[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EVA6K2PZNV0F88CHMBGSWR":3,"skill-same:01M7EVA6K2PZNV0F88CHMBGSWR":56},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":14,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":45},"01M7EVA6K2PZNV0F88CHMBGSWR","pstack-help",null,"Guides users through pstack setup, the workflow skill, and picking the skill, playbook, or principle for a task. Type \u002Fpstack-help with a question.","shreshthkapai\u002Fagent-workflows","shreshthkapai","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F73650860?v=4","01M7EV5VN4GMWQ8R39WQ7KYFZQ","Shreshth","01M7EV5VNTDGPW78K42W4TYBAJ",0,"MIT","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fpstack-help\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: pstack-help\ndescription: Guides users through pstack setup, the workflow skill, and picking the skill, playbook, or principle for a task. Type \u002Fpstack-help with a question.\ndisable-model-invocation: true\n---\n\n# pstack help\n\nAnswer the user's question about pstack, hand them a prompt they can send, and link the file the answer came from. For a help question, don't start the work. The user asked how, and a pstack run spends real tokens, so let them send the prompt.\n\nA message that asks for work, such as \"use pstack to fix this bug\", is not a help question. Do the work under the [`workflow`](..\u002Fworkflow\u002FSKILL.md) skill. Load it first if the session does not already have it.\n\nThis file maps questions to the skills and guide pages that hold the answers. Those files own the details. Read the file you route to before you quote it, and trust it when it disagrees with this map. The skill links here point into the installed skills, so give the user the installed file's absolute path. The README and guide pages are not installed, so give the user their upstream public copy: `https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002F` followed by its path. The guide describes Cursor, so say which steps differ in this CLI.\n\n## Find out what they need\n\nInfer the need from the message and the conversation. A named situation, such as \"which skill reviews a PR?\", goes straight to its section. If the need is still unclear, ask one multiple-choice question with these options, then answer only the section they pick:\n\n- Get set up\n- Start a task\n- Pick a skill for a situation\n- Fix a run that went wrong\n- Make pstack my own\n\nCheck the state that changes the answer, and mention it only when it does:\n\n- A `~\u002F.config\u002Fagent-workflows\u002Fmodels.toml` that is missing or still the shipped default means `\u002Fsetup-pstack` hasn't run for this user, so every role uses its default model.\n- No `verify-*` skill or other app harness in the project means agents have no scripted way to drive the app. Mention `\u002Fcreate-verification-skill` when the question is about proving a change works.\n\nWhen the model registry is missing or still the default and it matters, ask whether the user wants to pick a model for each role and a reasoning budget now. It matters when the user is new, the question is about setup or cost, or the answer depends on which models run. Ask at most once per chat. If the need is also unclear, ask both questions together. Offer two choices:\n\n- Now: give them `\u002Fsetup-pstack` to type, and answer their question too.\n- Later: answer their question, and add one line saying every role keeps its default model until they run `\u002Fsetup-pstack`.\n\n## Get set up\n\n1. Install with this package's `install.py`. It installs into every CLI it finds among Claude Code, Codex, and Grok, and puts `aw` on the `PATH`.\n2. Run [`\u002Fsetup-pstack`](..\u002Fsetup-pstack\u002FSKILL.md). It asks for a reasoning budget, maps a model to each role, and writes the model registry. The registry applies from the next spawn.\n3. Start a real task with a goal and a check that can pass or fail. Type `\u002Fworkflow` (`$workflow` in Codex) before it, or run `aw skills always workflow` once so every session loads the workflow.\n\nInstalling changes nothing until a skill runs. Auto skills load on their own when a task matches, and manual ones wait for the user or a workflow step. The workflow is manual unless the user runs `aw skills always workflow`. `aw skills` lists which is which, and `aw skills auto|manual NAME` changes one. The [README](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002FREADME.md) and [guide page 1](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F01-setup.md) have the details. Offer to word their first prompt with them, per [`references\u002Fprompting.md`](references\u002Fprompting.md).