SLOPSHOPPER

pstack

poteto's rigorous engineering skills (poteto-mode, how/why, architect, verification), ported from Cursor and kept in sync with upstream

newpanespinnerguardcommandtoast
★ 1v0.15.15+claude.7MITupdated 2026-10-08PickleBoxer/pstack
A shopper browsing a rack in a slop shop
Preview · a replayed session in a sandbox
claude · ~/work/app · pstack
│ ┃ pstack ✕ › fix the failing auth test and add an audit log c╭────────────────╮ │ ┃ No pstack skills found under │ pstack │ │ ┃ /plugins/pstack/skills. ⏺ Read(src/auth.ts) │ poteto mode on │ │ ⎿ Read 6 lines ╰────────────────╯ │ ⏺ Update(src/auth.ts) │ ⎿ Added 2 lines, removed 1 line │ ⏺ Bash(bun test) │ ⎿ 3 pass, 1 fail │ │ ● Done. refresh now rejects expired claims and logs an audit event. │ │ ✻ Worked for 42s · done 4:20 PM │ │ › /poteto │ │ ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── › ? for shortcuts auto-accept edits & [ poteto ] ♛ on

Draws

Pane · pstack
No pstack skills found under /plugins/pstack/skills.
README

pstack

i'm poteto. i'm not a president or ceo, but i've worked with millions of lines of code at Meta, Netflix, and Cursor. i'm also on the react core team where i help build and maintain react compiler.

there's a growing sense that ai writes too much slop code. i agree. i don't want to ship like a team of twenty slop artists. throughput without quality is not a goal i aspire to. if you want to go fast, go deep first.

pstack is my answer. these are the same skills i use everyday to ship high quality code at Cursor. this turns cursor into a real engineering team. the goal is not to maximize loc, in fact it's the opposite. pstack helps you write less, but higher quality code.

pstack gives you fearless parallelism. when you can go deep on one agent and trust it to write good, verifiable code, you can truly parallelize with confidence. start multiple agents up with poteto-mode and trust that they'll apply rigorous engineering principles to their work.

cursor gives you the best of all worlds. every frontier model has its strengths and weaknesses. use any model with pstack. in fact, many of my skills use multi-model workflows to take advantage of each model's unique strengths.

fork it. improve it. make it yours. PRs are welcome!

install

/add-plugin pstack

get started

two steps:

  1. run /setup-pstack, pick a reasoning budget, and choose which models you want.
  2. use /poteto-mode whenever you're doing anything that requires rigor.

new here? the pstack guide walks you through a first real task, from setup and prompting through verification and overnight runs. stuck, or unsure which skill fits? ask /poteto-help.

that's it. the other skills are situational; the mode skill uses them for you as needed. out of the box the mode splits work by model strength: code delegates (feature, refactoring, bug fix, perf, hillclimb) go to grok, while the hardest changes, prose, and judgment go to opus 5.5. the default panel is opus 5.5 / grok. /setup-pstack changes any of it.

usage

use /poteto-mode at the start of a task. it reads your request, picks from a set of playbooks, and runs the other skills as the steps need them.

just use /poteto-mode

this skill is the main shortcut. i use it whenever i need the agent to do rigorous engineering work. it comes with twenty-three playbooks:

/poteto-mode this pr has a subtle bug where the scroll drifts every 750ms even when idle. repro
first, then fix and verify.
/poteto-mode i'm going to bed. land the stack even if ci flakes. i want everything merged by
morning.
playbookfor
investigationa read-only question. how does x work, why was y built this way, are we sure.
bug fixreproduce a defect, root-cause it, and fix with runtime evidence.
perftrace a measured slowness and improve it against a baseline.
hillclimbsustained, scientific improvement of one metric against a target, looping hypotheses with before/after measurement and one commit per accepted win.
runtime forensicsdiagnose a live symptom (leak, idle-cpu spin, glitch) from instrumentation.
trace forensicsdiagnose a captured profiling artifact (cpuprofile, trace, spindump, heap snapshot).
featurenew or changed behavior, built from a named data shape.
refactoringa behavior-preserving change to structure or shape.
prototypea throwaway sketch to make a design or behavioral decision cheaply, or to settle an empirical fork by observing it.
visual paritypixel-exact ui equivalence between two implementations.
authoring a skillwriting or editing a SKILL.md.
evaltest how a skill or prompt change affects agent behavior, blinded.
babysitdrive a pr or a stack to merge-ready: conflicts, review threads, ci.
shippingindependently verify a green stack, then land the contiguous verified run bottom-up through github by default or origin when available.
autonomous rundrive a long task to completion without stopping.
orchestratea standing project handed to one coordinator chat: multi-day, many stacked prs, fleets of subagents.
autopilot-fullrun independent prs to merged with one owner per pr and a root swarm verdict on each round, from the code-ready head on.
autopilot-stackbuild and verify one linear base-branch stack for the operator to review and land.
session pickupresume or take over a prior agent's in-flight work.
pause safelysuspend in-flight work cleanly so it can be resumed later.
multi-phase planwork that spans phases or stacked PRs.
worktree cleanupreclaim disk by pruning merged or abandoned worktrees and stale ios simulators, safety-gated.
opening a propen a ready pr from small ordered commits with a conventional commits title and a briefing-style body. invoked at the end of every other playbook.

