skills.rest
cursor
cursorOFFICIAL
Explore all skills by this author →
WebsiteGitHub

Poteto Mode

Orchestrates agent workflows with playbooks, principles, subagents, and verified PR delivery.

6.6k|542|Updated Jan 23, 2026
Search Tags:#verification#agent workflow#subagents#playbooks#pr automation#cursor#Code review
One-click install
npx skills add https://github.com/cursor/plugins --skill poteto-mode
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Poteto Mode
Source: https://github.com/cursor/plugins/tree/main/pstack/skills/poteto-mode
Command: npx skills add https://github.com/cursor/plugins --skill poteto-mode

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires commander, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Agent-driven coding work often drifts into vague plans, unverified claims, bloated prose, and unreviewed subagent output. Poteto Mode enforces a disciplined operating style where every task routes through a named playbook, every principle cited traces to a real decision, and every change is verified against the real artifact before shipping.

Core Features & Use Cases

  • Playbook routing: Matches each task (bug fix, feature, investigation, perf, babysit, shipping, autonomous run, orchestration) to a step-by-step playbook whose steps are copied verbatim into the todolist.
  • Principle system: Applies named engineering principles (model the domain, prove it works, laziness protocol, sequence verifiable units) with mandatory leaf-skill reading before citation.
  • Subagent discipline: Routes delegates through poteto-agent with per-role model defaults, background execution, and mandatory diff review by the parent.
  • Use Case: A user says "get this PR stack green and ship it". The agent declares a babysit mode, works the merge frontier, triages Bugbot comments skeptically, then hands off to the Shipping playbook which independently verifies each PR before landing the contiguous verified run.

Quick Start

Ask the agent to work in poteto mode on your next bug fix or feature, for example by saying "use /poteto-mode to fix this bug and ship it as a PR".

Frequently Asked Questions about Poteto Mode

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I use poteto mode for a bug fix?▼

Invoke /poteto-mode and describe the defect. The agent matches the Bug fix playbook, reproduces the issue itself on the matching surface, binary-searches the root cause with runtime evidence, delegates the fix to a subagent, and verifies the original repro now passes before opening a PR.

What playbooks does poteto mode include?▼

It ships playbooks for investigation, bug fix, perf issue, hillclimb, runtime and trace forensics, feature, refactoring, prototype, visual parity, skill authoring, eval, babysit, shipping, autonomous run, orchestrate, autopilot variants, session pickup, pause safely, multi-phase plans, worktree cleanup, and opening PRs.

How does poteto mode handle Bugbot and automated review comments?▼

It triages each comment skeptically using the bugbot-triage reference, classifying it as fix, dismiss, or ask. Real findings get fixed in the lowest owning PR with a cited commit, while documented noise patterns are dismissed with a concrete disproof rather than churning code.

Can poteto mode run tasks autonomously overnight?▼

Yes. The Autonomous run playbook defines a checkable exit predicate, arms a wake mechanism via Cursor's /loop command, and iterates with verification and a decision trail until the predicate is met. Longer programs route to the Orchestrate or Autopilot playbooks.

When should I not use poteto mode?▼

Skip it for casual conversational turns or when the user opts out, as its own reminder states. Its heavyweight playbooks, subagent fan-out, and verification gates are designed for nontrivial engineering work, not quick questions or trivial edits.

Related Skills

Explore More Software Engineering Skills→
renderincrenderinc

render-mcp

Configure and connect the Render MCP server for Cursor, Claude Code, and Codex.

Official
Intermediate
binfen1binfen1

land-and-deploy

Merge PRs, wait for CI, deploy to production, and verify canary health.

Community
Advanced
GetFresh-VenturesGetFresh-Ventures

verify-execution

Verify code changes by observing runtime outputs on CLI, API, or GUI surfaces.

Official
Intermediate
tbqh511tbqh511

agent-md-refactor

Refactor oversized agent instruction files into a minimal root and linked topic documents.

Community
Intermediate