A tool for diagnosing and making one permitted repair to a failing end-to-end test, which tests a complete user flow through the application. It uses run artifacts such as traces, screenshots, logs, and the product change, then checks the repair with three reruns.

Use it for one failing end-to-end test inside a test workflow when evidence from the failed run is available.

It separates selector, timing, data, infrastructure, and real-product failures instead of masking a real regression. Its limits prevent weakening assertions or changing production code just to make a test pass.

test-healer — Agent for Claude Code by HoangNguyen0403

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo

git clone --depth 1 https://github.com/HoangNguyen0403/agent-skills-standard

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/agents/hoangnguyen0403/agent-skills-standard/test-healer/github.svg)](https://agentmods.dev/agents/hoangnguyen0403/agent-skills-standard/test-healer)

Your own site

<a href="https://agentmods.dev/agents/hoangnguyen0403/agent-skills-standard/test-healer"><img src="https://agentmods.dev/badge/agents/hoangnguyen0403/agent-skills-standard/test-healer/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/agents/hoangnguyen0403/agent-skills-standard/test-healer"><img src="https://agentmods.dev/badge/agents/hoangnguyen0403/agent-skills-standard/test-healer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 63 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 949 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00063	$0.00949
Opus 5.5	$0.00025	$0.00380
Sonnet 5.5	$0.00013	$0.00190
Haiku 4.5	$0.00006	$0.00095

Measured 1mo ago against content hash ccd06bba5454, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

test-healer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/agents/test-healer.md · 57 lines

How it starts

The opening of the file, as written. The whole thing — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Specialist: Test Healer

Priority: P0 (CRITICAL)

Role

For one failing test, decide SELECTOR_DRIFT | TIMING_SYNC | DATA_ENV | INFRA | REAL_REGRESSION from evidence, apply one repair the catalog allows, and prove it, per quality-engineering-test-healing.

Budget

One failing test per invocation; at most 15 tool calls.
Read: the failing run's trace, screenshot, DOM or hierarchy dump, console or device log, the test file, and the diff of the product since the last green run.
Write: only the test file, its page object, or its fixture; one repair per invocation.
Rerun: the single test, sequentially, in the foreground, 3 consecutive times after the repair; HEALED requires 3/3 green.
Retry without a change only for INFRA, once.
Never touch production code, weaken an assertion, widen a matcher, inflate a timeout past 2x, add test.skip/fixme, or run --update-snapshots.
No Git, no sub-agents.
Return BLOCKED (no evidence artifact) when no trace, screenshot, or log exists to classify from, or BLOCKED (no stable locator target) with ROUTE: testid-inserter.

Steps

Load the failure artifact and the product diff since the last green; refuse to classify from the error message alone.
Classify using the signal table in quality-engineering-test-healing failure taxonomy; a screenshot assertion failure is VISUAL_DIFF and is judged by quality-engineering-visual-baseline before it can be anything but REAL_REGRESSION.
REAL_REGRESSION: stop, no repair; verdict REAL_BUG_DO_NOT_HEAL, route dev-fix, attach the evidence.
Otherwise apply exactly one repair from the repair catalog: move the locator up the ladder, replace a sleep with a state wait, fix the seed or fixture; a missing stable id is VERDICT: BLOCKED with ROUTE: testid-inserter, never a CSS fallback.
Rerun 3 consecutive sequential foreground runs; compute ASSERTION_DELTA by comparing assertion count and matcher strength before and after; anything but none is not a heal. Fewer than 3/3: revert the repair; 1-2/3 green is QUARANTINE_CANDIDATE; 0/3 reclassifies once from the post-repair artifact.
Intermittent across the reruns with no code change: verdict QUARANTINE_CANDIDATE, route flaky-triage with the rerun tally.

Read the full file on GitHub · 57 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

1mo ago First seen · 57 lines · 63 tokens per session scan A ccd06bba5454

Subscribe to this mod's changes

test-healer is an agent published in the GitHub repository HoangNguyen0403/agent-skills-standard (577 stars, last pushed yesterday), licensed MIT. It adds 63 tokens to every session and 949 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-14.

E2E testing React Testing best-practices coding-standards flutter nestjs nextjs

delivery-validator

Runtime delivery verification agent. Boots applications, runs smoke tests, verifies developer experience and technology currency. Write access limited to running servers and install commands — never modifies source code.

irahardianto/awesome-agv · 37 tokens

test-automation-engineer

Hands-on test automation engineer. Invoke for writing E2E tests, building test infrastructure, managing test data, configuring coverage reporting, and investigating flaky tests. MCP-first Playwright automation. This agent writes test code — not production code.

irahardianto/awesome-agv · 53 tokens

uds-feature-implementer

Use to implement a feature whose goal and integration points are already defined, refactor a module toward a stated target structure, or write integration tests for a specified subsystem. Use when most steps are settled and only a bounded set of local choices remains. Do NOT use when the target structure itself is…

AsiaOstrich/universal-dev-standards · 74 tokens

testing-engineer

Use this agent when implementing comprehensive Flutter test suites with unit tests, widget tests, integration tests, golden tests, mocktail mocking, Riverpod testing, GoRouter testing, test coverage enforcement, and CI test pipeline optimization.

altmemy/claude-code-templates · 48 tokens

e2e-runner

Specialized agent that runs the full Playwright e2e suite in the project's Dockerized Chromium environment (docker-compose.e2e.yml) and reports pass/fail. Use before pushing to GitHub to guarantee e2e never regresses on main. Trigger on "run e2e", "e2e before push", "playwright check", or invoked by the pre-push git…

KaelSensei/MagicAIBuilder · 87 tokens

browser-tester

AI-driven browser testing agent. Takes ARIA snapshots, interacts via refs or accessible names, verifies with 2+ signals per assertion. Auto-detects MCP capability tier (batched playwright or individual browser calls). Supports spec-driven, basic, visual, and a11y modes.

ByeongminLee/nextjs-claude-code · 62 tokens