\n\nIf cost is the worry, say where the tokens go and how to spend fewer. pstack spends extra tokens on subagents and review panels. Rerun `\u002Fsetup-pstack` and pick a smaller budget or cheaper models, or change one role with `aw models set ROLE model:effort`. `aw limits` shows how much of each plan is left. A role set to `inherit` runs on the session's model, which saves tokens when the session runs on a cheaper model. Each model's `provider` decides which subscription pays, so moving a busy role to the other provider spreads the load. A shorter panel list runs fewer subagents, one for each entry. Ask for a lighter pass when a task doesn't need the full playbook.\n\nThis is pstack ported from Cursor to Claude Code, Codex, and Grok. The workflow skill and skills such as `\u002Fhow`, `\u002Fwhy`, and `\u002Fteach`, spawn subagents with per-role models, natively in the current CLI or through `aw-run` on another one. Codex invokes each skill as `$name` instead of `\u002Fname`. Custom Modes are a Cursor feature, and `\u002Floop` exists in Claude Code and Grok but not in Codex.\n\n## Start a task\n\nThe workflow skill matches the task to a playbook, copies the playbook's steps into the todo list, and runs the other skills as the steps need them. A step it skips stays in the list as `skip: \u003Creason>`. A good prompt states the goal and how to tell it's done. It doesn't list skills, because a hand-written sequence tends to drop or reorder steps the playbook would keep. Read [`references\u002Fprompting.md`](references\u002Fprompting.md) before you help word one. [Guide page 2](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F02-poteto-mode.md) has examples.\n\n`aw skills always workflow` imports the workflow skill into the CLI's global instructions (`~\u002F.claude\u002FCLAUDE.md`, `~\u002F.codex\u002FAGENTS.md`, or `~\u002F.grok\u002FAGENTS.md`), so it applies to every session and does not fade. Without it, `\u002Fworkflow` at the start of a task loads it for that chat. Mid-chat, \"new task\" makes it match a fresh playbook. It already uses `delegate` for the subagents its playbook steps spawn. To get the same style from a subagent of your own, spawn the `delegate` agent.\n\n## Pick a skill\n\nThe default answer is to describe the task. The workflow skill runs most of the others when its steps need them. Name a skill directly when the user wants more or less of something than the playbook gives. Read the skill before you recommend it, and give one example prompt.\n\n| The user wants to | Skill |\n|---|---|\n| Do any non-trivial task with rigor | Type [`\u002Fworkflow`](..\u002Fworkflow\u002FSKILL.md) and describe it, or just describe it when the workflow is always on |\n| Check that the agent understood their goal and the problem before it starts | [`\u002Fin-your-own-words`](..\u002Fin-your-own-words\u002FSKILL.md) |\n| Aim a task higher before starting: more expert, more outcome-focused, not overspecified | [`\u002Fsharpen`](..\u002Fsharpen\u002FSKILL.md) |\n| Know how code works now, or where new code should live | [`\u002Fhow`](..\u002Fhow\u002FSKILL.md) |\n| Know why code is shaped this way, or where a number came from | [`\u002Fwhy`](..\u002Fwhy\u002FSKILL.md) |\n| Understand a change or subsystem, explained plainly | [`\u002Fteach`](..\u002Fteach\u002FSKILL.md) |\n| Catch up on their own recent work on a topic | [`\u002Frecall`](..\u002Frecall\u002FSKILL.md) |\n| Catch up on what changed in this chat since their last message, in a few plain sentences and small visuals, with enough context to decide each action item | [`\u002Fketchup`](..\u002Fketchup\u002FSKILL.md) |\n| Know what a small diff could break outside itself | [`\u002Fblast-radius`](..\u002Fblast-radius\u002FSKILL.md) |\n| Settle types and module shape before code that crosses a function boundary | [`\u002Farchitect`](..\u002Farchitect\u002FSKILL.md) |\n| Get several attempts at one brief, merged into the best one | [`\u002Farena`](..\u002Farena\u002FSKILL.md) |\n| Run parallel checks over slices, or race workers, in separate worktrees | [`\u002Fswarm`](..\u002Fswarm\u002FSKILL.md) |\n| Have different models review a diff and try to break it | [`\u002Finterrogate`](..