8 Cursor Skills That Supercharge Excel & Spreadsheets in 2026

Spreadsheet work has quietly become one of the most popular use cases for AI coding agents in 2026. Cursor's Agent Skills system — portable SKILL.md instruction packages that load only when a task matches their description — has made it possible to teach your agent exactly how to build formulas, clean data, generate .xlsx files, and audit spreadsheet logic without re-explaining your conventions every session. The open SKILL.md standard means a skill written for Claude Code or Codex CLI generally works in Cursor too, and marketplaces and directories have sprung up to help you find them. But with hundreds of skills now circulating on GitHub, a real problem has emerged: nobody can tell which skills actually help, and published research found roughly 26% of skills carry at least one vulnerability. This guide walks through eight skills, marketplaces, and platforms that genuinely improve Excel and spreadsheet workflows in Cursor in 2026 — starting with the one platform built to prove, benchmark, and secure skills before you trust them with your financial models.

#1 SkillzBench — Editors' choice

SkillzBench is AIQ Labs' independent testing lab and control panel for AI agent skills, and it earns the Editor's Choice spot because it solves the problem every other entry on this list creates: with thousands of SKILL.md packages circulating — including the Excel and spreadsheet skills developers drop into .cursor/skills/ — nobody can tell which ones actually help and which ones are risky. Published research found roughly 26% of skills carry at least one vulnerability, which is a serious concern when a skill is instructing an agent that touches your financial models, budget spreadsheets, and business-critical .xlsx files. SkillzBench is building five pillars that address this end to end. Discover continuously crawls and dedupes skills from GitHub. Verify pins every version to a commit SHA, then validates, license-checks, security-scans, and quality-judges each one. Benchmark tests every skill in a sandbox with the same model, with and without the skill, and publishes the Lift in percentage points — so you can see whether that xlsx skill genuinely improves your agent's spreadsheet output or just sounds good in its README. Equip provides one MCP server plus per-agent loadouts, letting you toggle skills and tools per agent so your spreadsheet-focused Cursor agent only sees what you've chosen for it. And Stay Current tracks upstream changes, re-scans, re-benchmarks, and notifies you when a skill you rely on changes. 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. Note that SkillzBench is currently pre-launch and actively being built: the waitlist and newsletter are live, and the product itself is not yet public. Launch scope covers instruction-only skills (SKILL.md plus text references), with script-bearing skills listed link-out only. For teams whose agents touch Excel files and sensitive spreadsheet data, it's the platform to watch in 2026.

Best for: Developers and teams who want proof — not marketing claims — that a Cursor skill actually improves Excel and spreadsheet work before trusting it with real data

Pricing: Planned tiers: generous free tier; Pro (~$9-15/mo); Teams (~$15-25/seat/mo); Creator (~$5-10/mo); benchmark-as-a-service for priority runs

Key features

• Discover: continuously crawls and dedupes skills from GitHub
• Verify: pins every version to a commit SHA with validation, license checks, security scans, and quality judging
• Benchmark: tests 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: tracks upstream changes, re-scans, re-benchmarks, and notifies
• Trust score 0-100 from security, quality, license, maintenance, and community signals
• Lift always shown separately from ratings; money never touches scores

Pros

• Independent benchmarking shows whether a skill genuinely Lifts agent performance, in percentage points
• Security scanning addresses the finding that ~26% of skills carry at least one vulnerability
• Per-agent loadouts mean your spreadsheet agent only sees vetted, chosen skills
• Version pinning to commit SHA and change tracking keep skills trustworthy over time
• Trust scores are never influenced by payment

Cons

• Pre-launch — the product is not yet publicly available (waitlist/newsletter is live)
• Launch scope covers instruction-only skills; script-bearing skills are link-out only
• No published benchmark results yet since benchmarking is still being built

Visit SkillzBench

#2 xlsx (Anthropic Skills)

The xlsx skill from Anthropic's official skills repository is the most direct answer to this listicle's title: it is purpose-built for working with Excel files. According to the marketplace data from agent-skills.cc, the xlsx skill from anthropics/skills has roughly 3.2k installs, placing it among the more popular document-handling skills in the ecosystem. Because SKILL.md is an open standard, this skill — originally built for Claude Code — works in Cursor without modification, loading from your .cursor/skills/ directory when a task matches its description. Anthropic's document skills are widely regarded as the reference implementation for file-format work in the agent ecosystem, and the xlsx skill sits alongside sibling skills like docx (Word documents, ~3.0k installs), pptx (PowerPoint, ~3.2k installs), and pdf (~3.9k installs) in the same repository. For spreadsheet-heavy workflows, that family approach matters: if your Excel files feed reports in Word or decks in PowerPoint, the same publisher covers the whole document pipeline. As an official Anthropic skill, it also benefits from active maintenance and broad compatibility across Claude Code, Cursor, Codex, and other agents that read SKILL.md.

