Install pstack skills
Not yet tested by AgentSkillsDir. These are the upstream commands as written in the READMEs fetched on 1 Nov 2026; we have not run the installs, setup or /poteto-mode ourselves.
CursorOfficial
Cursor plugin
Run in Cursor chat (official README, “install”)
/add-plugin pstackThe official path. Pick /poteto-mode from the / menu and press option+enter (alt+enter on Windows) to keep it on across turns as a custom mode.
Claude CodeOfficial
Claude Code marketplace
Run inside Claude Code (official README, “install”)
/plugin marketplace add cursor/plugins
/plugin install pstack@cursor-pluginsClaude Code prefixes plugin commands: /poteto-mode becomes /pstack:poteto-mode and setup becomes /pstack:setup-pstack. Auto-update is off for third-party marketplaces; enable it under /plugin → marketplaces → cursor-plugins, or update by hand with claude plugin update pstack@cursor-plugins.
Codex · Pi · OpenCode · othersCommunity mirror
Standalone mirror (skills CLI)
Run in your terminal (mirror README, “Install”)
npx skills add backnotprop/pstackbacknotprop/pstack is a community mirror kept in sync with cursor/plugins for Claude Code, Codex, Pi, Cursor, OpenCode and other agents. The CLI lets you pick skills and agents. The mirror rewrites Cursor-specific skills to work in any harness, so it is not byte-identical to upstream. Its sponsor badges (Plannotator) are mirror sponsorship, not a required install.
First run: setup, then /poteto-mode
Two steps from the official README. In Claude Code, use the prefixed name on the right.
/setup-pstackClaude Code:/pstack:setup-pstackPick a reasoning budget and which models play each role. In Claude Code it writes the choices to ~/.claude/rules/pstack-models.md.
/poteto-modeClaude Code:/pstack:poteto-modeStart any task that needs rigor with it. It matches the task to a playbook, opens a todo list from that playbook’s steps and calls the other skills as the steps need them.
/poteto-helpClaude Code:/pstack:poteto-helpStuck, or unsure which skill fits? It asks what you are trying to do and hands back a prompt to type.
What it is, and what it is not
- It is a pack of plain Agent Skills (skills/<name>/SKILL.md) plus /poteto-mode, a router that matches your task to a playbook and runs the other skills when a step needs them.
- It is not the Linux pstack utility, not gstack, and not a hosted service. AgentSkillsDir does not host poteto-mode sessions or run multi-model panels for you.
- The README says there are no planning skills on purpose; Cursor’s plan mode or /poteto-mode covers planning when you want it.
Skill map
/poteto-mode is the entry point; it calls these when a step needs them. You can also type any of them directly. A sample, not the full list:
| Command | What it does |
|---|---|
/arena | N parallel attempts at the same thing, then keep the best parts of each. |
/interrogate | Different models try to break a diff; only findings a run proves are kept. |
/tdd | Write the failing test first, then the fix, when a cheap local test path exists. |
/unslop | Cleans up writing by removing AI tells (prose, not UI). |
/reflect | After a long task lands, captures the recipe as a skill edit. |
/how | A walkthrough of how a subsystem works. |
/swarm | N parallel workers across slices, then one aggregated report. |
See every skill in the official README
Upstream claims, not reproduced
The README states the following. We list them so you know what to expect; we have not reproduced any of them.
- /poteto-mode ships twenty-three playbooks (bug fix, perf, feature, refactoring, shipping, autonomous run and more).
- Twenty-four one-rule principle skills that /poteto-mode indexes at task start.
- Out of the box, code work goes to grok and the hardest changes, prose and judgment go to opus 5.5; the default review panel is opus 5.5 / grok. /setup-pstack changes any of it.
Before you install
- This is third-party code that can spawn subagents and run commands in your repository. Read the SKILL.md files before you install.
- Listed ≠ audited. AgentSkillsDir has not reviewed this code for security.
- Multi-model steps use the models you configure, and that usage may cost money with your providers.
FAQ
What are pstack skills?
Lauren Tan’s (poteto) open skill pack for rigorous AI-assisted engineering. Install it once, run setup, then start tasks with /poteto-mode so its playbooks invoke the other skills as needed.
Is this the Linux pstack command?
No. Linux pstack prints the stack trace of a running process. This page is about an AI coding skill pack that shares the name.
How is pstack different from gstack?
gstack is a separate skill pack by Garry Tan with its own install. Do not swap their commands; each has its own page in this directory.
Is pstack’s /unslop the same as the UI Unslop skill?
No. pstack’s /unslop cleans up writing and voice. The separately listed Unslop skill by heydamir redesigns UI. Same word, different jobs.
How do I install pstack in Cursor?
The official README gives /add-plugin pstack. We have not tested it ourselves yet.
How do I install pstack in Claude Code?
Add the cursor/plugins marketplace, then install pstack@cursor-plugins. Commands then carry a /pstack: prefix, such as /pstack:poteto-mode. We have not tested it ourselves yet.
What is npx skills add backnotprop/pstack?
The install for the standalone community mirror, for Codex, Pi, OpenCode and other agents. The official source remains the pstack folder in cursor/plugins.
Will AgentSkillsDir run /poteto-mode for me?
No. This is a source-linked directory entry. We do not host, run or proxy pstack or any multi-model panel. Listed ≠ audited.
Do I need Plannotator or another sponsor tool?
No. Plannotator sponsors the mirror. It is not a pstack dependency.
Are the star counts and playbook numbers guaranteed?
No. The source card shows the GitHub counts at the time we last checked; they change daily. Playbook and principle counts are what the upstream README says, not something we reproduced.
Related
Corrections and suggestions
Spotted something out of date, or want to suggest a skill? Email a correction. To suggest a resource, submit it for manual review.
