local-first architecture keeps content on your machine
• Large bundled skill and design-system library
• Broad agent compatibility beyond Cursor

Cons

• Design-centric rather than a pure document-manipulation tool
• Freshness signal on OpenAgentSkill shows 0/100 at time of research
• Pricing not published in the research data

Visit Open Design (nexu-io)

#6 Canvas Design (Anthropic)

Anthropic's Canvas Design skill, ranked #3 on OpenAgentSkill's best Cursor skills list with 181,205 installs, is described as enabling agents to create original visual art, posters, PNG assets, and PDF documents through a clear design philosophy. That last capability — PDF documents — makes it directly relevant to this list. According to the ranking data, it scores 100/100 on quality, 100/100 on popularity, and 100/100 on install readiness, with 88/100 freshness and strong evidence confidence. As an Anthropic-published skill in the open SKILL.md format, it loads in Cursor without modification, and its high install-readiness score means it's essentially copy-and-go. For developers who need to generate polished, design-forward PDFs — one-pagers, posters, visual reports — from a Cursor session, Canvas Design wraps a coherent design philosophy around the output rather than producing generic, template-looking documents. Its quality and popularity scores are among the highest of any design-oriented skill in the research data.

Best for: Developers generating design-forward PDFs, posters, and visual assets from Cursor sessions

Pricing: Free and open source

Key features

• Creates original visual art, posters, PNG assets, and PDF documents
• Guided by a clear design philosophy for non-generic output
• Open SKILL.md format — works in Cursor without modification
• 100/100 quality and install-readiness signals on OpenAgentSkill
• 181,205 installs per the ranking data

Pros

• Explicit PDF document creation capability
• Top-tier quality and popularity signals in the research data
• Backed by Anthropic's design philosophy for distinctive output
• Free and immediately installable

Cons

• Focused on visual/design output rather than document editing or analysis
• No built-in security verification of the skill itself

Visit Canvas Design (Anthropic)

#7 AgenticSkills Directory

AgenticSkills is a curated directory that, as of October 13, 2026, indexes 193+ curated agent skills across 16 categories and 8 platform targets, including Cursor, Claude Code, OpenAI Codex, and Gemini CLI. For document work, its featured listings include Anthropic's DOCX skill (create, edit, and analyze Word documents with tracked changes and formatting) and MCP Builder (a guide for creating MCP servers to integrate external APIs and services with AI agents — useful if your document workflow needs live connections to document storage systems). The directory's approach is notably principled: each listing links to the upstream repository, shows the real SKILL.md where one is published, and reports commit and contributor activity scoped to that skill's own path — what the site explicitly calls 'never a vanity install count.' That maintenance-focused signal is a meaningful step toward skill trust, and the directory also clearly explains the distinction between skills (knowledge) and MCP servers (access), which is essential context for anyone building a document pipeline in Cursor.

Best for: Developers who want a curated, maintenance-signal-driven directory to find document skills for Cursor

Pricing: Free to browse; individual skills are open source per their upstream licenses

Key features

• 193+ curated agent skills indexed across 16 categories and 8 platforms including Cursor
• Featured DOCX skill for creating, editing, and analyzing Word documents with tracked changes
• Each listing links to the upstream repository and shows the real SKILL.md
• Commit and contributor activity reported per skill's own path — not vanity install counts
• Install commands documented per skill
• Clear skills-vs-MCP guidance for building combined workflows

Pros

• Maintenance and contributor activity shown per skill rather than raw install counts
• Covers multiple agent platforms, not just Cursor
• Transparent methodology and links to upstream sources
• Includes Anthropic's DOCX document skill in featured listings

Cons

• A directory, not a testing lab — no sandboxed benchmarking or security scanning
• Smaller catalogue (193+) than the open GitHub ecosystem
• No per-agent loadout or equip functionality

Visit AgenticSkills Directory

#8 Awesome Cursor Skills (spencerpauly)

The Awesome Cursor Skills repository is a curated list of skills for Cursor, maintained on GitHub, containing ready-to-use SKILL.md files you copy into your .cursor/skills/ directory. While its catalogue leans toward engineering workflows rather than document manipulation specifically, it includes several skills that support document-adjacent work: building-skills-from-patterns (turn repeated multi-step workflows into a new SKILL.md so the agent can reuse them — ideal for codifying your team's document-generation process), saving-workspace-context (automatically persist research, decisions, and learnings to workspace files so knowledge survives across conversations), and suggesting-skills (the agent suggests installing a known skill when you're struggling with a task it covers). It also documents importing Anthropic's broader skills ecosystem. For teams assembling a document workflow in Cursor, Awesome Cursor Skills functions as a practical, community-maintained index — though, like any raw GitHub list, it carries no independent security verification, which is exactly the gap platforms like SkillzBench are being built to fill.

