rl-as

Plugin rl-as from marketplace agent-harness in rolopez23/agent-harness. Commands read from the repo's manifest — 1 month ago.

Skills (19)

• problem-spec

  First step for solving any complex coding problem. Defines what is and is not being solved, identifies all existing interfaces and contracts touched, and produces a structured spec doc. Trigger this skill whenever a user asks to build, implement, add, or solve something non-trivial. Do NOT trigger f

• btw-pull-request

  Human invoked like btw put makes a pull requests

• clean-code

  Use to apply coding best practices to working code. Generally you should call this skill whenever you have working code to make sure the code is readable and maintainable.

• command-center

  Show all Claude Code sessions as a table — title, active/idle, branch, and worktree. Trigger when the user asks to list, show, or review their Claude sessions, see which sessions are active/running, or "open the command center". Also trigger on "show my sessions", "which sessions are live", "session

• dispatching-parallel-agents

  Use when facing two or more independent tasks with no shared state or sequential dependency — delegate each to a separate agent with its own precisely-scoped context so they run concurrently. Reach for this whenever work splits into parallel tracks, even if the user doesn't explicitly ask to paralle

• frontend-design

  Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML/CSS layouts, or when styling/beautifying any web UI). Generates creative, polished code and UI design that avoids generi

• fullstack-boot

  Orchestrate a full-stack boot end to end: preflight, contract, parallel subagents that build frontend/ (next-react-boot) and backend/ (python-psql-boot), then link and verify in the main thread (fullstack-link), then harvest skill friction. Trigger on "boot the full stack", "boot frontend and backen

• fullstack-link

  Link a Next.js frontend/ and a FastAPI backend/ into one runnable repo: shared contract, Postgres via docker compose, root npm scripts (setup, db:up, dev, test, smoke), env files, and an end-to-end check that the browser shows the backend's health. Trigger on "link frontend and backend", "wire up th

• initialize

  Installs this harness's skills into a project by writing or updating AGENTS.md and/or CLAUDE.md with a skills reference and canonical workflow, plus a gitignored ai-sessions/ folder and .claude/settings.json read-permission defaults. Run this when setting up a new project, onboarding a repo to the s

• next-react-boot

  Scaffold a Next.js 16 / React 19 / TypeScript / Tailwind CSS v4 frontend from scratch, with Vitest + Testing Library, ESLint flat config, and a backend status indicator. Trigger when the user asks to bootstrap, scaffold, or create a new frontend, Next.js app, or React project ("boot frontend", "new

• plan

  Creates a structured implementation plan from a problem spec, breaking work into independently testable steps with dependency tracking and a living status dashboard. ONLY trigger this skill when a problem spec already exists at <artifacts>/spec.md — if no spec is present, simpler ad-hoc planning suf

• pr-interactive-walkthrough

  Human Invoked Skill

• python-psql-boot

  Scaffold a Python / FastAPI / PostgreSQL backend from scratch with a working /api/health endpoint, async SQLAlchemy, Alembic and pytest. Trigger when the user asks to bootstrap, scaffold, or create a new backend, API, or Python server ("boot backend", "new backend", "scaffold backend", "new API"). R

• refactor

  Restructures existing code for modularity and clarity by actively applying Martin Fowler's refactoring catalog. Distinct from `/clean-code` — refactor is invasive and assumes the user wants the structure changed, while clean-code is conservative and biased toward silence. Trigger when the user says

• review-comprehensive

  Use when reviewing written code to catch logical bugs and missed edge cases. You should aim to generally call this on any substantial code change (more than 50 lines excluding tests and comments).

• systematic-debugging

  Use when encountering any bug, test failure, or unexpected behavior — before proposing or writing a fix. Enforces root-cause diagnosis over quick patches: reproduce, isolate, trace to the underlying cause, then fix. Applies even when the user just says "it's broken" or pastes a stack trace without a

• tone

  Toggles the house conversation tone and coding standards on or off, independently. These are injected each turn by the tone-hooks.sh SessionStart/UserPromptSubmit hooks; this skill flips the persistent flag files that gate them. Trigger when the user says "turn off the tone", "disable coding standar

• verify

  Runs end-to-end verification of implemented code against a live system. For APIs, fires real curl requests. For web UIs, runs browser automation. Reports what passed, what failed, and what could not be verified. Returns "verification incomplete" when the system cannot be reached or the verification

• webapp-testing

  Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

Hooks (3)

• PostToolUse

  ${CLAUDE_PLUGIN_ROOT}/hooks/verify-evidence.sh

• SessionStart

  ${CLAUDE_PLUGIN_ROOT}/hooks/tone-hooks.sh session-start

• UserPromptSubmit

  ${CLAUDE_PLUGIN_ROOT}/hooks/tone-hooks.sh user-prompt-submit

Similar plugins