when invoked it:

  1. matches your task to a playbook and opens a todo list whose first items are its steps, copied in verbatim.
  2. routes to the other skills as the steps fire.
  3. writes unslopped replies framed for the consumer and the maintainer.

the full rules and playbooks live in skills/poteto-mode/SKILL.md.

to keep /poteto-mode on across turns, pick it from the / menu and press option+enter (mac) or alt+enter (windows) instead of enter. that makes it a custom mode, which cursor offers in the agents window and the cli. it stays in context every turn, applies itself when a playbook matches or the task needs rigor, and stays out of the way otherwise. plain enter attaches it to one message only. say so to opt out, or exit the mode to turn it off.

/poteto-mode works extremely well with cursor's /loop command. you can make cursor work for many hours without sacrificing rigor.

skills

/poteto-mode runs most of these for you when a step needs them (how, why, architect, arena, swarm, interrogate, unslop, no-comments, technical-writing, tdd, and the principles). the table below is for when you want one directly:

/how do we cancel runs? do we have an n+1 when we look up every run to cancel?
/interrogate review this pr.
skilluse it when
/poteto-modedefault entry point for any non-trivial task.
/poteto-helpyou're new to pstack, or unsure which skill, playbook, or principle fits. finds out what you're trying to do, answers that part, and hands you a prompt to type. runs only when you type /poteto-help.
/howyou want a walkthrough of how a subsystem works.
/whyyou want to know why something was built this way. discovers available MCPs at run time and queries each evidence category in parallel (source control, issue tracker, long-form docs, real-time chat, infra observability, error tracking, analytics warehouse).
/recallyou're starting or resuming work and want your recent context on a topic rebuilt from your own chat history and the shared record, handed back as a tight current-state brief.
/blast-radiusyou have a small-looking change and want to know what else it could break, with the one fact it's safe because of proven by running code, not asserted.
/architectyou're about to write code that crosses a function boundary and want the caller's usage, types, and module shape settled first.
/arenayou want N parallel attempts at the same thing, then to grab the best parts of each.
/swarmyou want N parallel workers across different slices or races, then one aggregated report.
/interrogateyou have a diff and want different models to try to break it, including a strict code-quality lens.
/automate-meyou want your own -mode skill, drafted from how you've actually worked.
/make-bot-uiyou want a page or dashboard whose buttons wake a Grok Bot over a webhook, including the sender-key handoff and Tailscale.
/setup-pstackyou want to pick which models pstack uses per role. detects your models and writes a config rule.
/reflecta long task landed and you want the recipe captured as a skill edit.
/correctyou keep correcting agents for the same mistakes. mines history for mistake classes, fixes each at the highest level that works (architecture, then types, lint, and ci, then tests, with docs last), and keeps a table pairing each rule with what enforces it.
/teachyou want to actually understand a change or subsystem, not just have it summarized. runs how + why and weaves one plain explanation, built up diagram by diagram.
/tddyou're fixing a bug and there's a cheap local test path. write the failing test first, then the fix.
/benchmark-checklistyou ran a benchmark or measured a speedup or regression. vets the number (limiter, tuning, errors, repeat runs, end-to-end relevance) before you report or act on it.
/no-commentsstrip comments before review; spawns Comment Sicko, fixes accepted findings, offers encodings for claimed constraints.
/typescript-best-practicesyou're reading or editing typescript. grounds the type-system-discipline principle in syntax.
/figure-it-outno bundled playbook fits. designs a rigorous, auditable playbook for the task.
/show-me-your-workyou want a reviewable decision trail. logs decisions to a tsv you can commit.
/create-verification-skillyour project has no scripted way to prove app behavior. generates a project-local verify skill with a feature map, for any language or platform.
/maintain-verification-skillyour verify skill's feature map has drifted from the app. source wave + one live pass, at most one PR of proven corrections.
/unslopyou're cleaning up writing. removes AI tells.
/broyou want the last message restated in plain human language, no jargon.
/technical-writinglayered doc standard (Diátaxis + Google developer style + STE + Global English) for docs, RFCs, readmes, PR descriptions, commit messages.

examples

mostly i type /poteto-mode at the start of a task and let it route to a playbook. the other skills fire as the steps need them. a few i reach for directly.