\u002Finterrogate\u002FSKILL.md) |\n| Fix a bug test-first when a cheap local test exists | [`\u002Ftdd`](..\u002Ftdd\u002FSKILL.md) |\n| Apply TypeScript rules to `.ts` or `.tsx` work | [`\u002Ftypescript-best-practices`](..\u002Ftypescript-best-practices\u002FSKILL.md) |\n| Strip comments before review, using a reviewer that didn't write them | [`\u002Fno-comments`](..\u002Fno-comments\u002FSKILL.md) |\n| Clean AI tells out of prose | [`\u002Funslop`](..\u002Funslop\u002FSKILL.md) |\n| Write docs, an RFC, a README, a PR description, or a commit message to a standard | [`\u002Ftechnical-writing`](..\u002Ftechnical-writing\u002FSKILL.md) |\n| Hear the last reply again in plain words | [`\u002Fbro`](..\u002Fbro\u002FSKILL.md) |\n| Give agents a scripted way to drive the app and prove behavior | [`\u002Fcreate-verification-skill`](..\u002Fcreate-verification-skill\u002FSKILL.md) |\n| Bring a verification skill and its feature map back in line with the app | [`\u002Fmaintain-verification-skill`](..\u002Fmaintain-verification-skill\u002FSKILL.md) |\n| Vet a performance number before reporting or acting on it | [`\u002Fbenchmark-checklist`](..\u002Fbenchmark-checklist\u002FSKILL.md) |\n| Run a large or cross-cutting change, or one to review after stepping away | [`\u002Ffigure-it-out`](..\u002Ffigure-it-out\u002FSKILL.md) |\n| Keep a decision log during a run, and review it afterward | [`\u002Fshow-me-your-work`](..\u002Fshow-me-your-work\u002FSKILL.md) |\n| Pick a model for each role and a reasoning budget | [`\u002Fsetup-pstack`](..\u002Fsetup-pstack\u002FSKILL.md) |\n| Turn their own working habits into a personal mode skill | [`\u002Fautomate-me`](..\u002Fautomate-me\u002FSKILL.md) |\n| Turn what a finished task taught into skill edits | [`\u002Freflect`](..\u002Freflect\u002FSKILL.md) |\n| Stop agents from repeating the same mistakes in this repo | [`\u002Fcorrect`](..\u002Fcorrect\u002FSKILL.md) |\n| Find their way around pstack | `\u002Fpstack-help` |\n\nIf a skill directory next to this one is missing from the table, read its frontmatter and route by its description. The `principle-*` directories are covered under principles below.\n\nClose calls:\n\n- `\u002Fhow` explains what the code does. `\u002Fwhy` explains the reasons. `\u002Fteach` runs one or both and explains the result plainly.\n- `\u002Farena` gives every worker the same brief and merges the best parts. `\u002Fswarm` splits work into slices or a race and returns one report.\n- `\u002Farchitect` implements right after it settles the design. Add \"with checkpoint\" to review the design before it writes code.\n- `\u002Finterrogate` reviews the diff. `\u002Fblast-radius` looks for breakage outside the diff and proves the one fact that makes the change safe.\n- `\u002Frecall` rebuilds context across recent chats. `\u002Fketchup` covers only this chat since the user's last message. Resuming one specific chat or branch is the Session pickup playbook.\n- `\u002Ffigure-it-out` designs one rigorous run. The Orchestrate playbook runs a program that spans days and many PRs. The Autonomous run playbook drives one task to a finish condition.\n\nNot in pstack:\n\n- `\u002Fdeslop`, `control-cli`, and `control-ui` ship in the `cursor-team-kit` plugin.\n- `\u002Floop` is built into Claude Code and Grok. Codex has none. `skill-creator` comes with the CLI.\n- pstack has no `\u002Forchestrate` skill. Orchestrate is a workflow playbook. If the slash menu shows `\u002Forchestrate`, another plugin provides it.\n\n## Playbooks and principles\n\nPlaybooks are step lists inside the workflow skill, not skills, so they have no slash command. Describing the task picks one, and these phrases name one directly:\n\n- \"babysit this pr\" or \"check on pr 123\" runs Babysit. It drives the PR to merge-ready and stops there. It doesn't merge unless the user asks to merge, land, or ship.\n- \"land the stack\" runs Shipping.\n- \"take over this branch\" runs Session pickup.\n- \"pause safely\" runs Pause safely.\n- \"full autopilot on this queue\" runs Autopilot-full. \"stack them, don't ship\" runs Autopilot-stack.\n- \"run the eval playbook\" runs Eval.\n\nThe Playbooks section of [`workflow`](..\u002Fworkflow\u002FSKILL.md) lists every playbook and when it applies. [Guide page 6](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F06-verify-and-ship.md) covers opening, babysitting, and landing a PR.