jorsisu/jor-stack

Review the finished change in the current coding task with Chrome DevTools MCP, preferably through one independent subagent. Use when the user asks for browser review of current work.

Diagnose bugs and performance regressions with an unknown cause. Build a tight feedback loop first, then reproduce, minimise, hypothesise, instrument, fix with a regression test, and clean up.

Shared vocabulary for designing deep modules. Use when designing or improving a module's interface, finding deepening opportunities, deciding where a seam goes, or making code more testable or AI-navigable.

Encode the real domain in a data structure instead of scattered conditionals. Use while writing stateful logic, or when code branches a lot or repeats a shape assumption across files.

Audit jor-stack consistency after skill or workflow changes. Use when asked to check the stack's skills, README, Mermaid flow, routing, metadata, and packaging together; not for application code review.

Investigate the current ticket or request and propose a fix or feature before implementation. Invoke manually after loading the ticket into the conversation.

Relentless interview that sharpens a plan or design one question at a time and writes no files. Use when the user starts from an idea instead of a ticket, or when a decision is still unclear.

Compact the current conversation into a handoff document for another agent to pick up.

Implement clear features and known bug fixes from an authorized request, spec, tickets, or approved proposal, following project rules and patterns.

Ask which jor skill or flow fits your situation. A router over the jor-stack skills.

Review current task changes after jor-implement or jor-bugfix checks pass, or on request. Use one independent agent when available; otherwise review in the current agent. Fix valid findings within scope and make one focused recheck.

Write a PR body: the smallest visual that shows the change, before/after evidence, and merge danger.

Build a throwaway prototype to answer a design question. Use to sanity-check whether a state model or logic feels right, or to explore what a UI should look like with variants screenshotted at the project's breakpoints.

Restate the user's goal and the problem they want solved before work starts. Use only when the user explicitly invokes /jor-restate or $jor-restate, or asks to use the jor-restate skill.

Review a coding session and propose ways to prevent repeated agent mistakes through code structure, automated checks, and minimal guidance. Use when the user asks for a retro.

Optional full review for large changes, when requested: review committed and uncommitted work since a fixed point along Standards and Spec axes with two parallel sub-agents.

Test-driven development. Build a feature or fix test-first, one red-green slice at a time, at existing public or user-agreed seams.

Explain a body of work plainly so a person actually understands it: what it is, how it works, and why it's built that way. Use for 'teach me this', 'help me really understand X', 'explain this change or subsystem to me'.

Turn the current conversation into a private spec under .scratch/, excluded locally from Git. Synthesize agreed requirements and confirm only unresolved test seams.

Break a plan, spec, or the current conversation into tracer-bullet tickets, each declaring its blockers, written as private files under .scratch/ and excluded locally from Git.

Generate a project-local verification skill with scripts for existing repo checks, Chrome DevTools MCP browser journeys, and a feature map. Run once per project, or when a project has no repeatable verification workflow.

Upkeep pass that keeps a project's verification skill and feature map honest: source readers per feature, one live Chrome DevTools MCP pass driving every feature, breakpoint screenshots as evidence. Run on demand or after UI-heavy sprints.

Create, edit, review, or consolidate agent skills and instruction files such as AGENTS.md and CLAUDE.md. Use when the requested output is reusable guidance for an agent.