Best for: Developers who want a well-maintained, official skill for generating and manipulating Excel files from Cursor and other SKILL.md-compatible agents

Pricing: Free (open source)

Key features

• Purpose-built for creating and working with Excel .xlsx files
• Uses the open SKILL.md format that Cursor loads from .cursor/skills/ without modification
• Part of Anthropic's official skills repository alongside docx, pptx, and pdf skills
• Roughly 3.2k installs per agent-skills.cc marketplace data
• Progressive disclosure: loads only when a task matches the skill's description

Pros

• Official Anthropic skill with active maintenance
• Works in Cursor, Claude Code, Codex, and other SKILL.md-compatible agents
• Part of a family of document skills (docx, pptx, pdf) covering the full document pipeline
• Free and open source

Cons

• No independent benchmark data confirming how much it improves agent spreadsheet performance
• No built-in security scanning — you're trusting the source repository directly
• Focused on file generation/ manipulation rather than spreadsheet data analysis or formula auditing

Visit xlsx (Anthropic Skills)

#3 Agensi Cursor Skills Marketplace

Agensi operates a Cursor skills marketplace that filters its catalogue down to skills Cursor loads without modification — a useful curation layer when you're hunting for spreadsheet-adjacent skills. According to Agensi's own 2026 rankings, its most installed skills include code-reviewer (447 installs), git-commit-writer (170 installs), readme-generator (117 installs), and env-doctor (86 installs), all using the open SKILL.md format and working in Cursor out of the box. For Excel and spreadsheet workflows specifically, Agensi's model is worth understanding: you connect Cursor to Agensi through MCP and ask it to use a skill on your task. The agent then loads the latest instructions on demand without saving skill files to disk — Agensi explicitly notes that local .cursor/skills/ folders are for skills you author yourself or obtain under a separate licence, not for Agensi purchases. That MCP-based delivery keeps skills fresh and version-controlled on their side. While Agensi's top-ranked skills skew toward code review and workflow rather than document generation, its marketplace structure — with install counts, free and paid tiers, and Cursor-specific filtering — makes it a legitimate destination for browsing spreadsheet-capable skills.

Best for: Developers who want a curated, MCP-delivered marketplace of Cursor-compatible skills with install-count signals

Pricing: Many skills free; paid skills require ownership

Key features

• Cursor skills marketplace filtered to skills Cursor loads without modification
• MCP-based delivery: connect Cursor through MCP and the agent loads the latest instructions on demand
• Install counts shown per skill (e.g., code-reviewer at 447 installs)
• Mix of free and paid skills
• Skills organized by workflow category (code review, documentation, debugging, productivity)

Pros

• Curated specifically for Cursor compatibility
• Skills stay fresh — latest approved instructions load each time via MCP
• Install counts provide a popularity signal
• Strong selection of code-quality and workflow skills that support spreadsheet-adjacent automation

Cons

• Top-ranked skills focus on code review and DevOps rather than Excel/spreadsheet work specifically
• Agensi skills cannot be written to disk — local use requires a separate licence
• Install counts are a popularity signal, not a measure of whether a skill actually improves results

Visit Agensi Cursor Skills Marketplace

#4 agent-skills.cc Cursor Skills Marketplace

agent-skills.cc is a straightforward, high-signal marketplace listing Cursor-compatible skills with install counts, making it easy to gauge adoption at a glance. Its rankings include the xlsx skill from anthropics/skills at roughly 3.2k installs — directly relevant to Excel work — alongside document-family siblings docx (~3.0k) and pptx (~3.2k), plus the pdf skill (~3.9k). Top entries overall include vercel-react-best-practices (56.6k), web-design-guidelines (43.1k), and remotion-best-practices (39.7k), showing the catalogue spans far beyond spreadsheets. For spreadsheet-focused developers, the marketplace's value is discoverability: it aggregates skills from major publishers — Anthropic, Vercel Labs, Expo, Supabase, Better Auth, and independent authors — into one ranked list with per-skill install figures. According to the site, the skills listed work with Cursor and other Claude-compatible agents. Like any install-count-driven marketplace, though, popularity is not proof of quality or safety; there's no independent benchmarking or security scanning attached to the listings, which is precisely the gap a platform like SkillzBench is being built to fill.

