Best Cursor Commands in 2026: 12 Copy-Paste Slash Commands for Solo Founders

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

Short answer: a Cursor command is a plain markdown file in .cursor/commands/, and the filename becomes the slash command you type in Agent (code-review.md becomes /code-review). Below are 12 copy-paste command files for solo founders, each with the full prompt, plus where commands can live (project vs global), when to convert one into a skill, and what to check when a command does not show up.

Cursor's own docs show only a few examples, and I have not verified every detail of how newer builds treat commands, so where behavior varies I say "check" rather than assert. The prompts below are ones I wrote and use as starting points. Edit them to your stack. For a personal walk-through of the full 35-command kit, see the field guide; this post is the ranked, copy-paste version.

What a Cursor command is

A command is a saved prompt. No frontmatter is required: the file body is the prompt, and whatever you type after the command name is passed along as extra context.

.cursor/commands/
  code-review.md      ->  /code-review
  pr-description.md   ->  /pr-description

Create the folder, drop in a .md file, type / in Agent chat. Use lowercase, hyphenated filenames so the command name is predictable. Commands run only when you invoke them, which is the main difference from rules (always on) and skills (loaded when relevant).

Want one built from a plain description instead of copy-pasting? The Cursor Commands Builder turns a task description and category into a ready .cursor/commands/<name>.md file.

Project vs global vs team commands

LocationScopeUse it for
.cursor/commands/ in the repoThis project, shared via gitCommands that reference your stack, paths and conventions
~/.cursor/commands/Every project on your machinePersonal habits: /standup, /explain-diff
Team commandsShared across a Cursor teamOrg-wide prompts managed centrally (a Cursor team feature; check the docs for your plan)

Rule of thumb: if the prompt mentions your folder structure or ORM, it is a project command. If it would work in any repo, make it global. When the same name exists in two places, do not rely on which one wins; give them distinct names.

The 12 commands

Ranked by how often a solo founder reaches for them. Each is a file you can paste as-is. Tell the agent what to look at in the text after the command, for example /code-review the diff on this branch.

1. /code-review

Review the current changes (staged, unstaged, or the branch diff if I name one).
Report only real problems, ordered by severity:
1. Correctness bugs and unhandled edge cases
2. Security issues (auth checks, input validation, secrets, tenant isolation)
3. Missing or weak tests for changed behavior
For each finding give: file:line, what breaks, and a concrete fix.
Do not comment on style or formatting. If there are no real problems, say so plainly.

2. /pr-description

Write a pull request description for the current branch.
Read the diff and commit messages first. Output:
- One-sentence summary of what changed and why
- Bullet list of notable changes grouped by area
- How to test it (exact steps)
- Risks and anything reviewers should look at closely
Do not invent changes that are not in the diff.

3. /db-migration

I will describe a schema change. Write the migration for our existing migration tool (detect it from the repo).
Requirements:
- Safe on a table with existing rows (nullable or defaulted columns first, backfill, then constrain)
- Reversible: include the down migration
- Add indexes for new foreign keys and any column I will filter on
- List any lock or downtime risk
Show the SQL or migration file, then list the app code that must change. Do not run it.

4. /write-tests

Write tests for the code I point at.
First look at the existing tests and copy their framework, naming and mocking style.
Cover: the happy path, each error branch, and boundary inputs.
Do not change production code. If something is untestable without a refactor, say what and stop.
Run the tests and report the result.

5. /security-pass

Do a security pass over the changed files (or the path I name).
Check: missing authorization on routes and server actions, user input reaching queries or shell,
secrets or keys in code or client bundles, unsafe redirects, missing rate limits on auth endpoints,
and data returned to the client that should not be.
Report findings with file:line and a fix. Separate confirmed issues from "worth checking".

For a deeper dive on the keys problem specifically, see finding exposed API keys in vibe-coded apps.

6. /release-notes

Write customer-facing release notes from the commits and merged PRs since the last tag (or the range I give).
Group into: New, Improved, Fixed. Write for users, not developers: say what they can now do.
Skip refactors and internal changes. One line per item, no marketing adjectives.

7. /fix-types

Run the type checker and fix every error.
Fix root causes. Do not use `any`, `@ts-ignore`, or non-null assertions to silence errors
unless you explain why it is correct. Re-run until clean and summarize what you changed.

8. /explain-diff

Explain the current diff to me as if I did not write it.
Start with the intent in two sentences. Then walk through each file: what changed and why it matters.
Flag anything surprising or any change unrelated to the stated intent.