bug fix:           /poteto-mode this pr has a subtle bug where the scroll drifts every 750ms even
                   when idle. repro first, then fix and verify.
perf:              /poteto-mode a big list takes a second or two to load even though we virtualize.
                   run a cpu trace and tell me why.
feature:           /poteto-mode build a small feature behind a feature flag. verify it really works.
prototype:         /poteto-mode build two prototypes of the markdown renderer so we can compare.
                   spawn an agent for each.
multi-phase:       /poteto-mode open source these skills as a plugin. nothing internal leaks, work
                   in a temp dir, show me the dependency graph first.
overnight run:     /poteto-mode i'm going to bed. land the stack even if ci flakes. i want
                   everything merged by morning.
babysit:           /poteto-mode check on pr 123. anything outstanding?
visual parity:     /poteto-mode the row spacing is too tall when this flag is on. the second image
                   is correct. repro and fix until it matches.
figure it out:     /poteto-mode i'm stepping away. migrate every caller from the synchronous store
                   to the new async one, keeping behavior identical. i want to trust it was done
                   right when i'm back.
how:               /how do we cancel runs? do we have an n+1 when we look up every run to cancel?
why:               /why is this feature flag not on yet?
architect:         design this instrumentation to be high signal with no false positives. /architect
                   this first.
arena:             /arena take my prompt to the arena verbatim. i want to compare their proposals
                   with yours.
swarm:             /swarm check every package under packages/ against its check.sh. one worker per
                   package. one report.
interrogate:       /interrogate review this pr.
tdd:               /tdd implement
unslop:            can we unslop and tighten the new changes?
reflect:           /reflect that took too long. capture what we learned so the next run doesn't
                   repeat it.
correct:           /correct
show-me-your-work: /show-me-your-work keep a decision trail i can review when i'm back.
automate-me:       /automate-me
help:              /poteto-help which skill should i use to review this branch?

the poteto-agent and Comment Sicko subagents

pstack also ships a subagent that runs my style end to end. spawn it from a parent agent via subagent_type: "poteto-agent". it reads poteto-mode in full, including its inline principles index, before doing any work. substituting generalPurpose skips that read and drifts.

/poteto-mode and subagent_type: "poteto-agent" route through the same wrapper.

pstack also ships Comment Sicko, a read-only comment reviewer available as subagent_type: "Comment Sicko". usually invoke it through /no-comments, not directly.

principles

twenty-four short skills, one principle each. poteto-mode indexes them inline and reads that index at task start. the standalone files are there so other skills can reference a principle by name, and so the index can point at the full rule for each.

principlegrouprule
laziness-protocolcoreBias toward deletion and the smallest change that solves the problem.
foundational-thinkingcoreApply before writing logic: choosing core types and data structures, sequencing scaffold-vs-feature work, asking what concurrent actors share. Get the data structures right so downstream code becomes obvious.
redesign-from-first-principlescoreRedesign as if the requirement had been a foundational assumption from day one, instead of bolting it on.
attack-the-premisecoreApply when two or more fixes that share one premise have failed the same gate. Take a census of which actors hold the imbalance before the next fix, then question the premise instead of writing another fix that assumes it.
subtract-before-you-addcoreRemove dead weight, redundant validators, and stub references first, then build on the simpler base.
minimize-reader-loadcoreCount layers between question and answer, and hidden state in the reader's head; collapse one-caller wrappers and shrink mutable scope.
outcome-oriented-executioncoreApply during planned rewrites and migrations with explicit phase boundaries. Converge on the target architecture; don't preserve smooth intermediate states with throwaway compatibility code.
experience-firstcoreChoose user delight over implementation convenience; ship fewer polished features over more rough ones.
exhaust-the-design-spacecoreBuild 2-3 competing prototypes and compare side by side before committing.
build-the-levercoreApply to any non-trivial work, not just bulk work: edits, migrations, analyses, checks. Build the tool that does it or proves it (codemod, script, generator, or a skill your subagents follow) instead of working by hand. The tool is the artifact a reviewer can rerun.
model-the-domainarchitectureEncode the domain in a structure instead of scattered conditionals.
boundary-disciplinearchitectureConcentrate guards at system boundaries (CLI, config, network, external APIs); trust internal types and keep business logic in pure functions.
type-system-disciplinearchitectureMake illegal states unrepresentable, brand semantic primitives, parse external data at boundaries, refuse to lie to the compiler, exhaust variants, derive from authoritative schemas.
make-operations-idempotentarchitectureConverge to the same end state regardless of partial prior runs.
migrate-callers-then-delete-legacy-apisarchitectureMigrate callers and delete the old API in the same wave instead of preserving compatibility layers.
separate-before-serializing-shared-statearchitectureEliminate the sharing first; serialize structurally only when one shared writer is a real invariant.
prove-it-worksverificationApply after completing a task, before declaring done. Verify against the real artifact (run the feature, read the actual value, inspect the diff), not a proxy, self-report, or 'it compiles.'.
fix-root-causesverificationTrace each symptom to its root cause and fix it there; reproduce first, ask why until you reach it, resist nil-check guards that silence crashes.
sequence-verifiable-unitsverificationApply to multi-step work (sweeps, migrations, runs of similar edits) and to how you stack commits and PRs. Break work into small units that each end in a verifiable state, check each before the next, and order delivery so the sequence proves itself to a reviewer.
test-behavior-not-implementationverificationApply when you write, change, or keep a test. Call the code the way its users do and assert the result they observe against a literal expected value. If the test would still pass when every imported function returns undefined, rewrite the assertion or delete the test.
explain-the-numberverificationApply before you trust, report, or act on a number you measured: a speedup, a regression, a throughput, a latency, or an eval result. Find what limits it, and rule out that it measured something other than the work you think.
guard-the-context-windowdelegationRoute bulk to subagents; keep summaries in the main thread, not raw payloads.
never-block-on-the-humandelegationProceed, present the result, let the human course-correct after the fact; reserve confirmation for irreversible actions.
encode-lessons-in-structuremetaEncode the rule as a lint, metadata flag, runtime check, or script instead of more text.

