pstack is poteto's set of engineering skills for coding agents, installed with /add-plugin pstack. This page is an unofficial reading aid.
@@ -329,14 +357,26 @@
detail:"Shape the trigger, the steps, and the reply contract. Encode lessons in structure rather than more text."},
{id:"eval", n:"Eval", t:"Testing how a skill, structure, or prompt change affects agent behavior before promoting it.",
detail:"Blinded comparison against a baseline. Measure behavior change before you promote the new version."},
+ {id:"babysit", n:"Babysit", t:"Driving a PR or stack to merge-ready: conflicts, review threads, and CI.",
+ detail:"Own the merge frontier one PR at a time. Clear conflicts and review threads before CI, then stop at a merge-ready verdict."},
+ {id:"shipping", n:"Shipping", t:"Landing a green stack after independent verification.",
+ detail:"Verify every PR from the root, land only the contiguous verified run, then let Graphite drain it without touching the queue."},
{id:"autonomous-run", n:"Autonomous run", t:"A long task to drive to completion without stopping. Run until done, /loop until X, going to bed.",
detail:"Keep a decision trail via show-me-your-work so the human can audit it when they return. Commit the trail when stakes need a record."},
+ {id:"orchestrate", n:"Orchestrate", t:"A standing project for many stacked PRs and agents, run by one coordinator over multiple days.",
+ detail:"Run a program, not one task. Frame the finish predicate, pilot the unit, scale workers, drain the queue, and keep the frontier green."},
+ {id:"autopilot-full", n:"Autopilot-full", t:"A queue of independent PRs run to merged with one owner per PR.",
+ detail:"Owners build through merge. The root swarm-verifies every merge-ready head before the owner merges."},
+ {id:"autopilot-stack", n:"Autopilot-stack", t:"A queue built and verified as one linear reviewed stack for the operator to land.",
+ detail:"Owners build and prove each PR, while the root owns topology. Deliver one verified Graphite chain and never auto-ship it."},
{id:"session-pickup", n:"Session pickup", t:"Resuming or taking over a prior agent's in-flight work from a transcript, cloud-agent URL, or pushed branch.",
detail:"Reconstruct state from the record. Fire a fresh subagent with consolidated scope rather than trusting a done summary."},
{id:"pause-safely", n:"Pause safely", t:"Suspending in-flight work cleanly so it can be resumed. On an explicit pause, going offline, a restart, or imminent compaction.",
detail:"The complement to Session pickup. Leave the work in a state another agent can pick up without guessing."},
{id:"multi-phase-plan", n:"Multi-phase plan", t:"Work that spans phases or stacked PRs.",
detail:"Break the work into verifiable units. Order delivery so the sequence proves itself, each unit green before the next."},
+ {id:"worktree-cleanup", n:"Worktree and simulator cleanup", t:"Reclaiming local disk by pruning safe worktrees and stale iOS simulators.",
+ detail:"Audit usage, uncommitted work, and pinned chats before deleting. Recheck disk space and report what stayed behind."},
{id:"figure-it-out", n:"Figure it out", t:"No bundled playbook fits. Design a bespoke, rigorous playbook for the task.",
detail:"Used for large or cross-cutting efforts like a migration across many call sites, or work the user steps away from to trust later."},
{id:"opening-a-pr", n:"Opening a PR", t:"Invoked at the end of every other playbook.", meta:true,
@@ -344,7 +384,8 @@
];
const SKILLS = [
- {cmd:"/poteto-mode", when:"Default entry point for any non-trivial task. Reads your request, picks a playbook, and routes the other skills as the steps fire."},
+ {cmd:"/poteto-mode", when:"Default entry point for any non-trivial task. Routes through poteto-agent, picks a playbook, and calls the other skills as the steps fire."},
+ {cmd:"/bro", when:"The last message is dense or full of jargon and you want it restated plainly."},
{cmd:"/how", when:"You want a walkthrough of how a subsystem works."},
