Best Cursor Slash Commands in 2026: 20 Commands Every Solo Founder Should Steal

By Rakshit Yadav (@yadavrakshit60)•Oct 2026•9 min read

Short answer: a Cursor slash command is a markdown file in .cursor/commands/ — type / in Agent chat and the filename shows up as the command name. Below are 20 slash commands worth having, grouped by job, each with the exact prompt you can paste into a new file, plus the 3 most founders forget to install and how to write your own in five minutes.

This is the command-only version of a bigger question. If you want the difference between commands, rules, and skills explained first, read that section below before copying anything. Running Claude Code instead (or alongside)? The same 12 catalog commands, with Claude Code's own frontmatter options, are in Best Claude Code Commands for Shipping a SaaS MVP in 2026.

Slash commands vs rules vs skills, in one paragraph

A rule (.cursor/rules/*.mdc) applies automatically, always or by file glob — you never type it. A slash command (.cursor/commands/*.md) is a saved prompt that only runs when you type /name. A skill (SKILL.md) sits in between: the agent loads it on its own when the description matches, and it can bundle reference files a command can't. If you keep forgetting to type a command, that's a sign it should be a skill instead. For the longer version with 10 copy-paste rule files, see Best Cursor Rules for Next.js and SaaS in 2026; for the skills side, see Best Cursor Skills in 2026.

How custom commands actually work

.cursor/commands/
  seo-audit.md     ->  /seo-audit
  ship.md          ->  /ship
  release.md       ->  /release

No frontmatter is required for a basic command — the file body is the prompt, and anything you type after the command name (/seo-audit /pricing) gets passed along as extra context. Two scopes matter:

LocationScopeUse it for
.cursor/commands/ in the repoThis project, committed to gitCommands that reference your stack, folders, or conventions
~/.cursor/commands/Every project on your machinePersonal habits that don't depend on the repo

Cursor's CLI and cloud/background agents may not see your global (~/.cursor/commands) folder — commit anything you rely on to the repo. The mechanics (frontmatter, folders, project vs global) are covered in more depth in the 12 copy-paste commands post; this one is the broader 20-item list targeted at picking a starting set, not a single file's anatomy.

20 slash commands worth having

Ranked loosely by how often a solo founder reaches for them. 12 of these ship in the AgenticKit commands catalog with a full agent pipeline behind them; the rest are plain single-file prompts you can write yourself in the next five minutes. Either way, start with the ones you'd actually type this week.

Build

/ship — plan through a working, tested slice for one S/M complexity feature. Full pipeline: /ship.

/feature-spec — turn a vague ask into acceptance criteria before you touch code. /feature-spec.

/build-api — a route with a clear contract. /build-api.

/build-page — a view against an API that already exists. /build-page.

/db-migration — a migration safe on a table with existing rows (backfill-then-constrain), with the down migration included.

/fix-types — run the type checker and fix every error at the root cause, no @ts-ignore escape hatches.

Verify

/code-review — review the diff for correctness, security, and missing tests, file:line, no style nitpicks.

/audit — a pre-release sweep of a folder of routes, not a daily linter. /audit.

/security-pass — check auth on routes, input reaching queries, secrets in client bundles, missing rate limits.

/e2e-suite — end-to-end coverage for a path that touches money or auth, not every page. /e2e-suite.

/write-tests — tests that copy your existing framework and mocking style, happy path plus error branches.

Operate

/pr-description — a PR description written from the actual diff and commit messages, nothing invented.

/release — a changelog generated from git since the last tag. /release.

/billing-fix — Stripe dunning, proration, seat math, or a broken customer portal fix, without changing live prices unannounced. /billing-fix.

/sre-incident — triage for something actually down, not a typo on the pricing page.

/ci-pipeline — CI wired to the scripts already in your CURSOR.md, not a second test runner.

Grow

/whats-new — turn a real shipped diff into changelog, email, and social copy, claims-checked against the diff. /whats-new.

/seo-audit — an on-page SEO report for a route: title, canonical, headings, schema, internal links. /seo-audit.

/blog-post — a post from a real reader problem and real sources, not a keyword-stuffed filler page. /blog-post.

/standup — Done / In progress / Blocked, summarized from git activity since your last standup, under 8 lines.

3 commands most solo founders are missing

These three don't make most "top commands" lists because they only pay off once you've shipped something real — which is exactly the point.