not shipped here

a few things poteto-mode references but doesn't bundle:

  • /deslop and the deslop skill ship in the cursor-team-kit plugin.
  • control-cli (for CLIs and TUIs) and control-ui (for browser, Electron, web) ship in cursor-team-kit too.
  • /create-skill is a cursor built-in. cursor also ships a built-in /babysit; inside poteto-mode, the babysit playbook supersedes it for pr-status requests.

install cursor-team-kit alongside pstack if you want the full set.

why are there no planning skills?

cursor already has a great plan mode which works great with pstack. but personally, i don't believe in planning. the best spec is code. if you do want to make a plan, /poteto-mode covers it, but it's not a default.

make it yours

poteto-mode is my style. you may not want exactly that.

type /automate-me. it mines your recent transcripts, drafts a <your-name>-mode skill from how you've actually worked, and routes through pstack underneath. you keep pstack as the base and end up with your own routing skill alongside poteto-mode.

models are configurable too. type /setup-pstack. it detects the models you have access to and writes a small always-applied rule mapping each role (code, judgment, the review panels) to a model. every skill reads it and falls back to sensible defaults when the rule is absent, so you override only what you want.

when a default changes, a rule written before the change still pins the old default. delete those role lines, or delete the file, then run /setup-pstack again. a rerun keeps any role whose model diff