{cmd:"/why", when:"You want to know why something was built this way. Discovers MCPs at runtime and queries each evidence category in parallel."},
{cmd:"/recall", when:"You are starting or resuming work and want your recent context on a topic rebuilt from your chat history and the shared record."},
@@ -359,7 +400,13 @@
{cmd:"/typescript-best-practices", when:"You are reading or editing TypeScript. Grounds the type-system-discipline principle in syntax."},
{cmd:"/figure-it-out", when:"No bundled playbook fits. Designs a rigorous, auditable playbook for the task."},
{cmd:"/show-me-your-work", when:"You want a reviewable decision trail. Logs decisions to a TSV you can commit."},
- {cmd:"/unslop", when:"You are cleaning up writing. Removes AI tells."}
+ {cmd:"/unslop", when:"You are cleaning up writing. Removes AI tells."},
+ {cmd:"/teach", when:"Explain a body of work plainly so a person actually understands it. Runs `how` and `why` and weaves what they find into one clear explanation."},
+ {cmd:"/create-verification-skill", when:"Generate a project-local verification skill that drives your app the way a user does — any language, framework, or platform."},
+ {cmd:"/maintain-verification-skill", when:"Periodic pass that keeps a project's verification skill and feature map honest."},
+ {cmd:"/swarm", when:"You want N parallel workers over slices or races, drained into one aggregated report."},
+ {cmd:"/no-comments", when:"You want Comment Sicko to hunt comments, then fix accepted findings at the root cause."},
+ {cmd:"/technical-writing", when:"You are writing docs and need Diátaxis, Google style, STE, and Global English applied together."}
];
const GROUPS = [
@@ -386,6 +433,7 @@
{g:"arch", n:"Make Operations Idempotent", applies:"Designing commands, lifecycle steps, or loops that run amid crashes and retries.", rule:"Converge to the same end state regardless of partial prior runs."},
{g:"arch", n:"Migrate Callers Then Delete Legacy APIs", applies:"Introducing a new internal API while old callers exist.", rule:"Migrate callers and remove the old API in the same wave. No compatibility layers."},
{g:"arch", n:"Separate Before Serializing Shared State", applies:"Concurrent actors might write the same file, branch, key, or object.", rule:"Eliminate the sharing first. When sharing is real, enforce serialization structurally."},
+ {g:"arch", n:"Model the Domain", applies:"When writing stateful logic, or when code branches a lot or repeats a shape assumption across files.", rule:"Encode the real domain in a data structure instead of scattering it across conditionals. Reach for state machines, typed models, lookup tables, reducers, or small module boundaries that gather repeated behavior. Don't force an abstraction if the current shape is already clear and local."},
{g:"verif", n:"Prove It Works", applies:"After a task, before declaring done.", rule:"Verify against the real artifact. Not a proxy. Not that it compiles."},
{g:"verif", n:"Fix Root Causes", applies:"Debugging.", rule:"Trace each symptom to its root cause. Reproduce first. Ask why until you reach it."},
@@ -403,6 +451,7 @@
{r:"perf-issue", k:"single", o:"Trace measured slowness against a baseline."},
{r:"hillclimb", k:"single", o:"Loop hypotheses against a target metric."},
{r:"judgment and prose", k:"single", o:"Where judgment and writing quality matter most."},
+ {r:"hardest tasks", k:"single", o:"The default model for the hardest tasks."},
{r:"how explorer", k:"single", o:"Maps the subsystem inside the /how skill."},
{r:"how explainer", k:"single", o:"Writes the /how walkthrough."},
{r:"how critics", k:"panel", o:"One subagent per model reviews the explanation."},
@@ -411,6 +460,8 @@
{r:"reflect tooling", k:"single", o:"Captures the tooling recipe in /reflect."},
{r:"reflect judgment, divergent, synthesizer", k:"single", o:"Captures the judgment recipe in /reflect."},
{r:"arena runners", k:"panel", o:"One subagent per model produces a competing attempt."},