\n\npstack has no planning skill. Claude Code's plan mode works alongside it. For work that spans phases or stacked PRs, asking for a plan runs the [Multi-phase plan playbook](..\u002Fworkflow\u002Fplaybooks\u002Fmulti-phase-plan.md), which writes the plan and doesn't implement it. For a design question, the Prototype playbook or `\u002Farchitect` settles it in code first.\n\nPrinciples are one-rule skills that the workflow skill reads and cites in its replies. The user rarely invokes one. They steer with the names instead, as in \"apply prove it works. show me the real output.\" Typing `\u002Fprinciple-\u003Cname>` still loads one on demand. [Guide page 8](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F08-principles.md) lists them.\n\n## Fix a run that went wrong\n\n| Symptom | Fix |\n|---|---|\n| The workflow stopped applying | Check that the install's block is still in `~\u002F.claude\u002FCLAUDE.md`, `~\u002F.codex\u002FAGENTS.md`, or `~\u002F.grok\u002FAGENTS.md`. |\n| A question got treated as the next step of the last task | Say \"new task\", or say the turn doesn't need a playbook. |\n| A new model choice had no effect | Check the role with `aw-run --resolve \u003Crole>`. A role set to `inherit` follows the session's model. |\n| Runs cost more than expected | See the cost paragraph under Get set up. |\n| A skill didn't load on its own | Only `\u002Fsetup-pstack` loads from the user's words. The others load when the user types them or when the workflow skill runs them, and it doesn't run every skill. |\n| Parallel agents overwrote each other | Give each agent its own worktree. |\n| An overnight run moved but finished nothing | `\u002Floop` (Claude Code) needs a check that can pass or fail, not a duration. See [guide page 7](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F07-overnight.md). |\n| The reply claims success from a green build | Ask for the real command, flow, stored value, or profile. That's the prove-it-works principle. |\n\nFor a run that drifts, [`references\u002Fprompting.md`](references\u002Fprompting.md) has one-line steers. [Guide page 10](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F10-recipes-and-pitfalls.md) has more pitfalls and the recipes worth copying.\n\n## Make pstack my own\n\n- [`\u002Fautomate-me`](..\u002Fautomate-me\u002FSKILL.md) drafts a personal mode skill from the user's own history, to use alongside the workflow skill.\n- [`\u002Freflect`](..\u002Freflect\u002FSKILL.md) after a session turns its lessons into skill edits the user approves.\n- `write a skill for \u003Cworkflow>` runs the authoring playbook. The eval playbook tests a skill change blind.\n- Fix a misbehaving skill in its own PR, not inside the feature work where it went wrong.\n\n[Guide page 9](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F09-make-it-yours.md) covers each of these.\n\n## Reply\n\nLead with the answer. Give at most one example prompt in a code block, adapted from [`references\u002Frecipes.md`](references\u002Frecipes.md) when one fits, then the link to that file. Keep it short unless the user asked for the whole map.\n","# pstack help\n\nAnswer the user's question about pstack, hand them a prompt they can send, and link the file the answer came from. For a help question, don't start the work. The user asked how, and a pstack run spends real tokens, so let them send the prompt.\n\nA message that asks for work, such as \"use pstack to fix this bug\", is not a help question. Do the work under the [`workflow`](..\u002Fworkflow\u002FSKILL.md) skill. Load it first if the session does not already have it.\n\nThis file maps questions to the skills and guide pages that hold the answers. Those files own the details. Read the file you route to before you quote it, and trust it when it disagrees with this map. The skill links here point into the installed skills, so give the user the installed file's absolute path. The README and guide pages are not installed, so give the user their upstream public copy: `https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002F` followed by its path. The guide describes Cursor, so say which steps differ in this CLI.\n\n## Find out what they need\n\nInfer the need from the message and the conversation. A named situation, such as \"which skill reviews a PR?