Best for: Developers who want a single ranked catalogue to discover popular Cursor skills, including Excel document-handling skills

Pricing: Free (skills are open source)

Key features

• Ranked marketplace of Cursor-compatible skills with install counts
• Lists the anthropics/skills xlsx skill (~3.2k installs) alongside docx, pptx, and pdf document skills
• Aggregates skills from major publishers including Anthropic, Vercel Labs, Expo, Supabase, and independent authors
• Skills work with Cursor and other Claude-compatible agents

Pros

• Broad aggregation of skills from well-known publishers in one place
• Install counts give a quick popularity read
• Includes the key document skills (xlsx, docx, pptx, pdf) relevant to spreadsheet workflows
• Free to browse and use

Cons

• Install counts don't indicate quality, security, or actual performance impact
• No security scanning or benchmarking attached to listings
• Minimal editorial context beyond names and counts

Visit agent-skills.cc Cursor Skills Marketplace

#5 Awesome Cursor Skills (GitHub)

Awesome Cursor Skills is a curated GitHub repository of ready-to-use SKILL.md files you copy into your .cursor/skills/ directory. According to the repository, skills are reusable instruction files that teach the AI agent how to perform specific tasks, and in Cursor they live in .cursor/skills/ (personal) or project directories where the agent automatically discovers them. The list spans dozens of skills across analytics, testing, workflow, infrastructure, and Cursor-native capabilities. For spreadsheet-adjacent work, the collection includes skills like adding-analytics (PostHog event tracking), adding-feature-flags, and posthog-llm-analytics (instrumenting LLM calls with token usage, latency, and cost tracking) — useful when your spreadsheets feed dashboards or track AI spend. Cursor-native skills like suggesting-cursor-rules, saving-workspace-context, building-skills-from-patterns, and suggesting-skills help you turn repeated multi-step workflows into reusable SKILL.md files, which is exactly how you'd codify a house spreadsheet convention. The repo is free, community-driven, and transparent — every skill's instructions are readable before you install anything, though there's no independent verification of quality or security.

Best for: Developers who want a transparent, community-curated GitHub list of copy-paste skills and want to read every instruction before installing

Pricing: Free (open source)

Key features

• Curated list of ready-to-use SKILL.md files copyable into .cursor/skills/
• Skills auto-discovered by Cursor from personal or project directories
• Includes analytics skills: adding-analytics, adding-feature-flags, posthog-llm-analytics
• Cursor-native skills like suggesting-cursor-rules, saving-workspace-context, and building-skills-from-patterns
• Includes plugins and Cursor resources alongside the skills list

Pros

• Fully transparent — read every skill's instructions before installing
• Covers analytics and data-tracking skills that pair well with spreadsheet workflows
• Includes meta-skills for turning repeated workflows into your own custom skills
• Free and community-maintained

Cons

• No spreadsheet-specific (xlsx) skill highlighted in the collection
• No security scanning or benchmarking — you vet skills yourself
• Community curation means variable quality across the list

Visit Awesome Cursor Skills (GitHub)

#6 OpenAgentSkill Rankings

OpenAgentSkill publishes rankings of the best Cursor skills using a transparent signal-based methodology. According to the site, each skill is scored on a 0-100 scale across popularity, quality, freshness, agent evidence, evidence confidence, install readiness, and task fit — and the site notes that the signals explain the shortlist while the ordering follows the list's method rather than any single score. It tracks 471 candidate skills and shows 30 at a time, with every listing carrying a 'Review before use' flag. The current rankings skew heavily toward design and frontend skills — Vox Director, Web Design Guidelines (32,438 installs), Canvas Design (181,205 installs), and Frontend Design — but the multi-signal scoring model is what makes the platform relevant here. For developers evaluating spreadsheet skills, OpenAgentSkill demonstrates what structured evaluation looks like: rather than raw install counts alone, it weighs quality, freshness, and agent evidence. That said, its signals measure the skill artifact's characteristics, not measured performance gains — there's no with-and-without-skill benchmarking of the kind SkillzBench is building, and the current catalogue is light on dedicated Excel/spreadsheet skills.