  1. /billing-fix — most solo founders wire up Stripe once with /stripe-billing and never touch it again until a payment_failed webhook silently drops a customer. /billing-fix exists specifically for the follow-up bugs: dunning emails that never fire, proration that double-charges a downgrade, seat counts that drift from the Stripe subscription object.
  2. /release — if you tag releases by hand (or don't tag at all), you have no changelog a user or a future-you can read. /release reads git since the last tag and writes one. Pair it with /whats-new so the same ship also produces customer-facing copy instead of staying invisible.
  3. /whats-new — the gap between "we shipped it" and "customers know we shipped it" is where retention quietly leaks. This command turns a diff into a why-care blurb for email, changelog, and social, and is built to refuse inventing features that aren't actually in the diff.

How to write your own command in 5 minutes

  1. Create .cursor/commands/your-name.md in the repo (not .cursor/command/, not a subfolder — Cursor only scans the top level of that folder).
  2. Write the prompt as the entire file body. No frontmatter needed for a basic command.
  3. Keep it to one job. A command that reads like a ten-item checklist gets partially ignored; split it into two commands instead.
  4. State what NOT to do as explicitly as what to do — "do not invent entries," "do not change production code" — since that's usually the failure mode you're trying to prevent.
  5. Reload the Cursor window if / doesn't show the new file, and confirm you're in the repo root the command folder lives in.

A minimal example:

Summarize the current branch's commits into a one-paragraph PR description.
Do not invent changes that aren't in the diff. List risk areas a reviewer should check.

Save that as .cursor/commands/pr-description.md and /pr-description is live.

Mistakes that make a command list useless

  • Writing 20 commands before you've typed any of them twice. A command earns its place by repetition, not by looking complete in a folder listing. Start with the 5-8 above you'd actually reach for this week and add the rest only when you notice yourself retyping a prompt.
  • One command, ten jobs. /ship that also does security review, writes the changelog, and drafts a tweet will do all four badly. Split it — /ship, /security-pass, /release, /whats-new — so each one has a single job to get right.
  • Grow commands in an engineering-only install. /whats-new and /blog-post assume a filled-in brand voice and positioning doc exist somewhere. If they don't, the command either stalls or invents a tone nobody approved.
  • No "do not" line. "Write tests" invites an agent to also refactor the function under test. "Write tests, do not change production code" closes that door. Every command above has at least one explicit don't for this reason.
  • Relying on ~/.cursor/commands/ for something a CI run or cloud agent needs. That folder lives on your laptop. If a command matters outside your editor, it has to be in the repo.

FAQ

What is a Cursor slash command? A markdown file in .cursor/commands/ (or ~/.cursor/commands/ for global) whose filename becomes the /name you type in Agent chat. The file body is the prompt.

How many slash commands should I have? Start with 5-8 you'd actually type weekly: /ship, /code-review, /pr-description, /fix-types, /release. Add more as real friction shows up — don't pre-build 20 you'll never run.

Do Cursor slash commands need frontmatter? No for a basic command. Frontmatter becomes relevant for more structured setups (argument hints, allowed tools) like the ones in the commands catalog.

What's the difference between a slash command and a Cursor rule? A rule applies automatically (always, or by file pattern). A command only runs when you type it. Rules are for standing constraints; commands are for jobs you do on demand.

Why doesn't my command show up in the CLI or a cloud agent? Those surfaces may not read your global ~/.cursor/commands/ folder. Commit commands you depend on to .cursor/commands/ in the repo so every environment checks them out.

Get all of them in one install

AgenticKit installs 48 commands — including every one named above — with the agents and skills behind them, into .cursor/ and .claude/. Browse the full commands catalog, or start from a blank file with the Slash Prompt Builder.

npx agentickit-cursor init --target cursor

Sources

Written by Rakshit Yadav. Published 2026-10-26, last updated 2026-10-26.

Related Tools & Agents

🛠️ Free Tool: slash-prompt🛠️ Free Tool: rules-grader🤖 Agent: billing-engineer🤖 Agent: product-marketer🤖 Agent: code-reviewer⚡ Command: /billing-fix⚡ Command: /release⚡ Command: /whats-newSkill: project-conventionsSkill: billing-dunningSkill: feature-narrative

Get the whole command set

48 commands land in .cursor/ and .claude/ with one install.

Browse commands →