\", goes straight to its section. If the need is still unclear, ask one multiple-choice question with these options, then answer only the section they pick:\n\n- Get set up\n- Start a task\n- Pick a skill for a situation\n- Fix a run that went wrong\n- Make pstack my own\n\nCheck the state that changes the answer, and mention it only when it does:\n\n- A `~\u002F.config\u002Fagent-workflows\u002Fmodels.toml` that is missing or still the shipped default means `\u002Fsetup-pstack` hasn't run for this user, so every role uses its default model.\n- No `verify-*` skill or other app harness in the project means agents have no scripted way to drive the app. Mention `\u002Fcreate-verification-skill` when the question is about proving a change works.\n\nWhen the model registry is missing or still the default and it matters, ask whether the user wants to pick a model for each role and a reasoning budget now. It matters when the user is new, the question is about setup or cost, or the answer depends on which models run. Ask at most once per chat. If the need is also unclear, ask both questions together. Offer two choices:\n\n- Now: give them `\u002Fsetup-pstack` to type, and answer their question too.\n- Later: answer their question, and add one line saying every role keeps its default model until they run `\u002Fsetup-pstack`.\n\n## Get set up\n\n1. Install with this package's `install.py`. It installs into every CLI it finds among Claude Code, Codex, and Grok, and puts `aw` on the `PATH`.\n2. Run [`\u002Fsetup-pstack`](..\u002Fsetup-pstack\u002FSKILL.md). It asks for a reasoning budget, maps a model to each role, and writes the model registry. The registry applies from the next spawn.\n3. Start a real task with a goal and a check that can pass or fail. Type `\u002Fworkflow` (`$workflow` in Codex) before it, or run `aw skills always workflow` once so every session loads the workflow.\n\nInstalling changes nothing until a skill runs. Auto skills load on their own when a task matches, and manual ones wait for the user or a workflow step. The workflow is manual unless the user runs `aw skills always workflow`. `aw skills` lists which is which, and `aw skills auto|manual NAME` changes one. The [README](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002FREADME.md) and [guide page 1](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F01-setup.md) have the details. Offer to word their first prompt with them, per [`references\u002Fprompting.md`](references\u002Fprompting.md).\n\nIf cost is the worry, say where the tokens go and how to spend fewer. pstack spends extra tokens on subagents and review panels. Rerun `\u002Fsetup-pstack` and pick a smaller budget or cheaper models, or change one role with `aw models set ROLE model:effort`. `aw limits` shows how much of each plan is left. A role set to `inherit` runs on the session's model, which saves tokens when the session runs on a cheaper model. Each model's `provider` decides which subscription pays, so moving a busy role to the other provider spreads the load. A shorter panel list runs fewer subagents, one for each entry. Ask for a lighter pass when a task doesn't need the full playbook.\n\nThis is pstack ported from Cursor to Claude Code, Codex, and Grok. The workflow skill and skills such as `\u002Fhow`, `\u002Fwhy`, and `\u002Fteach`, spawn subagents with per-role models, natively in the current CLI or through `aw-run` on another one. Codex invokes each skill as `$name` instead of `\u002Fname`. Custom Modes are a Cursor feature, and `\u002Floop` exists in Claude Code and Grok but not in Codex.\n\n## Start a task\n\nThe workflow skill matches the task to a playbook, copies the playbook's steps into the todo list, and runs the other skills as the steps need them. A step it skips stays in the list as `skip: \u003Creason>`. A good prompt states the goal and how to tell it's done. It doesn't list skills, because a hand-written sequence tends to drop or reorder steps the playbook would keep. Read [`references\u002Fprompting.md`](references\u002Fprompting.md) before you help word one. [Guide page 2](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F02-poteto-mode.md) has examples.\n\n`aw skills always workflow` imports the workflow skill into the CLI's global instructions (`~\u002F.claude\u002FCLAUDE.md`, `~\u002F.codex\u002FAGENTS.md`, or `~\u002F.grok\u002FAGENTS.md`), so it applies to every session and does not fade. Without it, `\u002Fworkflow` at the start of a task loads it for that chat. Mid-chat, \"new task\" makes it match a fresh playbook. It already uses `delegate` for the subagents its playbook steps spawn. To get the same style from a subagent of your own, spawn the `delegate` agent.