Source 5 files
hooks/register.tsx 334 lines
1import { atom, read, update } from 'claude-code'
2import type { EngineInterface, Register } from 'claude-code'
3
4import type { PstackLoaded, PstackSkill } from '../types'
5import { descriptionFrom, grouped } from './catalog'
6import { POTETO_OFF, potetoSection, storeKey } from './poteto'
7import { agentModel, agentType, harnessNote, modelsFrom } from './translate'
8
9// The org's security plugin can bypass user-tier skill.prompt hooks, so the note rides prompt.submit and Read instead
10const PSTACK_COMMAND = /^\/pstack:(\S+)/
11const PANE = 'pstack'
12const HIDDEN_KEY = 'hidden'
13const LOADED_TAB = 'Loaded'
14const LOADED_NOTES: Record<PstackLoaded['via'], string> = {
15  read: 'read this session',
16  run: 'run this session',
17  skill: 'loaded by Claude this session',
18}
19const WORDMARK =['█▀█ █▀ ▀█▀ ▄▀█ █▀▀ █▄▀', '█▀▀ ▄█  █  █▀█ █▄▄ █ █']
20
21const poteto = atom({ plugin: 'pstack', key: 'poteto' } as const, false)
22const skills = atom({ plugin: 'pstack', key: 'skills' } as const, [])
23const hidden = atom({ plugin: 'pstack', key: 'hidden' } as const, [])
24const group = atom({ plugin: 'pstack', key: 'group' } as const, '')
25const loaded = atom({ plugin: 'pstack', key: 'loaded' } as const, [])
26const pinned = atom({ plugin: 'pstack', key: 'pinned' } as const, false)
27
28// Mirrors the project's stored choice into session state, which prompt.submit reads
29async function loadPoteto($: EngineInterface, isDefault: boolean): Promise<boolean> {
30  const stored = await $.store.get(storeKey(await $.session.root()))
31  const isOn = typeof stored === 'boolean' ? stored : isDefault
32  await update($, poteto, () => isOn)
33
34  return isOn
35}
36
37async function togglePoteto($: EngineInterface, isDefault: boolean): Promise<void> {
38  const isOn = !(await loadPoteto($, isDefault))
39  await $.store.set(storeKey(await $.session.root()), isOn)
40  await update($, poteto, () => isOn)
41  $.ui.toast(isOn ? 'poteto mode on' : 'poteto mode off')
42}
43
44// Every folder under skills/ with its frontmatter description, read once per session
45async function loadSkills($: EngineInterface): Promise<void> {
46  const entries = await $.fs.list(`${$.plugin.root}/skills`).catch(() => [])
47  const found: PstackSkill[] = await Promise.all(
48    entries
49      .filter(entry => entry.kind === 'dir')
50      .map(async entry => ({
51        name: entry.name,
52        description: descriptionFrom(String(await $.fs.read(`${$.plugin.root}/skills/${entry.name}/SKILL.md`).catch(() => ''))),
53      })),
54  )
55  await update($, skills, () => found)
56
57  const stored = await $.store.get(HIDDEN_KEY)
58  await update($, hidden, () => (Array.isArray(stored) ? stored.filter(name => typeof name === 'string') : []))
59}
60
61// Only the pane header shows it, so a manifest that does not parse must not stop session.start
62function versionFrom(manifest: string): string {
63  try {
64    const version: unknown = JSON.parse(manifest).version
65
66    return typeof version === 'string' ? version : ''
67  } catch {
68    return ''
69  }
70}
71
72async function markLoaded($: EngineInterface, name: string, via: PstackLoaded['via']): Promise<void> {
73  await update($, loaded, list => (list.some(one => one.name === name) ? list : [...list, { name, via }]))
74}
75
76async function toggleHidden($: EngineInterface, name: string): Promise<void> {
77  const next = await update($, hidden, names => (names.includes(name) ? names.filter(one => one !== name) : [...names, name]))
78  await $.store.set(HIDDEN_KEY, next)
79  $.ui.invalidate('command.describe')
80}
81
82export const register: Register = (on, options) => {
83  const models = modelsFrom(options)
84  const isPotetoDefault = options.poteto_default === true
85  // Loops (main is '') that already carry the note this turn
86  const noted = new Set<string>()
87  let potetoSkill = ''
88  let version = ''
89
90  on('session.start', async ($, e, next) => {
91    potetoSkill = String(await $.fs.read(`${$.plugin.root}/skills/poteto-mode/SKILL.md`).catch(() => ''))
92    version = versionFrom(String(await $.fs.read(`${$.plugin.root}/.claude-plugin/plugin.json`).catch(() => '')))
93    await update($, loaded, () => [])
94    await update($, pinned, () => false)
95    await $.command.register({ name: 'poteto', description: 'Toggle poteto mode for this project' })
96    await $.command.register({ name: 'pstack', description: 'Browse, run and hide pstack skills' })
97    await loadPoteto($, isPotetoDefault)
98    await loadSkills($)
99
100    return next(e)
101  })
102
103  on('command.run', { command: 'poteto' }, async $ => {
104    await togglePoteto($, isPotetoDefault)
105
106    return {}
107  })
108
109  on('command.run', { command: 'pstack' }, async $ => {
110    await $.ui.open({ id: PANE, title: 'pstack', closeOnEscape: true })
111
112    return { text: 'pstack pane opened' }
113  })
114
115  on('command.describe', async ($, e, next) => {
116    const name = e.command.startsWith('pstack:') ? e.command.slice('pstack:'.length) : undefined
117
118    if (name === undefined || !(await read($, hidden)).includes(name)) {
119      return next(e)
120    }
121
122    return next({ ...e, isHidden: true })