+ {r:"arena cross-judge pool", k:"panel", o:"Arena picks a judge from a different model family than the parent when possible."},
+ {r:"swarm workers", k:"single", o:"The default model for every swarm worker."},
{r:"architect runners", k:"panel", o:"One subagent per model explores a design."},
{r:"interrogate reviewers", k:"panel", o:"One subagent per model tries to break the diff."}
];
@@ -428,10 +479,16 @@
"visual-parity":["visual parity","pixel","match the design","spacing","styling system","row spacing","matches","pixel-exact"],
"authoring-a-skill":["write a skill","edit a skill","skill.md","authoring a skill","new skill","modify a skill"],
eval:["eval","evaluate the skill","affects agent behavior","blinded","promote the skill","skill change","prompt change"],
+ babysit:["check on pr","get it green","address the bugbot comments","babysit this","merge-ready"],
"autonomous-run":["going to bed","run until","overnight","autonomous","don't stop","loop until","many hours","i'm away","while i sleep"],
+ shipping:["land the stack","ship the stack","ship it","enable merge when ready"],
+ orchestrate:["run this whole project","own this migration until it lands","standing project","many stacked prs"],
+ "autopilot-full":["autopilot this queue","full autopilot","one-owner-per-pr"],
+ "autopilot-stack":["autopilot-stack","stack them, don't ship","build the stack, i'll land it"],
"session-pickup":["resume","take over","prior agent","in-flight","in flight","transcript","pickup","cloud-agent","pick up where"],
"pause-safely":["pause","suspend","going offline","restart","cleanly so it can be resumed","step away","context compaction"],
"multi-phase-plan":["multi-phase","stacked pr","spans phases","migration across","many call sites","multi-pr","across the codebase"],
+ "worktree-cleanup":["clean up worktrees","what's using my disk","free up space","delete old simulators","prune safe-to-prune worktrees"],
"figure-it-out":["no playbook","bespoke","cross-cutting","ambitious","step away","trust it when","trust it was done","when i'm back","migrate every caller","migrate all callers","keeping behavior identical","figure it out","unprecedented"]
};
diff --git a/llms-full.txt b/llms-full.txt
index a158eb3..6af8897 100644
--- a/llms-full.txt
+++ b/llms-full.txt
@@ -6,7 +6,7 @@ This site is an unofficial reading aid at https://hustlecoding.github.io/pstack-
## Overview
Type `/poteto-mode` and describe a task. pstack matches the task to a playbook, then runs the appropriate skills as the steps fire.
-Twenty principles apply across every playbook, biasing the work toward small diffs and real verification.
+Twenty-one principles apply across every playbook, biasing the work toward small diffs and real verification. The [official guide](https://github.com/cursor/plugins/blob/main/pstack/docs/guide/README.md) covers setup, routing, design, building, verification, overnight work, principles, customization, and recipes.
## Playbooks
@@ -60,10 +60,30 @@ Detail: Shape the trigger, the steps, and the reply contract. Encode lessons in
Trigger: Testing how a skill, structure, or prompt change affects agent behavior before promoting it.
Detail: Blinded comparison against a baseline. Measure behavior change before you promote the new version.
+### Babysit
+Trigger: Driving a PR or stack to merge-ready: conflicts, review threads, and CI.
+Detail: Own the merge frontier one PR at a time. Clear conflicts and review threads before CI, then stop at a merge-ready verdict.
+
+### Shipping
+Trigger: Landing a green stack after independent verification.
+Detail: Verify every PR from the root, land only the contiguous verified run, then let Graphite drain it without touching the queue.
+
### Autonomous run
Trigger: A long task to drive to completion without stopping. Run until done, /loop until X, going to bed.
Detail: Keep a decision trail via show-me-your-work so the human can audit it when they return. Commit the trail when stakes need a record.
