Search skills/

pstack (standalone mirror) Claude skills

A standalone mirror of Cursor's pstack skill stack, rewritten to work in any agent harness: 61 skills for rigorous, verified engineering work.

backnotprop/pstack 61 skills Code review & qualityPlanning, TDD & debugging
Our verdict Recommended with caveats

A good way to pick individual pstack skills, but the full stack is large and several skills depend on each other.

2/5 Depth expertise shipped ! Security Broad access 1d Maintenance Updated today 3.1K Adoption stars OK License MIT, commercial OK

What pstack (standalone mirror) does

This repository mirrors cursor/plugins/pstack, the opinionated skill collection by Lauren Tan (Poteto), and keeps it usable outside Cursor. Many of the original skills depended on Cursor-specific tools, so they were rewritten to work in any harness. The mirror also carries a community fix to the no-comments skill, which now keeps comments about outside rules it cannot verify and marks them for review rather than deleting them.

There are 61 skills averaging about 4.5 KB each. A group works alone, such as unslop, bro, how, tdd, typescript-best-practices, arena, swarm, interrogate, reflect, correct and statechart. Others call each other, and the README gives a table of which to install together, for example architect needs arena and how, and poteto-mode needs the principle-* skills and most others. poteto-mode is the router: tell it your goal and it picks a workflow. There is also an automations folder with Slack-driven issue triage and reproduce-and-fix skills that are marked as not model-invocable, fail closed without their configuration, and open only bounded draft pull requests after before-and-after proof. Install with npx skills add backnotprop/pstack and pick skills and agents from the list.

Good for

  • Teams who want pstack outside Cursor, installed skill by skill
  • Disciplined design, verification and review workflows
  • Choosing a few standalone skills instead of the whole stack

Not for

  • Users who want a small, self-contained skill set
  • Anyone who wants the fully pre-translated Claude Code build with a routing hook

What's inside depth 2/5

61 skills averaging 4.5 KB with some references, no scripts in the skill folders we inspected, no tests or evals.

Skills
61
Reference docs
35
Scripts
63
Tests / evals
381
Templates
0
Avg SKILL.md
4.4 KB

Counted from the repository file tree. How we score depth.

Dependencies and costs

  • GitHub CLI login for PR-related skills
  • Slack and an app-control adapter only for the optional automations

Security review

Broad access

This skill asks for more access than most, or sends data to another service. That is usually a design choice rather than bad intent, but you should know before you install it.

  • Pushes or merges without askingPlaybook has the agent merge verified PRs (gh pr merge --squash, optional --auto), retarget bases and push with --force-with-lease; SKILL.md says external/reversible actions proceed without asking and only force-push to shared branches always pauses (SKILL.md:84-86).skills/poteto-mode/playbooks/shipping.md:8

A large collection of workflow skills plus a few helper scripts. The scripts read PR status through the gh CLI and audit git worktrees locally; the shipping playbook lets the agent merge verified PRs and force-with-lease push stacks without a per-merge confirmation, and make-bot-ui tells the agent to install Tailscale via curl|sudo sh. We only have part of the repository (478 files missing), so this review is incomplete.

Worth knowing

  • Installs with curl | shIf Tailscale is missing, the skill tells the agent to run curl -fsSL https://tailscale.com/install.sh | sudo sh, then sudo tailscale up (third-party vendor host, with sudo).skills/make-bot-ui/SKILL.md:89
  • Installs packages or toolsRuns bun install --frozen-lockfile in the scripts folder on first use (commander 14.0.0 pinned; bun-types/typescript devDependencies are latest).skills/poteto-mode/scripts/bootstrap.ts:29
  • Relies on a CLI or package we could not reviewRelies on gh, optional origin CLI, bun, jq, rg and git; the audit script is read-only but scans ~/.cursor, ~/.claude and ~/.pi chat-transcript folders (file listing and mtimes only) to date the last session per worktree.skills/poteto-mode/scripts/worktree-audit.sh:63
  • Makes network requestsQueries the GitHub API via the authenticated gh CLI (read-only PR/check/review-thread data); worktree-audit.sh also runs a best-effort git fetch origin.skills/poteto-mode/scripts/watch-pr/github.ts:444

What we checked

The repository (581 files) on Oct 29, 2026. Not a full review: 478 of 581 files were not downloaded (large repository; we reviewed the skill folders); we read 6 of 15 code files; it relies on code outside the repository: 478 text files of the repo are not in our snapshot (including most SKILL.md files, agents/, automations/benny/ and parts of the playbooks), so those could not be checked; npm packages commander 14.0.0, bun-types, typescript (installed by bun at first run); gh, bun, jq, rg and the optional `origin` CLI; Tailscale install script from https://tailscale.com/install.sh.

NVIDIA SkillSpector static scan: 104 pattern hits (Agent Snooping 88, MCP Rug Pull 4, Supply Chain 3, Tool Misuse 3, Memory Poisoning 2, Excessive Agency 2). SkillSpector hits are mostly 'skill enumeration' from README links, doc mentions of git push --force / reset --hard as things to avoid or do carefully in playbooks, and one real curl|sh for Tailscale in make-bot-ui. The 'memory poisoning' hits are phrases like 'reset state' in docs.

A review, not a certification. Skills change after we check them — read the files before installing.

License

MIT — commercial use allowed.

Can it run unattended?

Not really. Most skills are interactive methods; the optional automations run on a schedule but depend on external configuration we could not verify.

Alternatives to pstack (standalone mirror)

pstack-claude

michael-denyer/pstack-claude

A port of Lauren Tan's pstack workflow skills to Claude Code, Codex, Copilot and others, with a poteto-mode router and 67 skills.

With caveats Check first Depth 3/5

Superpowers

obra/superpowers

A 15-skill development methodology that makes coding agents clarify the goal, write a plan, work through it with subagents and verify before calling it done.

Recommended Check first Depth 3/5

Matt Pocock's Skills

mattpocock/skills

Small, composable engineering skills: grill-me interviews, shared glossary, TDD, triage, prototyping, code review, handoff and a flow router.

Recommended Check first Depth 2/5

Agent Skills (Addy Osmani)

addyosmani/agent-skills

25 engineering-lifecycle skills and 9 slash commands (spec, plan, build, test, review, ship) that make coding agents follow senior-engineer workflows.

Recommended Check first Depth 3/5

FAQ

What is backnotprop/pstack?

It is a standalone mirror of Cursor's pstack skill stack by Lauren Tan, rewritten so the skills work in Claude Code, Codex, Pi, Cursor, OpenCode and other agents.

How do I install pstack?

Run npx skills add backnotprop/pstack and pick the skills and agents you want from the list. Install related skills together, as the README's dependency table shows.

backnotprop/pstack vs michael-denyer/pstack-claude?

Both carry pstack to non-Cursor agents. This mirror is a plain skills folder installed with the skills CLI, while pstack-claude pre-translates tool names for Claude Code and adds a routing hook and helper scripts.

Sources: GitHub repository (stars, license, file tree, last push Oct 28, 2026). Review based on the README and SKILL.md files as published.