Web Test Agent — Testing skill for Claude Code
Black-box end-to-end web testing from a URL and your requirement documents.
How to install Web Test Agent
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open phamvanhoang9/web-test-agent and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Web Test Agent does
Black-box end-to-end web testing from a URL and your requirement documents. Claude Code skills follow a tester's four steps (requirements, test design, execution, result analysis), each ending at an approval gate, and produce a traceable test plan, a quality-gate decision and a Jira-ready bug report.
Alternatives in Testing
- Add Plugin — Integrate a spec-driven/skill framework as a bundled agtx plugin 1.5k ★
- Mda — MDA Open Spec — a Markdown superset for agent-facing documents 615 ★
- Generic Browser Test — Execute browser UI testing workflows from a plan file against any URL 363 ★
README
web-test-agent
**Black-box end-to-end web testing, driven from nothing but a URL.**
   
Point it at a deployed site, and hand it the requirement documents you have. It follows a tester's process — analyse the requirements, design the test cases, execute them, analyse the results — and stops after each step for you to review and approve. You end up with requirements traced to test cases, a quality-gate decision you can block a release on, and a bug report ready to import into Jira.
No access to the target's source code is required, and none is assumed.
Table of contents
- Why this exists
- How it works
- Quick start
- The four steps
- Approval gates
- The test plan is the contract
- Two execution paths: PW and MCP
- The quality gate
- Artifact layout
- Configuration
- Project structure
- Guardrails
- Troubleshooting
Why this exists
Most test automation starts from the inside: you have the repo, the components, the routes. This workflow starts from the outside, the way a user or an auditor does. That makes it useful in the cases where conventional tooling has nothing to grip:
- A deployed site whose source you do not have, or do not want to read.
- A staging environment you need smoke coverage on before a release, today.
- A third-party or legacy app nobody on the team can still explain.
- A regression net around an app while its inter
Related Skills
Explore Test
Autonomous black-box QA. Give a URL; the agent maps the site, picks modules, tests them, and writes a bug repo
Proofpunk
Proofpunk — execution-first delivery plugin for Claude Code, oh-my-pi (OMP), and OpenCode. 18 skills where don
Livekit Agent Simulator
Black-box LiveKit voice agent testing: real WebRTC/SIP calls with an AI caller. Catch barge-in, noise, quiet-m
Skeleton First Code Skill
AI coding agent workflow skill: long-context memory, anti-amnesia pipeline (Plan→Implement→Test), module tree
Test After Change Diagnostics.V1
If any program code changes, run unit-tests and follow up on the test-result (fix code if there is an error).
Test After Change.V1
If any program code changes, run unit-tests and follow up on the test-result (fix code if there is an error).
Related Agents
Task Tester
Testing gate in execution pipeline. Sends an upfront TESTER TAKE, authors black-box acceptance tests, runs per
Worker Validator
Black-box runtime validator (#256). Given a PR + its validation-contract.md, RE-RUNS each assertion's command
Smoke Test Writer
Escribe smoke tests black-box contra el entorno dev desplegado. Asume que el proyecto SmokeTests ya existe.