pr

Plugin pr from marketplace pr-cloud-plugins in PouriaRouzrokh/pr-claude-plugins. Commands read from the repo's manifest — 13 days ago.

Skills (20)

• agentation

  Adds the Agentation visual feedback toolbar to a React 18+ app (Next.js App or Pages Router, Vite, React Router, Remix) so the user can click elements in the browser and send precise, selector-level annotations to the agent, optionally synced live through the agentation-mcp server. Use when the user

• atlas-development

  ATLAS (Architect, Trace, Link, Assemble, Stress-test) is this toolkit's development philosophy for building production-ready software instead of demos: define the problem and success metric, design data and integrations before code, validate every connection before building, assemble in thin end-to-

• clean-codebase

  Cleans and professionalizes code without changing behavior: records a test/lint/build baseline, finds dead code with project tooling (knip, ruff, vulture, tsc) and pr:code-reviewer agents, confirms the plan with the user, then removes dead code, fixes stale comments and docstrings, applies /simplify

• commit-push

  Commit changes, prepend an entry to commitlog.md, and push to the remote, optionally merging into main (--merge) or opening a pull request (--pr).

• create-prd

  Generates a Product Requirements Document through an interactive interview and current web research, then saves it to .claude/checkpoints/checkpoint-0/prd.md. Works from notes, an ideas file, or an existing codebase. Covers problem, users, success metrics, MVP scope, architecture, verified stack ver

• create-snapshot

  Writes a comprehensive technical snapshot of a codebase into a new checkpoint (.claude/checkpoints/checkpoint-N/snapshot.md): reads the PRD, previous snapshots, RFDs, and references, explores the code with four parallel pr:code-explorer agents (architecture, features, development context, and UI/UX

• feature-dev

  Guided feature development for an existing codebase: loads project context from .claude/checkpoints, explores the code with agents, asks clarifying questions, designs and gets the architecture approved, implements, verifies with evidence, reviews with Claude Code's /code-review and /simplify, and re

• frontend-design

  Creates distinctive, production-grade frontend interfaces with intentional visual design, choosing aesthetic direction, typography, color, layout, motion, and interface copy that avoid templated AI defaults. Use when building or redesigning web components, pages, landing pages, dashboards, artifacts

• generate-image

  Generates and edits raster images (photos, illustrations, product shots, blog headers, thumbnails, banners, posters, infographics, stickers, logos with transparent backgrounds, textures) with two model families - Google's Nano Banana (2.1 and Pro) and OpenAI's GPT Image 2.5 (the ChatGPT image model

• generate-svg

  Generates and edits SVG graphics (hero illustrations, UI icons, logo marks, illustrations, UI component graphics, CSS-animated loaders) by having a Gemini model write the markup, then validates, sanitizes, optionally optimizes with svgo, and renders a PNG preview for review. Use when the user asks f

• generate-video

  Generates, edits, and extends short videos (3-10 s clips with audio) with Google's Gemini Omni Flash model through the Gemini API - text-to-video, image-to-video (first frame, first and last frame, reference images), editing an uploaded or previously generated clip, and extending clips up to 40 s. U

• handle-claude-md

  Initializes or updates CLAUDE.md and the .claude/checkpoints documentation structure. For new projects it creates the structure and places or generates the PRD; for existing codebases it sets up checkpoints and triggers a first snapshot; for initialized projects it checks structure integrity and ref

• mvp-dev

  Builds an MVP from the PRD at .claude/checkpoints/checkpoint-0/prd.md: extracts and orders the features, designs and validates the foundation (architecture RFD, Link checks), then implements one feature at a time, each with its own RFD, tests, review, and user sign-off. Resumes from existing RFDs. U

• run-local

  Starts the application locally for development: detects the stack and services, prepares the environment, stops stale processes, starts databases, backends, and frontends in dependency order as background tasks, and verifies each responds. Use when the user asks to run, start, or restart the app or

• run-public

  Deploy the application publicly (preview, staging, or production), verify it is healthy, and report URLs and rollback steps. Confirms before any production deploy.

• systematic-debugging

  Enforces root-cause debugging before any fix - reproduce, gather evidence at component boundaries, trace bad values to their source, test one hypothesis at a time, then fix with a failing regression test first. Use when encountering any bug, test failure, build failure, flaky test, performance regre

• test-driven-development

  Red-green-refactor test-driven development plus a verification gate that forbids claiming work is done, fixed, or passing without fresh command output as evidence. Covers when full TDD applies, writing honest tests (JS/TS with Vitest or Jest, Python with pytest, ML code), and verifying before commit

• update-docs

  Updates project documentation to match the current code: README, API docs, guides, and RFDs. Uses pr:code-explorer agents to find recent changes and documentation gaps, checks that every recent code change is recorded in an RFD, proposes a plan for approval, then applies it. Never edits snapshots, a

• web-quality-audit

  Measures the performance (Core Web Vitals), accessibility (WCAG 2.2 AA), SEO, and best practices of a running web app with Lighthouse and axe-core, then produces a prioritized fix report in which every finding cites measured evidence, a concrete fix, and the skill that owns it. Includes Next.js-spec

• writing-clearly-and-concisely

  Applies Strunk's rules (active voice, positive form, concrete language, omit needless words) and a catalog of AI-writing tells to make prose clear and direct. Use when writing or editing prose humans will read - documentation, READMEs, RFDs, PRDs, CLAUDE.md, commit messages, PR descriptions, error m

Subagents (3)

• code-architect

  Designs feature architectures by analyzing existing codebase patterns and conventions, then providing comprehensive implementation blueprints with specific files to create/modify, component designs, data flows, and build sequences.

• code-explorer

  Deeply analyzes existing codebase features by tracing execution paths, mapping architecture layers, understanding patterns and abstractions, and documenting dependencies to inform new development. Returns the 5-10 files most essential to the topic.

• code-reviewer

  Reviews code for bugs, logic errors, security vulnerabilities, code quality issues, and adherence to project conventions, using confidence-based filtering to report only high-priority issues that truly matter.

Similar plugins