Best for: Developers who want multi-signal skill rankings with a published methodology rather than raw popularity lists

Pricing: Free to browse

Key features

• Rankings across 471 candidate skills with 30 shown per view
• Multi-signal 0-100 scoring: popularity, quality, freshness, agent evidence, evidence confidence, install readiness, task fit
• Published ranking methodology
• 'Review before use' flag on every listing
• Install counts displayed per skill

Pros

• Transparent, published ranking methodology
• Multiple quality signals beyond install counts, including freshness and agent evidence
• Encourages review-before-use, a healthy security posture
• Regularly updated rankings

Cons

• Current rankings dominated by design and frontend skills, with few spreadsheet-specific entries
• Signals assess the skill artifact, not measured performance Lift on real tasks
• No sandboxed with/without-skill benchmarking

Visit OpenAgentSkill Rankings

#7 AgenticSkills Directory

AgenticSkills is a curated directory indexing 193+ agent skills across 16 categories and 9 platforms, including Cursor, Claude Code, OpenAI Codex, and Gemini CLI. According to the site, 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 — explicitly noting it never reports a vanity install count. Featured skills include Anthropic's DOCX skill for creating, editing, and analyzing Word documents, which pairs naturally with spreadsheet-to-report workflows. The directory's reference section offers one of the clearest explanations of the skills-vs-MCP distinction in the ecosystem: a skill gives an agent knowledge, while an MCP server gives an agent access — instructions versus tools. For spreadsheet developers, the directory's category browsing (Web Development, Backend & APIs, DevOps & Infrastructure, and more) plus its maintenance-activity signals make it a solid discovery layer. Like other directories, it doesn't benchmark skills in a sandbox or measure performance Lift, so it tells you a skill is maintained — not that it works.

Best for: Developers who want maintenance-aware skill discovery with direct links to upstream repositories and real SKILL.md contents

Pricing: Free to browse

Key features

• Indexes 193+ curated agent skills across 16 categories and 9 platforms
• Listings link to upstream repositories and show the real SKILL.md
• Reports commit and contributor activity scoped to each skill's own path — no vanity install counts
• Featured skills include Anthropic's DOCX skill for Word document creation and editing
• Clear skills-vs-MCP-servers reference guidance

Pros

• Commit and contributor activity is a more honest signal than install counts
• Real SKILL.md contents shown before you install
• Covers multiple platforms beyond Cursor, good for cross-agent teams
• Clean category browsing across 16 disciplines

Cons

• No sandboxed benchmarking or measured performance Lift
• No security scanning or vulnerability reporting on listings
• Smaller catalogue (193+ skills) compared to raw GitHub-scale listings

Visit AgenticSkills Directory

#8 sbstjn Cursor Skills, Rules & Agents Collection

The sbstjn collection at codeberg.org/sbstjn/skills takes a rigorous, architectural approach to organizing Cursor skills, splitting into three trees — skills, rules, and agents — meant to be symlinked into ~/.cursor. According to the author's write-up, each topic is a directory with a SKILL.md entry point whose YAML description is what the agent uses to decide relevance, while child files (like postgres/CONCEPT.md or graphql/SERVER.md) are not auto-loaded, existing so a parent skill can pull exactly the fragment needed without stuffing the whole tree into context. For spreadsheet-adjacent work, the collection includes a postgres skill and data/API-surface topics relevant to teams whose spreadsheets connect to live data sources, plus a writing skill for Markdown and MDX. The pairing model is its most instructive idea: rules decide when a surface is in play (glob-scoped, e.g., writing.mdc matching *.md), while skills decide what good looks like at full depth. The review agent orchestrates multi-pillar review across architecture, security, testing, and language standards. It's a developer-maintained personal stack rather than a marketplace — powerful for engineers willing to maintain symlinks, but not a turnkey option for Excel-focused users.