Best for: Developers who want a broad, community-maintained index of Cursor skills and the ability to codify their own document workflows

Pricing: Free and open source

Key features

• Curated list of ready-to-use SKILL.md files for Cursor's .cursor/skills/ directory
• building-skills-from-patterns: turn repeated document workflows into reusable skills
• saving-workspace-context: persist research and decisions to workspace files across conversations
• suggesting-skills: agent recommends installing a known skill when relevant
• Categorized catalogue spanning Cursor-native, testing, workflow, and infrastructure skills
• Documents importing skills from the broader ecosystem including Anthropic's repositories

Pros

• Wide, well-organized catalogue of copy-paste-ready skills
• building-skills-from-patterns is a genuine differentiator for codifying document workflows
• Free and community-maintained
• Skills are automatically discovered by Cursor's agent once installed

Cons

• Few skills directly targeting PDF or document manipulation
• No security scanning, license checking, or benchmarking of listed skills
• Quality varies since it's a raw community list

Visit Awesome Cursor Skills (spencerpauly)

Conclusion

Document and PDF work is one of the clearest, highest-value use cases for Cursor skills in 2026 — and the ecosystem has matured to the point where you can generate a polished PDF, edit a Word file with tracked changes, or codify an entire document workflow into a reusable SKILL.md. Anthropic's official document-skills collection and Canvas Design cover the fundamentals, chrisboden's cursor-skills MCP framework gives you a structured way to discover and invoke them, and marketplaces and directories like Agensi, OpenAgentSkill, and AgenticSkills help you find more. What all of these have in common is a missing trust layer: with published research finding roughly 26% of skills carry at least one vulnerability, installing an unvetted skill that touches your contracts, financials, or internal documents is a risk nobody should be taking blindly. That's the gap SkillzBench is being built to close — commit-SHA-pinned verification, security scanning, license checks, and sandboxed benchmarking that publishes real Lift in percentage points, plus an MCP server with per-agent loadouts so your document agent only ever sees the skills you've chosen. If you're serious about document workflows in Cursor, join the SkillzBench waitlist and be first in line when the trust layer for agent skills goes live. Skills, proven.

Frequently asked questions

What makes SkillzBench different from other skill directories?

SkillzBench isn't just a directory — it's an independent testing lab. Every skill version is pinned to a commit SHA, validated, license-checked, security-scanned, and quality-judged, then benchmarked in a sandbox with the same model, with and without the skill, to publish a measurable Lift in percentage points. Trust scores (0-100) come from security, quality, license, maintenance, and community signals, and Lift is always shown separately from ratings. Money never touches scores, so the rankings can't be bought.

Do skills built for Claude Code work in Cursor?

Yes. SKILL.md is an open standard, so skills built for Claude Code or Codex CLI generally work in Cursor without modification. Cursor reads skills from .cursor/skills/ in your project directory, and the agent automatically discovers them. The same skill, with minor path adjustments, can also run in GitHub Copilot and other agents that have adopted the format.

Is SkillzBench available now?

SkillzBench is pre-launch and actively being built. The waitlist and newsletter are live, but the product itself is not yet publicly available. Launch scope covers instruction-only skills (SKILL.md plus text references); script-bearing skills will be listed link-out only. Join the waitlist at skillzbench.com to get launch updates and early access details.

Why does skill security matter for document workflows?

Document workflows often involve sensitive material — contracts, financial statements, internal reports. Published research found approximately 26% of agent skills carry at least one vulnerability. Installing an unvetted skill that reads or generates your documents means an unreviewed instruction package is influencing how your agent handles that sensitive content. Security scanning, license checking, and commit-SHA pinning are what let you equip document skills with confidence.

How much do document skills for Cursor cost?

Most of the skills in this list are free and open source, including Anthropic's document-skills collection, Canvas Design, and the Awesome Cursor Skills catalogue. Agensi lists its documentation skills as free, with paid skills requiring ownership. SkillzBench is planning a generous free tier, with Pro (~$9-15/mo), Teams (~$15-25/seat/mo), and Creator (~$5-10/mo) tiers, plus benchmark-as-a-service for priority runs.

What is the difference between a skill and an MCP server?

A skill gives an agent knowledge — instructions for how to perform a task, packaged as a SKILL.md file. An MCP server gives an agent access — a live connection to a database, API, or file system. Skills are instructions; MCP servers are tools. Most real document workflows use both: a skill describes the procedure (how to format a PDF), while an MCP server supplies the live connection (where the document is stored).

How does SkillzBench's benchmarking work?

SkillzBench is building sandboxed benchmarking that tests every skill with the same model, with and without the skill, and publishes the resulting Lift in percentage points. This means you see whether a PDF or document skill actually improves agent performance on real tasks, rather than relying on install counts or star ratings. Lift is always displayed separately from trust scores, and money never influences either.