\n\n## Pick a skill\n\nThe default answer is to describe the task. The workflow skill runs most of the others when its steps need them. Name a skill directly when the user wants more or less of something than the playbook gives. Read the skill before you recommend it, and give one example prompt.\n\n| The user wants to | Skill |\n|---|---|\n| Do any non-trivial task with rigor | Type [`\u002Fworkflow`](..\u002Fworkflow\u002FSKILL.md) and describe it, or just describe it when the workflow is always on |\n| Check that the agent understood their goal and the problem before it starts | [`\u002Fin-your-own-words`](..\u002Fin-your-own-words\u002FSKILL.md) |\n| Aim a task higher before starting: more expert, more outcome-focused, not overspecified | [`\u002Fsharpen`](..\u002Fsharpen\u002FSKILL.md) |\n| Know how code works now, or where new code should live | [`\u002Fhow`](..\u002Fhow\u002FSKILL.md) |\n| Know why code is shaped this way, or where a number came from | [`\u002Fwhy`](..\u002Fwhy\u002FSKILL.md) |\n| Understand a change or subsystem, explained plainly | [`\u002Fteach`](..\u002Fteach\u002FSKILL.md) |\n| Catch up on their own recent work on a topic | [`\u002Frecall`](..\u002Frecall\u002FSKILL.md) |\n| Catch up on what changed in this chat since their last message, in a few plain sentences and small visuals, with enough context to decide each action item | [`\u002Fketchup`](..\u002Fketchup\u002FSKILL.md) |\n| Know what a small diff could break outside itself | [`\u002Fblast-radius`](..\u002Fblast-radius\u002FSKILL.md) |\n| Settle types and module shape before code that crosses a function boundary | [`\u002Farchitect`](..\u002Farchitect\u002FSKILL.md) |\n| Get several attempts at one brief, merged into the best one | [`\u002Farena`](..\u002Farena\u002FSKILL.md) |\n| Run parallel checks over slices, or race workers, in separate worktrees | [`\u002Fswarm`](..\u002Fswarm\u002FSKILL.md) |\n| Have different models review a diff and try to break it | [`\u002Finterrogate`](..\u002Finterrogate\u002FSKILL.md) |\n| Fix a bug test-first when a cheap local test exists | [`\u002Ftdd`](..\u002Ftdd\u002FSKILL.md) |\n| Apply TypeScript rules to `.ts` or `.tsx` work | [`\u002Ftypescript-best-practices`](..\u002Ftypescript-best-practices\u002FSKILL.md) |\n| Strip comments before review, using a reviewer that didn't write them | [`\u002Fno-comments`](..\u002Fno-comments\u002FSKILL.md) |\n| Clean AI tells out of prose | [`\u002Funslop`](..\u002Funslop\u002FSKILL.md) |\n| Write docs, an RFC, a README, a PR description, or a commit message to a standard | [`\u002Ftechnical-writing`](..\u002Ftechnical-writing\u002FSKILL.md) |\n| Hear the last reply again in plain words | [`\u002Fbro`](..\u002Fbro\u002FSKILL.md) |\n| Give agents a scripted way to drive the app and prove behavior | [`\u002Fcreate-verification-skill`](..\u002Fcreate-verification-skill\u002FSKILL.md) |\n| Bring a verification skill and its feature map back in line with the app | [`\u002Fmaintain-verification-skill`](..\u002Fmaintain-verification-skill\u002FSKILL.md) |\n| Vet a performance number before reporting or acting on it | [`\u002Fbenchmark-checklist`](..\u002Fbenchmark-checklist\u002FSKILL.md) |\n| Run a large or cross-cutting change, or one to review after stepping away | [`\u002Ffigure-it-out`](..\u002Ffigure-it-out\u002FSKILL.md) |\n| Keep a decision log during a run, and review it afterward | [`\u002Fshow-me-your-work`](..\u002Fshow-me-your-work\u002FSKILL.md) |\n| Pick a model for each role and a reasoning budget | [`\u002Fsetup-pstack`](..\u002Fsetup-pstack\u002FSKILL.md) |\n| Turn their own working habits into a personal mode skill | [`\u002Fautomate-me`](..\u002Fautomate-me\u002FSKILL.md) |\n| Turn what a finished task taught into skill edits | [`\u002Freflect`](..\u002Freflect\u002FSKILL.md) |\n| Stop agents from repeating the same mistakes in this repo | [`\u002Fcorrect`](..\u002Fcorrect\u002FSKILL.md) |\n| Find their way around pstack | `\u002Fpstack-help` |\n\nIf a skill directory next to this one is missing from the table, read its frontmatter and route by its description. The `principle-*` directories are covered under principles below.