Plugin Claude Code
Open-source Claude Code plugin for QA/QC engineers, QA Managers and Product Ops using AI: scaffold a standard POM project from scratch (one command → a compiling Playwright-Web or Appium-App framework + example test), web/mobile automation, test-case generation, exploratory testing, review, plus quality management…
Instructions file
Claude Code instructions for hominhtuong/qa-claude-plugins, covering qaclaudeplugins — project rules and lark login & authentication policy (mandatory).
Plugin Claude Code
All-in-one QA + quality-management toolkit for Claude Code; commands are namespaced /qa: . Serves QA/QC engineers, QA Managers/Leads and Product Ops. scaffold creates a standard Page Object Model project from scratch (Playwright-Web or Appium-App) — a fresh repo gets a compiling framework + an example login test.
MCP server Claude CodeCodexCursor
MCP server "figma", hosted remotely at mcp.figma.com, as configured in hominhtuong/qa-claude-plugins.
MCP server Claude CodeCodexCursor
MCP server "playwright" as configured in hominhtuong/qa-claude-plugins. Runs ${CLAUDE_PLUGIN_ROOT}/scripts/mcp_playwright.py with python3.
MCP server Claude CodeCodexCursor
Intelligent MCP server providing AI assistants with powerful tools and resources for Appium mobile automation. Runs locally from the appium-mcp npm package. Needs 1 environment variable to run.
Agent
Read Figma design screens and extract structured UI summary (layout, elements, states). Use when a Figma URL is provided and design context needs to be fetched. Returns structured markdown.
Agent
Read a single Lark wiki/docx document and extract structured content (text, tables, images, comments, embedded links). Reads via the plugin's Python helper (app/tenant token with auto auth — NOT the Lark MCP, which fails with token-expired 99991668). Use when a Lark URL is provided and document content needs to be…
Agent
Discover elements from live app via Appium MCP, save as elements.json, then create/update Screen classes. Use when elements are unknown or need verification.
Agent
Read ONE spec source (local file, generic web URL, or pasted text) and extract a QC-relevant structured summary (functional requirements, business rules, validation, UI states, data, edge cases, questions). Collects embedded Lark/Figma/external links and reports them back so the main agent can fan out lark-reader /…
Command
Analyze a requirement/spec doc from a QA perspective (testable items, ambiguity, risk, questions for the PO) into an md file used as input for /qa:plan-gen-testcases and /qa:gen-testcases (read-only, no edits).
Command
Analyze codebase, test results or project structure for AUTOMATION (analysis only, no code changes).
Command
Answer questions about the codebase, architecture, configuration, or testing approach (answer only, do not modify code).
Command
Optional Figma token setup for /qa:exploratory-ui — most users DON'T need it (exploratory-ui reads Figma through the bundled Figma MCP, just log in when asked). A FIGMATOKEN is only a fallback for headless/no-MCP runs or to fetch the exact design text+font oracle. When needed, you (Claude) open the Figma token page in…
Command
Authenticate Lark (Feishu) in BOTH token modes and verify permissions — tenant (app token from LARKAPPID/LARKAPPSECRET) and/or user (UAT via OAuth login). Probes what each mode can do (Bitable / Drive / Docs), resolves the effective read mode (preference user(default)|auto|tenant; user prefers the real-person token so…
Command
Deep-analyze any bug/ticket board over a time range. Reads every record from a board URL (or the active board), ADAPTIVELY classifies that board's own status names into readiness groups (not-ready-to-test = New/Rework/Reopen/Fixing/DevDone-style vs ready vs closed), then analyzes the not-ready backlog — counts by…
Command
Check for a duplicate bug on the Lark Bitable board BEFORE logging a new one. Parses free text or a structured payload, extracts 2–3 keywords, searches the board (read-only), drops closed records + keyword-only false positives, and returns a decision (safe to create / potential duplicate → you pick). Never creates a…
Command
Write AUTOMATION test code from a plan or a request following the design pattern; auto-routes by platform (web Playwright / app Appium), reading only the matching platform skill.
Command
Count the existing AUTOMATION regression testcases (@Test methods) of a feature + estimate the total, broken down by sub-feature.
Command
Count the MANUAL test cases in a created sheet/plan and report coverage by feature/section.
Command
Estimate Story Points for QC effort (write TCs + execute + log bugs + retest, NOT dev effort) from a plan or a feature description, factored by the executor's role (USERROLE in .plugin.env). Counts/estimates TCs, assesses complexity/scope/risk to a base SP, applies the role multiplier, rounds to Fibonacci, outputs the…
Command
Translate" a poorly-written bug into a clear, structured summary (summary, repro steps, actual vs expected, preliminary severity, Severity/Priority sanity check). Handles raw/no-diacritics/vague text, a screenshot, a Bug ID, or a full Lark record link (reads record + comments + history from the board, read-only).…
Command
Design-conformance exploratory — open a feature screen (web/android/ios), capture screenshots, render the Figma design to PNGs, then compare each app screen vs its design using a LOCAL CV engine (color CIEDE2000 Delta-E / SSIM / palette / perceptual-hash) so heavy image work runs locally and only tiny JSON verdicts…
Command
Navigate to a feature screen (web/android/ios), read & analyze any spec provided (local file / URL / Figma / pasted prompt) into an expected-behavior oracle, then explore & hunt bugs like a senior QA against that oracle, capture evidence + produce a bug report; GATE for writing tests. Auto-route by platform, reading…