+### Orchestrate
+Trigger: A standing project for many stacked PRs and agents, run by one coordinator over multiple days.
+Detail: Run a program, not one task. Frame the finish predicate, pilot the unit, scale workers, drain the queue, and keep the frontier green.
+
+### Autopilot-full
+Trigger: A queue of independent PRs run to merged with one owner per PR.
+Detail: Owners build through merge. The root swarm-verifies every merge-ready head before the owner merges.
+
+### Autopilot-stack
+Trigger: A queue built and verified as one linear reviewed stack for the operator to land.
+Detail: Owners build and prove each PR, while the root owns topology. Deliver one verified Graphite chain and never auto-ship it.
+
### Session pickup
Trigger: Resuming or taking over a prior agent's in-flight work from a transcript, cloud-agent URL, or pushed branch.
Detail: Reconstruct state from the record. Fire a fresh subagent with consolidated scope rather than trusting a done summary.
@@ -76,6 +96,10 @@ Detail: The complement to Session pickup. Leave the work in a state another agen
Trigger: Work that spans phases or stacked PRs.
Detail: Break the work into verifiable units. Order delivery so the sequence proves itself, each unit green before the next.
+### Worktree and simulator cleanup
+Trigger: Reclaiming local disk by pruning safe worktrees and stale iOS simulators.
+Detail: Audit usage, uncommitted work, and pinned chats before deleting. Recheck disk space and report what stayed behind.
+
### Figure it out
Trigger: No bundled playbook fits. Design a bespoke, rigorous playbook for the task.
Detail: Used for large or cross-cutting efforts like a migration across many call sites, or work the user steps away from to trust later.
@@ -89,7 +113,10 @@ Detail: Not a task shape of its own. It fires once the matched playbook's work i
Skills are the individual moves. Reach for one directly when you want a specific result instead of the full route.
### /poteto-mode
-When to use: Default entry point for any non-trivial task. Reads your request, picks a playbook, and routes the other skills as the steps fire.
+When to use: Default entry point for any non-trivial task. Routes through poteto-agent, picks a playbook, and calls the other skills as the steps fire.
+
+### /bro
+When to use: The last message is dense or full of jargon and you want it restated plainly.
### /how
When to use: You want a walkthrough of how a subsystem works.
@@ -136,9 +163,27 @@ When to use: You want a reviewable decision trail. Logs decisions to a TSV you c
### /unslop
When to use: You are cleaning up writing. Removes AI tells.
+### /teach
+When to use: Explain a body of work plainly so a person actually understands it. Runs `how` and `why` and weaves what they find into one clear explanation.
+
+### /create-verification-skill
+When to use: Generate a project-local verification skill that drives your app the way a user does — any language, framework, or platform.
+
+### /maintain-verification-skill
+When to use: Periodic pass that keeps a project's verification skill and feature map honest.
+
+### /swarm
+When to use: You want N parallel workers over slices or races, drained into one aggregated report.
+
+### /no-comments
+When to use: You want Comment Sicko to hunt comments, then fix accepted findings at the root cause.
+
+### /technical-writing
+When to use: You are writing docs and need Diátaxis, Google style, STE, and Global English applied together.
+
## Principles
-Twenty rules in five families. Each principle has a name, when it applies, and the rule to follow.
+Twenty-one rules in five families. Each principle has a name, when it applies, and the rule to follow.
### Core
How to sequence work and scope it. The bias to delete before you add.
@@ -202,6 +247,10 @@ Boundaries, types, and shared state across concurrent actors.
- Applies when: Concurrent actors might write the same file, branch, key, or object.
- Rule: Eliminate the sharing first. When sharing is real, enforce serialization structurally.
+- **Model the Domain**
+ - Applies when: When writing stateful logic, or when code branches a lot or repeats a shape assumption across files.
+ - Rule: Encode the real domain in a data structure instead of scattering it across conditionals. Reach for state machines, typed models, lookup tables, reducers, or small module boundaries that gather repeated behavior. Don't force an abstraction if the current shape is already clear and local.