\n\nClose calls:\n\n- `\u002Fhow` explains what the code does. `\u002Fwhy` explains the reasons. `\u002Fteach` runs one or both and explains the result plainly.\n- `\u002Farena` gives every worker the same brief and merges the best parts. `\u002Fswarm` splits work into slices or a race and returns one report.\n- `\u002Farchitect` implements right after it settles the design. Add \"with checkpoint\" to review the design before it writes code.\n- `\u002Finterrogate` reviews the diff. `\u002Fblast-radius` looks for breakage outside the diff and proves the one fact that makes the change safe.\n- `\u002Frecall` rebuilds context across recent chats. `\u002Fketchup` covers only this chat since the user's last message. Resuming one specific chat or branch is the Session pickup playbook.\n- `\u002Ffigure-it-out` designs one rigorous run. The Orchestrate playbook runs a program that spans days and many PRs. The Autonomous run playbook drives one task to a finish condition.\n\nNot in pstack:\n\n- `\u002Fdeslop`, `control-cli`, and `control-ui` ship in the `cursor-team-kit` plugin.\n- `\u002Floop` is built into Claude Code and Grok. Codex has none. `skill-creator` comes with the CLI.\n- pstack has no `\u002Forchestrate` skill. Orchestrate is a workflow playbook. If the slash menu shows `\u002Forchestrate`, another plugin provides it.\n\n## Playbooks and principles\n\nPlaybooks are step lists inside the workflow skill, not skills, so they have no slash command. Describing the task picks one, and these phrases name one directly:\n\n- \"babysit this pr\" or \"check on pr 123\" runs Babysit. It drives the PR to merge-ready and stops there. It doesn't merge unless the user asks to merge, land, or ship.\n- \"land the stack\" runs Shipping.\n- \"take over this branch\" runs Session pickup.\n- \"pause safely\" runs Pause safely.\n- \"full autopilot on this queue\" runs Autopilot-full. \"stack them, don't ship\" runs Autopilot-stack.\n- \"run the eval playbook\" runs Eval.\n\nThe Playbooks section of [`workflow`](..\u002Fworkflow\u002FSKILL.md) lists every playbook and when it applies. [Guide page 6](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F06-verify-and-ship.md) covers opening, babysitting, and landing a PR.\n\npstack has no planning skill. Claude Code's plan mode works alongside it. For work that spans phases or stacked PRs, asking for a plan runs the [Multi-phase plan playbook](..\u002Fworkflow\u002Fplaybooks\u002Fmulti-phase-plan.md), which writes the plan and doesn't implement it. For a design question, the Prototype playbook or `\u002Farchitect` settles it in code first.\n\nPrinciples are one-rule skills that the workflow skill reads and cites in its replies. The user rarely invokes one. They steer with the names instead, as in \"apply prove it works. show me the real output.\" Typing `\u002Fprinciple-\u003Cname>` still loads one on demand. [Guide page 8](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F08-principles.md) lists them.\n\n## Fix a run that went wrong\n\n| Symptom | Fix |\n|---|---|\n| The workflow stopped applying | Check that the install's block is still in `~\u002F.claude\u002FCLAUDE.md`, `~\u002F.codex\u002FAGENTS.md`, or `~\u002F.grok\u002FAGENTS.md`. |\n| A question got treated as the next step of the last task | Say \"new task\", or say the turn doesn't need a playbook. |\n| A new model choice had no effect | Check the role with `aw-run --resolve \u003Crole>`. A role set to `inherit` follows the session's model. |\n| Runs cost more than expected | See the cost paragraph under Get set up. |\n| A skill didn't load on its own | Only `\u002Fsetup-pstack` loads from the user's words. The others load when the user types them or when the workflow skill runs them, and it doesn't run every skill. |\n| Parallel agents overwrote each other | Give each agent its own worktree. |\n| An overnight run moved but finished nothing | `\u002Floop` (Claude Code) needs a check that can pass or fail, not a duration. See [guide page 7](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F07-overnight.md). |\n| The reply claims success from a green build | Ask for the real command, flow, stored value, or profile. That's the prove-it-works principle. |\n\nFor a run that drifts, [`references\u002Fprompting.md`](references\u002Fprompting.md) has one-line steers. [Guide page 10](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F10-recipes-and-pitfalls.md) has more pitfalls and the recipes worth copying.