123  })
124
125  on('turn.start', ($, e, next) => {
126    noted.clear()
127
128    return next(e)
129  })
130
131  on('prompt.submit', async ($, e, next) => {
132    const command = e.text.match(PSTACK_COMMAND)?.[1]
133    const context = [...(e.context ?? [])]
134
135    if (command !== undefined) {
136      noted.add('')
137      await markLoaded($, command, 'run')
138      context.push(harnessNote($.plugin.root, models))
139    }
140
141    // The conversation catches up with the toggle on the next prompt the model reads
142    if (command !== undefined || !e.text.startsWith('/')) {
143      const isOn = await read($, poteto)
144
145      if (isOn !== (await read($, pinned))) {
146        context.push(isOn ? potetoSection($.plugin.root, potetoSkill, harnessNote($.plugin.root, models)) : POTETO_OFF)
147        await update($, pinned, () => isOn)
148      }
149    }
150
151    return next(context.length === (e.context ?? []).length ? e : { ...e, context })
152  })
153
154  on('tool.call', { tool: 'Read' }, async ($, e, next) => {
155    const result = await next(e)
156    const loop = e.agentId ?? ''
157    const skillsDir = `${$.plugin.root}/skills/`
158    const isPstackFile = e.file_path.startsWith(skillsDir) || e.file_path.startsWith(`${$.plugin.root}/agents/`)
159
160    // A subagent's reads sit in its own context, not the conversation the pane reports on
161    if (e.file_path.startsWith(skillsDir) && result.deny === undefined && e.agentId === undefined) {
162      await markLoaded($, e.file_path.slice(skillsDir.length).split('/')[0] ?? '', 'read')
163    }
164
165    if (!isPstackFile || result.deny !== undefined || noted.has(loop)) {
166      return result
167    }
168
169    noted.add(loop)
170
171    return { ...result, context: [...(result.context ?? []), harnessNote($.plugin.root, models)] }
172  })
173
174  on('tool.call', { tool: 'Skill' }, async ($, e, next) => {
175    const result = await next(e)
176    const name = e.skill.match(/^pstack:(principle-[a-z-]+)$/)?.[1]
177
178    if (name !== undefined && result.deny === undefined && e.agentId === undefined) {
179      await markLoaded($, name, 'skill')
180    }
181
182    return result
183  })
184
185  on('session.compact', async ($, e, next) => {
186    const result = await next(e)
187
188    if (e.agentId === undefined && e.trigger !== 'precompute' && !('skip' in result)) {
189      await update($, loaded, () => [])
190      await update($, pinned, () => false)
191    }
192
193    return result
194  })
195
196  on('agent.spawn', ($, e, next) =>
197    next({ ...e, subagentType: agentType(e.subagentType), model: agentModel(e.model, models) }),
198  )
199
200  // Draws the footer's modes itself: the desktop skips plain mode labels, and the poteto one is a toggle
201  on('ui.render', { component: 'SessionMode' }, async ($, e) => {
202    const { Box, Button, Text } = $.ui.resolve(e)
203    const isOn = await read($, poteto)
204
205    return (
206      <Box flexDirection="row">
207        {e.props.modes.map(mode => (
208          <Text key={mode} dimColor>
209            {mode} &{' '}
210          </Text>
211        ))}
212        <Button
213          key="poteto"
214          variant="primary"
215          dimColor={!isOn}
216          label="poteto"
217          onPress={() => void togglePoteto($, isPotetoDefault)}
218        />
219        <Text color={isOn ? 'success' : undefined} dimColor={!isOn}>
220          {isOn ? ' ♛ on' : ' ♛ off'}
221        </Text>
222      </Box>
223    )
224  })
225
226  on('ui.render', { component: 'Pane', requestId: PANE }, async ($, e) => {
227    const { Box, Button, Text } = $.ui.resolve(e)
228    const all = await read($, skills)
229    const hiddenNames = await read($, hidden)
230    const isOn = await read($, poteto)
231
232    if (all.length === 0) {
233      return <Text dimColor>No pstack skills found under {$.plugin.root}/skills.</Text>
234    }
235
236    const isPinned = await read($, pinned)
237    const pinnedRows = isPinned ? [{ name: 'poteto-mode', note: 'in context while poteto is on' }] : []
238    const sessionLoaded = (await read($, loaded))
239      .filter(one => !(isPinned && one.name === 'poteto-mode'))
240      .map(one => ({ name: one.name, note: LOADED_NOTES[one.via] }))
241    const loadedRows = [...pinnedRows, ...sessionLoaded]
242    const loadedNames = new Set(loadedRows.map(one => one.name))
243
244    const groups = grouped(all).slice(0, 8)
245    const tabs = [...groups.map(one => ({ title: one.title, label: one.title })), { title: LOADED_TAB, label: `${LOADED_TAB} ${loadedRows.length}` }]
246    const chosen = await read($, group)
247    const selected = tabs.find(one => one.title === chosen) ?? tabs[0]
248    const rows =
249      selected?.title === LOADED_TAB
250        ? loadedRows.map(one => ({ name: one.name, description: one.note }))
251        : (groups.find(one => one.title === selected?.title)?.skills ?? [])
252
253    return (
254      <Box flexDirection="column">
255        <Box flexDirection="column">
256          {WORDMARK.map((line, index) => (
257            <Text key={`wordmark-${index}`} color="warning">
258              {line}
259            </Text>
260          ))}