+
### Verification
Proving it works against the real artifact, and ordering the work so each step can be checked.
@@ -245,6 +294,7 @@ Run `/setup-pstack` to detect your models and write the rule. Skills read it and
- **perf-issue** (single) — Trace measured slowness against a baseline.
- **hillclimb** (single) — Loop hypotheses against a target metric.
- **judgment and prose** (single) — Where judgment and writing quality matter most.
+- **hardest tasks** (single) — The default model for the hardest tasks.
- **how explorer** (single) — Maps the subsystem inside the /how skill.
- **how explainer** (single) — Writes the /how walkthrough.
- **how critics** (panel) — One subagent per model reviews the explanation.
@@ -253,31 +303,33 @@ Run `/setup-pstack` to detect your models and write the rule. Skills read it and
- **reflect tooling** (single) — Captures the tooling recipe in /reflect.
- **reflect judgment, divergent, synthesizer** (single) — Captures the judgment recipe in /reflect.
- **arena runners** (panel) — One subagent per model produces a competing attempt.
+- **arena cross-judge pool** (panel) — Arena picks a judge from a different model family than the parent when possible.
+- **swarm workers** (single) — The default model for every swarm worker.
- **architect runners** (panel) — One subagent per model explores a design.
- **interrogate reviewers** (panel) — One subagent per model tries to break the diff.
## Recent upstream changelog
Source: https://github.com/cursor/plugins/tree/main/pstack
-Last synced: 2026-07-13T11:10:51+00:00
-
-- [2026-07-12] pstack: bump version to 0.11.2 (#154) (a29f5a8) by poteto
-- [2026-07-12] pstack: add the teach skill (compose how + why into one explanation) (#153) (8f008c4) by poteto
-- [2026-07-12] pstack: bump version to 0.11.1 (#152) (20bdb6c) by poteto
-- [2026-07-12] maintain-verification-skill: cleanup granularity, re-doctor, evidence checks (#151) (6714489) by poteto
-- [2026-07-12] pstack: add create-verification-skill and maintain-verification-skill (#150) (e42d29f) by poteto
-- [2026-07-11] pstack: give poteto-mode a human display name (#149) (9d2a3f2) by poteto
-- [2026-07-11] pstack: lead the README with the two-step quickstart (#148) (9251b26) by poteto
-- [2026-07-11] pstack: add model-the-domain principle; true up README (#147) (a814542) by poteto
-- [2026-07-09] pstack: make poteto-mode a sticky mode with a conditional reminder (#144) (0dda29e) by poteto
-- [2026-07-08] pstack: route hardest tasks to claude-fable-5-thinking-max (#143) (9b80b53) by poteto
-- [2026-07-08] pstack: route composer slots to grok-4.5-fast-xhigh (#142) (dc2fae6) by poteto
-- [2026-06-24] pstack: add Benny issue automation pack (#137) (0452e08) by poteto
-- [2026-06-17] pstack: add recall and blast-radius skills (#135) (e46364b) by poteto
-- [2026-06-14] pstack: bump to 0.9.1, list the hillclimb playbook in the README (#133) (cfd81b3) by poteto
-- [2026-06-14] poteto-mode: add the Hillclimb playbook and tighten Autonomous run stop semantics (#132) (b64f02a) by poteto
-- [2026-06-07] pstack: sync README with shipped principles and playbooks (#126) (74dd229) by poteto
-- [2026-06-07] pstack: bump version to 0.9.0 (#125) (a5cda8b) by poteto
-- [2026-06-06] poteto-mode: shape refactoring commits before opening the PR (#124) (9170757) by poteto
-- [2026-06-06] pstack: self-unblock empirical forks with a sketch instead of asking (#122) (318e8f7) by poteto