\n\n## Make pstack my own\n\n- [`\u002Fautomate-me`](..\u002Fautomate-me\u002FSKILL.md) drafts a personal mode skill from the user's own history, to use alongside the workflow skill.\n- [`\u002Freflect`](..\u002Freflect\u002FSKILL.md) after a session turns its lessons into skill edits the user approves.\n- `write a skill for \u003Cworkflow>` runs the authoring playbook. The eval playbook tests a skill change blind.\n- Fix a misbehaving skill in its own PR, not inside the feature work where it went wrong.\n\n[Guide page 9](https:\u002F\u002Fgithub.com\u002Fcursor\u002Fplugins\u002Fblob\u002Fmain\u002Fpstack\u002Fdocs\u002Fguide\u002F09-make-it-yours.md) covers each of these.\n\n## Reply\n\nLead with the answer. Give at most one example prompt in a code block, adapted from [`references\u002Frecipes.md`](references\u002Frecipes.md) when one fits, then the link to that file. Keep it short unless the user asked for the whole map.\n",{"name":5,"description":7,"disable-model-invocation":24},true,"main","cafd398384400c1a0ce1da059a134394f28a690e","skills\u002Fpstack-help","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows","https:\u002F\u002Fgithub.com\u002Fshreshthkapai","https:\u002F\u002Fshreshth-personal-website.vercel.app\u002F","Lost....in neural networks","2026-10-30 08:24:11",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":24},[35,38,41],{"tool":36,"command":37},"npx","npx skills add shreshthkapai\u002Fagent-workflows --skill pstack-help",{"tool":39,"command":40},"bunx","bunx skills add shreshthkapai\u002Fagent-workflows --skill pstack-help",{"tool":42,"command":43},"pnpm","pnpm dlx skills add shreshthkapai\u002Fagent-workflows --skill pstack-help","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fpstack-help\u002FSKILL.md",[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fprompting.md","REFERENCE","参考文档",3166,"https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fpstack-help\u002Freferences\u002Fprompting.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Frecipes.md",3082,"https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fpstack-help\u002Freferences\u002Frecipes.md",[57,64,71,78,85,92],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M7EVBMVS4ZXTSPK19XNMYNT5","workflow","Concise, detailed responses, deliberate subagents, unslopped prose, simple code, and verified work, with each task routed to a playbook. Use for \u002Fworkflow, or for requests to work in this style. `aw skills always workflow` loads it into every session.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fworkflow\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M7EVBEEEVNYKRCGBH6V5SSFA","why","Use for 'why does X work this way', 'why we picked Y', design rationale, regressions, postmortems, or data-backed thresholds. Discovers available MCPs and queries each evidence category (source control, issue tracker, long-form docs, real-time chat, infrastructure observability, error tracking, product analytics warehouse) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fwhy\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M7EVBBK5G5QKHXWVYHTRKJ0B","unslop","Cut AI tells from any writing.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Funslop\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M7EVB9ED5F0RVKN70D2ZKV9W","typescript-best-practices","TypeScript best practices. Use when reading or editing any .ts or .tsx file.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Ftypescript-best-practices\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M7EVB568STK9X05C00C83KBD","technical-writing","Layered technical-writing standard: Diátaxis structure, Google developer style sentences, STE instruction rules, Global English syntax. Use for \u002Ftechnical-writing or when writing or reviewing docs, RFCs, readmes, PR descriptions, or commit messages.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Ftechnical-writing\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":19},"01M7EVB0H4V10C6GQHXX4599Y2","teach","Explain a body of work plainly so a person actually understands it. Runs the `how` and `why` skills and weaves what they find into one clear explanation. Use for 'teach me this', 'help me really understand X', 'explain this change or subsystem to me'.","https:\u002F\u002Fgithub.com\u002Fshreshthkapai\u002Fagent-workflows\u002Fblob\u002Fmain\u002Fskills\u002Fteach\u002FSKILL.md",[],[]]