261          <Text color={isOn ? 'success' : undefined} dimColor={!isOn} wrap="truncate-end">
262            {isOn ? '♛ poteto on: poteto-mode rides the conversation, every task gets the playbooks' : '♛ poteto off: /poteto or the footer toggle pins poteto-mode'}
263          </Text>
264          <Text dimColor wrap="truncate-end">
265            new here? press 1, then /poteto-help
266          </Text>
267          <Text dimColor wrap="truncate-end">
268            {`${loadedRows.length} loaded · ${hiddenNames.length} hidden · ${all.length} skills${version === '' ? '' : ` · v${version}`}`}
269          </Text>
270        </Box>
271        <Box flexDirection="row" flexWrap="wrap" columnGap={2} marginTop={1}>
272          {tabs.map((tab, index) => {
273            const isSelected = tab.title === selected?.title
274            const hotkey = String(index + 1)
275            // The terminal draws a plain Button with a hotkey as `1: label`
276            const width = hotkey.length + 2 + tab.label.length
277
278            return (
279              <Box key={`tab-${tab.title}`} flexDirection="column">
280                <Button
281                  key={`group-${tab.title}`}
282                  plain
283                  hotkey={hotkey}
284                  dimColor={!isSelected}
285                  label={tab.label}
286                  onPress={() => void update($, group, () => tab.title)}
287                />
288                <Text color={isSelected ? 'suggestion' : undefined} dimColor={!isSelected}>
289                  {(isSelected ? '━' : '─').repeat(width)}
290                </Text>
291              </Box>
292            )
293          })}
294        </Box>
295        <Box flexDirection="column" marginTop={1}>
296          {rows.length === 0 && <Text dimColor>No pstack skill is in the context yet.</Text>}
297          {rows.map(skill => {
298            const isHidden = hiddenNames.includes(skill.name)
299            const isLoaded = loadedNames.has(skill.name)
300
301            return (
302              <Box key={skill.name} flexDirection="row" gap={1}>
303                <Text color={isLoaded ? 'success' : undefined}>{isLoaded ? '●' : ' '}</Text>
304                <Button
305                  key={`run-${skill.name}`}
306                  plain
307                  dimColor={isHidden}
308                  label={`/${skill.name}`}
309                  onPress={() => void $.prompt.fill({ text: `/pstack:${skill.name} `, mode: 'replace' })}
310                />
311                <Box flexGrow={1} flexShrink={1} overflow="hidden">
312                  <Text dimColor wrap="truncate-end">
313                    {skill.description}
314                  </Text>
315                </Box>
316                <Button
317                  key={`hide-${skill.name}`}
318                  plain
319                  dimColor
320                  label={isHidden ? 'show' : 'hide'}
321                  onPress={() => void toggleHidden($, skill.name)}
322                />
323              </Box>
324            )
325          })}
326        </Box>
327        <Box marginTop={1}>
328          <Text dimColor>Press a skill to put it in the prompt · 1-9 switch tabs · hidden skills stay out of the / menu but still run when typed</Text>
329        </Box>
330      </Box>
331    )
332  })
333}
334
hooks/catalog.ts 47 lines
1// The /pstack pane's grouping of upstream skills; a skill upstream adds later lands in Other until it is placed here
2
3import type { PstackSkill } from '../types'
4
5const GROUPS: ReadonlyArray<readonly [string, readonly string[]]> = [
6  ['Start here', ['poteto-mode', 'poteto-help']],
7  ['Understand', ['how', 'why', 'teach', 'recall', 'blast-radius', 'bro']],
8  ['Design', ['architect', 'arena', 'interrogate', 'figure-it-out']],
9  [
10    'Build and verify',
11    ['tdd', 'swarm', 'create-verification-skill', 'maintain-verification-skill', 'benchmark-checklist', 'show-me-your-work'],
12  ],
13  ['Clean and write', ['unslop', 'no-comments', 'technical-writing', 'typescript-best-practices']],
14  ['Improve the setup', ['correct', 'reflect', 'automate-me']],
15]
16
17export type Group = { title: string; skills: PstackSkill[] }
18
19export function descriptionFrom(skill: string): string {
20  const frontmatter = skill.match(/^---\n([\s\S]*?)\n---/)?.[1] ?? ''
21  const description = frontmatter.match(/^description:\s*(.+)$/m)?.[1]?.trim() ?? ''
22
23  return description.replace(/^(["'])(.*)\1$/, '$2')
24}
25
26function groupOf(name: string): string {
27  if (name.startsWith('principle-')) {
28    return 'Principles'
29  }
30
31  return GROUPS.find(([, names]) => names.includes(name))?.[0] ?? 'Other'
32}
33
34export function grouped(skills: readonly PstackSkill[]): Group[] {
35  const titles = [...GROUPS.map(([title]) => title), 'Principles', 'Other']
36  const order = (name: string) => GROUPS.flatMap(([, names]) => names).indexOf(name)
37
38  return titles
39    .map(title => ({
40      title,
41      skills: skills
42        .filter(skill => groupOf(skill.name) === title)
43        .sort((a, b) => order(a.name) - order(b.name) || a.name.localeCompare(b.name)),
44    }))
45    .filter(group => group.skills.length > 0)
46}