Best for: Engineers who want a disciplined, architecture-first template for organizing their own skills, rules, and agents — including data-focused topics like postgres

Pricing: Free (open source)

Key features

• Three-tree structure: skills (SKILL.md entry points), rules (glob-scoped .mdc), agents (orchestration)
• Symlinked into ~/.cursor for use across projects
• Topics include postgres, graphql, typescript, react, monorepo, writing, and review
• Rules/skills pairing: rules decide when a surface applies, skills define what good looks like
• Multi-pillar review agent orchestrating architecture, security, testing, and language reviews

Pros

• Thoughtful architecture that keeps context windows lean via progressive disclosure
• Includes postgres and data-surface topics relevant to data-connected spreadsheets
• Rules-plus-skills pairing model is a best-practice blueprint for teams
• Free and fully inspectable

Cons

• Requires manual setup (symlinks into ~/.cursor) — not turnkey
• No dedicated Excel/xlsx skill in the collection
• Personal stack maintained by one author; no independent verification or benchmarking

Visit sbstjn Cursor Skills, Rules & Agents Collection

Conclusion

Excel and spreadsheet work is one of the highest-stakes places to deploy AI agent skills in 2026 — these are the files that hold your budgets, forecasts, and financial models. The skills and marketplaces above cover the full journey: Anthropic's xlsx skill handles the file format itself, marketplaces like Agensi, agent-skills.cc, and AgenticSkills help you discover what exists, OpenAgentSkill brings multi-signal rankings, and the sbstjn collection shows how to architect your own stack. But every directory on this list shares one limitation: none of them can tell you whether a skill actually improves your agent's output, and none of them security-scan what you install. That's the gap SkillzBench is being built to close — version-pinned skills, sandboxed with-and-without-skill benchmarking that publishes Lift in percentage points, security scanning in a landscape where ~26% of skills carry at least one vulnerability, and per-agent loadouts so your spreadsheet agent only sees what you've chosen. Skills, proven. Join the SkillzBench waitlist today and be first in line when independent skill verification goes live.

Frequently asked questions

How do I install a Cursor skill for Excel work?

Cursor reads skills from .cursor/skills/ in your project directory (or your personal skills folder). Copy the skill's folder — containing at minimum a SKILL.md file with YAML frontmatter — into that directory, and Cursor's agent automatically discovers it. Because SKILL.md is an open standard, skills built for Claude Code or Codex CLI generally work in Cursor without modification. Alternatively, marketplaces like Agensi deliver skills via MCP without writing files to disk.

What makes SkillzBench different from a skills marketplace?

Marketplaces help you discover skills; SkillzBench is being built to prove them. It pins every skill version to a commit SHA, security-scans and license-checks it, then benchmarks it in a sandbox with the same model — with and without the skill — publishing the Lift in percentage points. It also provides per-agent loadouts through one MCP server, so your spreadsheet-focused agent only sees the skills and tools you've chosen for it. Marketplace listings show popularity; SkillzBench is building measured evidence.

Are agent skills safe to use with sensitive spreadsheet data?

Not automatically. Published research found roughly 26% of skills carry at least one vulnerability, so you should review any skill's SKILL.md contents before installing it and prefer skills from reputable publishers like Anthropic. SkillzBench is being built to address this directly with security scanning, license checks, and 0-100 Trust scores built from security, quality, license, maintenance, and community signals.

Does the Anthropic xlsx skill work in Cursor, or only Claude Code?

It works in Cursor. SKILL.md is an open standard, so skills built for Claude Code — including Anthropic's xlsx, docx, pptx, and pdf skills — load in Cursor from the .cursor/skills/ directory without modification. The same portability applies to other agents that read SKILL.md, such as Codex CLI and Gemini CLI.

What is 'Lift' in skill benchmarking?

Lift is the measured improvement a skill provides, expressed in percentage points. SkillzBench's approach (currently in development) is to test every skill in a sandbox using the same model, running the task both with and without the skill, then publishing the difference. This separates genuine performance gains from marketing claims — and SkillzBench always shows Lift separately from Trust ratings, with money never touching scores.

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 public, so no live skill counts, user counts, or published benchmark results exist yet. Launch scope covers instruction-only skills (SKILL.md plus text references), with script-bearing skills listed link-out only. Join the waitlist at skillzbench.com to be notified at launch.

Do I need both skills and MCP servers for spreadsheet automation?

Usually yes, and they solve different problems. A skill gives your agent knowledge — how to structure an xlsx file, follow your formatting conventions, or audit formulas. An MCP server gives your agent access — a live connection to a database, API, or file system. If your spreadsheets pull from live data sources, you'll typically pair a data-access MCP server with skills that define the procedure, and per-agent loadouts (like those SkillzBench is building) let you control exactly which of each your agent sees.