-- [2026-06-06] pstack: add 'sequence work into verifiable units' principle (#123) (27daaa3) by poteto
+Last synced: 2026-08-17T09:34:56+00:00
+
+- [2026-08-13] chore(pstack): bump Grok default from 4.5 to 4.6 (#210) (63d938c) by poteto
+- [2026-08-02] docs(pstack): bring guide current with new skills and playbooks (#188) (424829e) by poteto
+- [2026-08-02] pstack: add bro, babysit/shipping/orchestrate/worktree-cleanup, and catch-up ports (0.14.0) (#187) (99559f2) by poteto
+- [2026-08-02] pstack: add autopilot playbooks, /no-comments, Comment Sicko, and /technical-writing (0.13.0) (#185) (b047069) by poteto
+- [2026-07-30] Address swarm review feedback (91dd7b7) by cursoragent
+- [2026-07-30] docs(pstack): refresh guide for 0.12.0 (0b7ef5b) by cursoragent
+- [2026-07-30] Add swarm skill to pstack (b79f8ca) by cursoragent
+- [2026-07-30] Add feature map reference to create-verification-skill (#178) (4483dcd) by poteto
+- [2026-07-29] feat(pstack): deepen architect interface guidance (#175) (45c66fd) by poteto
+- [2026-07-27] feat(pstack): teach constructive type modeling (#174) (91be0f9) by poteto
+- [2026-07-26] Clarify ownership of autonomous run discoveries (#170) (ba7b590) by poteto
+- [2026-07-26] pstack: add Opus 5 to model panels (#169) (d45ad02) by poteto
+- [2026-07-23] pstack: cut model-config prose to its definition sites (#167) (04166ac) by poteto
+- [2026-07-22] pstack: add public usage tutorial (#164) (02c03a9) by poteto
+- [2026-07-22] pstack: default explorer roles to grok (#166) (03e087a) by poteto
+- [2026-07-22] pstack: default panels to fable sol grok (#165) (e1007b1) by poteto
+- [2026-07-22] pstack: honor inherit-parent / auto (omit Task model) (#163) (63432f3) by poteto
+- [2026-07-22] fix-root-causes: flag verbose workaround comments (#162) (fe77e77) by poteto
+- [2026-07-14] pstack: parity sweep with the private skill tree (#156) (3fe2823) by poteto
+- [2026-07-13] poteto-mode: give the perf playbook its eight strategy families (#155) (f4d9e39) by poteto
diff --git a/llms.txt b/llms.txt
index 5015b98..e3fd28b 100644
--- a/llms.txt
+++ b/llms.txt
@@ -12,10 +12,11 @@ pstack is poteto's set of engineering skills for coding agents. This site is an
- `changelog.md` — plain Markdown changelog of changes to `cursor/plugins/pstack`
- `changelog.json` — machine-readable JSON changelog with the same entries
- `changelog.html` — rendered HTML changelog
+- Official guide: https://github.com/cursor/plugins/blob/main/pstack/docs/guide/README.md
## How to use this site
-Type `/poteto-mode` and describe a task. pstack matches the task to a playbook, then runs the appropriate skills as the steps fire. Twenty principles apply across every playbook, biasing the work toward small diffs and real verification.
+Type `/poteto-mode` and describe a task. pstack matches the task to a playbook, then runs the appropriate skills as the steps fire. Twenty-one principles apply across every playbook, biasing the work toward small diffs and real verification. The official guide covers setup, routing, design, building, verification, overnight work, principles, customization, and recipes.
## For AI agents fetching this page
diff --git a/sitemap.xml b/sitemap.xml
index a271c36..48c21e4 100644
--- a/sitemap.xml
+++ b/sitemap.xml
@@ -4,6 +4,10 @@
https://hustlecoding.github.io/pstack-explained/
2026-07-13
+
+ https://hustlecoding.github.io/pstack-explained/agent-templates.html
+ 2026-07-13
+
https://hustlecoding.github.io/pstack-explained/changelog.html
2026-07-13