9. /refactor-small

Refactor only the function or file I point at.
Keep behavior identical. No new dependencies, no API changes, no edits outside the target.
Prefer smaller functions, clear names, and removing dead code.
Run the existing tests before and after. If there are none, tell me before you start.

10. /seo-audit

Audit the page or route I name for on-page SEO.
Check: one H1, title and meta description present and unique, canonical URL, heading order,
image alt text, internal links in and out, structured data, and whether the first paragraph answers the target query.
Output a prioritized fix list with the exact change for each. Do not rewrite the whole page.

AgenticKit ships a fuller /seo-audit command backed by an SEO skill and agent; the version above is the single-file cut.

11. /changelog

Update CHANGELOG.md with the changes since the last entry.
Follow the format already in the file. Add a new section at the top with today's date.
Use the commit history as the source and do not invent entries.

12. /standup

Summarize what I did since my last standup from git activity (commits, branches, merged PRs) over the last working day.
Format: Done, In progress, Blocked. Keep it under 8 lines. If something looks unfinished, list it under In progress.

Commands vs skills

Commands are manual prompts. Skills carry the same kind of instructions, but the agent can load them on its own when the description matches, and they can bundle supporting files. Convert a command into a skill when you keep forgetting to run it, or when it needs reference material (a style guide, a checklist file) next to it. Keep it a command when you want it to fire only on demand, like /release-notes. The full breakdown is in Cursor skills vs commands, and the skills side is covered in Best Cursor Skills in 2026. For a ranked shortlist of which skills to install first, see 25 Best Cursor Skills for Solo SaaS Founders in 2026 (Ranked). If you also want hooks in the picture, see Rules vs Skills vs Commands vs Hooks: The Only Diagram You Need.

Troubleshooting: commands not showing up

Work down this list in order.

  1. Wrong folder or extension. It must be .cursor/commands/<name>.md, at the root of the folder you opened, not a subfolder or .cursor/command/.
  2. Wrong working directory. If you opened a parent folder or a different worktree, Cursor looks for .cursor/ there. Open the repo root, or confirm which worktree the agent is running in.
  3. Not reloaded. Reload the window after adding files if the / menu looks stale.
  4. Global vs project name clash. Rename one of them.
  5. CLI or Cloud agents. Cursor's CLI and cloud/background agents run in a different environment from your editor window. Do not assume ~/.cursor/commands on your laptop exists there. Commit commands to the repo (.cursor/commands/) so every environment checks them out, and consult the current Cursor docs and forum for what each surface supports, since this has been changing. A forum thread on slash commands in the CLI is a good place to check current behavior.
  6. Empty or unsaved file. An empty .md does not make a useful command.

If a command exists but the agent ignores half of it, shorten it. One job per file works better than a checklist of ten.

FAQ

Where do Cursor commands go? In .cursor/commands/ in your project, or ~/.cursor/commands/ for commands you want in every project. The filename becomes the command name.

Do commands need frontmatter? No. The file body is the prompt.

Can I pass arguments? Anything you type after the command name is added as context, so /write-tests src/billing/invoice.ts works as an instruction, not a formal parameter.

What is the difference between a command and a rule? Rules apply automatically (always, or by file pattern). Commands run only when you type them.

Why does a command work in the editor but not in the CLI or cloud agent? Those environments may not have your global folder. Commit the command to the repo and check which surface you are using.

Want these commands paired with the rule that should back them (so /security-pass isn't your only line of defense)? See 25 Cursor Slash Commands and Rules Combos.

Get all of them in one command

AgenticKit installs 47 ready-made commands (including /seo-audit, /audit, /ship and /fix) with their agents and skills into your repo. Browse the commands catalog or the best Cursor commands comparison, or build a single one with the Slash Prompt Builder or the Cursor Commands Builder — see 13 Best Free Cursor Tools in 2026 for how it compares to other free generators.

npx agentickit-cursor init --target cursor

Sources

Written by Rakshit Yadav. Published 2026-10-25, last updated 2026-10-25.

Related Tools & Agents

🛠️ Free Tool: slash-prompt🛠️ Free Tool: rules-grader🤖 Agent: code-reviewer🤖 Agent: security-auditor🤖 Agent: seo-specialist⚡ Command: /audit⚡ Command: /seo-audit⚡ Command: /shipSkill: project-conventionsSkill: security-basicsSkill: on-page-seo

Get the whole command set

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

Browse commands →