47
hooks/poteto.ts 18 lines
1// Cursor pins poteto mode's SKILL.md as a Custom Mode. Desktop sessions raise no prompt.compose, so here it rides a prompt's context once
2
3export function potetoSection(root: string, skill: string, note: string): string {
4  const body = skill.replace(/^---\n[\s\S]*?\n---\n*/, '').trim()
5
6  return [
7    `Poteto mode is on for this project. Follow the poteto-mode skill below on every turn until a later note says poteto mode is off; no Read of its SKILL.md needed. Its playbooks and references live under \`${root}/skills/poteto-mode/\`.`,
8    note,
9    body,
10  ].join('\n\n')
11}
12
13export const POTETO_OFF = 'Poteto mode is now off for this project. Stop following the poteto-mode skill given earlier in this conversation and work in your normal style.'
14
15export function storeKey(projectRoot: string): string {
16  return `poteto:${projectRoot}`
17}
18
hooks/translate.ts 57 lines
1// Pure mapping from upstream's Cursor vocabulary to Claude Code, kept free of `$` so tests can call it directly
2
3export type Models = {
4  search: string
5  code: string
6  judgment: string
7}
8
9const AGENT_TYPES: Record<string, string> = {
10  generalPurpose: 'general-purpose',
11  'poteto-agent': 'pstack:poteto-agent',
12  'Comment Sicko': 'pstack:Comment Sicko',
13}
14
15const INHERIT = new Set(['inherit', 'inherit-parent', 'auto'])
16
17export function modelsFrom(options: Readonly<Record<string, unknown>>): Models {
18  const pick = (key: string, fallback: string) => (typeof options[key] === 'string' ? (options[key] as string) : fallback)
19
20  return {
21    search: pick('search_model', 'haiku'),
22    code: pick('code_model', 'sonnet'),
23    judgment: pick('judgment_model', 'opus'),
24  }
25}
26
27export function agentType(type: string): string {
28  return AGENT_TYPES[type] ?? type
29}
30
31// undefined means omit `model`, so the subagent runs on its own or the parent's model
32export function agentModel(model: string | undefined, models: Models): string | undefined {
33  if (model === undefined || INHERIT.has(model)) {
34    return undefined
35  }
36
37  const mapped = model.startsWith('grok-') ? models.code : model.startsWith('claude-opus-') ? models.judgment : model
38
39  return INHERIT.has(mapped) ? undefined : mapped
40}
41
42function named(model: string): string {
43  return INHERIT.has(model) ? 'the parent model (omit `model`)' : `\`${model}\``
44}
45
46export function harnessNote(root: string, models: Models): string {
47  return [
48    'pstack was written for Cursor. You are in Claude Code, so translate its instructions as you follow them:',
49    `- Another pstack skill: Read \`${root}/skills/<name>/SKILL.md\` and follow it. Principles are the exception: load one with the Skill tool as \`pstack:principle-<slug>\`. The Skill tool refuses every other pstack skill. Read pstack files one per Read call, in parallel when there are several, never through \`cat\`, \`sed\` or \`head\` in Bash.`,
50    '- `Task` tool or call: the `Agent` tool. Subagent types: `generalPurpose` is `general-purpose`, `poteto-agent` is `pstack:poteto-agent`, `Comment Sicko` is `pstack:Comment Sicko`. Readonly agent mode: the `Explore` agent. `environment: "cloud"`: `isolation: "worktree"`.',
51    '- `AskQuestion`: `AskUserQuestion`.',
52    `- Models: \`~/.cursor/rules/pstack-models.mdc\` and \`/setup-pstack\` do not exist here, so use the defaults below. Grok slugs mean ${named(models.code)}. \`claude-opus-*\` slugs mean ${named(models.judgment)}. Fast searches (the recall fan-out, readonly exploration) use ${named(models.search)}. \`inherit\`, \`inherit-parent\` and \`auto\` mean omit \`model\`. A panel of Opus and Grok is one judgment-model and one code-model agent.`,
53    '- Not available: cursor-team-kit (`deslop`, `control-ui`, `control-cli`), `/make-bot-ui`, and Cursor\'s `create-skill` (use `skill-creator` if installed). Use the project\'s own verification skill or browser tools instead, and say what you skipped.',
54    "- The user's CLAUDE.md files and rules win over pstack's Autonomy guidance. If they say never commit, do not commit.",
55  ].join('\n')
56}
57
types/index.d.ts 26 lines
1// Whether poteto mode is on for the session's project, mirrored from $.store
2export type PstackPoteto = boolean
3
4// One skill folder under skills/, as the /pstack pane lists it
5export type PstackSkill = { name: string; description: string }
6
7// A pstack skill whose text sits in the main conversation: its files read, or typed as /pstack:<name>
8export type PstackLoaded = { name: string; via: 'read' | 'run' | 'skill' }
9
10declare module 'claude-code' {
11  interface PluginState {
12    pstack: {
13      poteto: PstackPoteto
14      skills: PstackSkill[]
15      // Skill names hidden from the / menu, mirrored from $.store
16      hidden: string[]
17      // The group the /pstack pane shows on the right; '' is the first group
18      group: string
19      // Cleared at compaction and at session start; the pinned poteto-mode is derived from poteto instead
20      loaded: PstackLoaded[]
21      // Whether the conversation holds the poteto-mode text; cleared at compaction and session start
22      pinned: boolean
23    }
24  }
25}
26