Adoption

Agent Skills are supported by leading AI development tools.

VS Code Gemini CLI GitHub Goose Amp Cursor Claude Code Letta OpenCode Claude OpenAI Codex Factory VS Code Gemini CLI GitHub Goose Amp Cursor Claude Code Letta OpenCode Claude OpenAI Codex Factory

astra-technology-company-limited/user-test

Name: user-test
Author: astra-technology-company-limited

skills/user-test/SKILL.md

npx skillsauth add astra-technology-company-limited/astra-methodology user-test

Clean

TrivyContainer and dependency vulnerability scanner

Clean

SemgrepStatic code analysis for vulnerabilities

Clean

mcp-scan (Snyk)Model Context Protocol security validation

Skipped

Snyk (dep)Open source security scanning

Skipped

Socket.devSupply chain security analysis

Skipped

VirusTotalMulti-engine malware detection

Skipped

CrowdStrikeAdvanced threat intelligence

Skipped

OSV-ScannerOpen Source Vulnerability database check

Skipped

OWASP Dep-Check

ASTRA /user-test — AI-assisted UAT

Drives Chrome MCP to run end-user UAT flows. Claude executes each step, self-verifies with hard assertions, and writes a session folder containing an HTML report, session.json, and (when failures occur) issues.md. Output language follows the project's /select-language setting; default is Vietnamese (this skill was designed for Vietnamese-speaking teams).

Difference from sibling skills:

/test-run — developer-authored technical integration tests (server launch + scenario verify).
/test-scenario — generates E2E scenarios from blueprints.
/user-test — UAT: input is natural-language Vietnamese or normalized UAT case files; output is a QA-readable report.

Detailed references (load on demand):

references/assertion-guide.md — assertion syntax + severity rules.
references/i18n-strings.md — vi/en/ko translation table for HTML report + issues.md.
assets/report-template.html — HTML report template (uses {{LANG}} + {{T_*}} placeholders).

1. Mode Decision

| Trigger | Mode | Input | |---|---|---| | $ARGUMENTS contains --auto | Auto | UAT case files under docs/tests/uat-cases/ (or --from) | | $ARGUMENTS starts with http:// or https:// | Interactive | URL + natural-language flow description | | Otherwise | Interactive | Prompt user for URL + description via AskUserQuestion |

Argument parsing

--auto: enable Auto mode.
--from {path}: Auto-only. Restrict to a single case file. Default: glob docs/tests/uat-cases/*.md.
--priority {critical|high|medium|low}: Auto-only. Filter by frontmatter priority.
--feature {name}: Auto-only. Filter by frontmatter feature.
--lang {vi|en|ko}: Report-output language for index.html, issues.md, and console logs. If omitted, see Step 0 below.

2. Pipeline (shared by both modes)

Step 0 — Language Selection (report output language)

Determine LANG_CODE ∈ {vi, en, ko} — used for index.html (<html lang> + visible labels), issues.md headings, and console log messages. UAT case file contents themselves are not translated.

Resolution order:

--lang flag in $ARGUMENTS → normalize case-insensitive: vi|vie|vietnamese → vi, en|eng|english → en, ko|kor|korean → ko. If recognized, skip to Step 1.
Persisted CLAUDE.md ## Language in the project (set by /select-language) → if it resolves to ko, vi, or en, use it silently.
Otherwise → ask the user via AskUserQuestion with the trilingual prompt below. Default selection is Vietnamese (preserves the skill's original design).

Chọn ngôn ngữ cho báo cáo UAT.
Select the language for the UAT report.
UAT 보고서 언어를 선택하세요.

Options (single-select, header Lang):

Tiếng Việt — Vietnamese (Recommended)
English — English
한국어 — Korean

Map: Tiếng Việt → vi, English → en, 한국어 → ko.

Under --auto: if no --lang and no persisted CLAUDE.md ## Language, default to vi silently (no prompt) so /autorun stays unattended.

Once resolved, hold LANG_CODE in memory for all downstream steps. Load references/i18n-strings.md to get the strings dictionary for that language.

Step 1 — Load input and prepare test cases

Interactive:

Extract URL and Vietnamese flow description from $ARGUMENTS.
Decompose the description into steps. Each step needs an action (navigate / click / fill / wait) and expected hard assertions.
If the description is vague (e.g., "đăng ký"), reason from the URL: open page → find primary CTA → fill form → submit → verify next page. Do not interrupt to ask the user — reason and proceed.
Hold the draft case in memory; do not write to disk yet.

Auto:

Glob test case files (or honor --from).
Read each YAML frontmatter; apply --priority / --feature filters.
Announce: "Tìm thấy N test case sẽ chạy" then continue (no confirmation prompt).

Step 2 — Initialize session folder

SESSION_ID  = {YYYY-MM-DD-HHmm}     # e.g. 2026-05-21-1030
SESSION_DIR = docs/tests/uat-reports/{SESSION_ID}/

Create:

{SESSION_DIR}/screenshots/

{SESSION_DIR}/session.json initialized with:

{
  "session_id": "{SESSION_ID}",
  "mode": "interactive | auto",
  "started_at": "{ISO}",
  "test_cases": []
}

Step 3 — Test loop (Claude executes + self-verifies)

For each test case → for each step:

A. Execute action

| Action syntax | Chrome MCP tool | |---|---| | navigate {url} | mcp__chrome-devtools__navigate_page | | click {selector} | mcp__chrome-devtools__click | | fill {selector} value={value} | mcp__chrome-devtools__fill | | wait {ms} or wait {selector} | mcp__chrome-devtools__wait_for | | press {key} | mcp__chrome-devtools__press_key | | hover {selector} | mcp__chrome-devtools__hover |

Replace literal {timestamp} in any value with Date.now() to avoid duplicate-data collisions.

Before the very first navigate, ensure a page exists via list_pages; if none, call new_page.

B. Screenshot every step

After each action, call take_screenshot into {SESSION_DIR}/screenshots/step-{NN}-{slug}.png. {slug} = step name lowercased, spaces → -, Vietnamese diacritics removed.

C. Verify hard assertions

Read the step's expected list and verify each. See references/assertion-guide.md for full syntax. Summary:

url: equals|contains|matches {x} → compare via evaluate_script (window.location.href).
network: METHOD {path} → {status} → list_network_requests, match method+path, check status.
dom: {selector} exists|visible|contains "..."|has value matching {re} → take_snapshot or evaluate_script.
console: no error|no warning → list_console_messages, filter by level.

All assertions for a step must PASS for the step to PASS.

D. Record result

PASS → append step result to session.json; continue. Log: [N/M] {step_name} ✅ PASS.
FAIL → assign severity per references/assertion-guide.md, append to issues.md, log: [N/M] {step_name} ❌ FAIL — {SEVERITY}, skip remaining steps in this test case, move to the next test case.

Step 4 — Generate reports

After the loop completes. All visible strings come from references/i18n-strings.md for the resolved LANG_CODE.

A. index.html — load assets/report-template.html and substitute placeholders:

{{LANG}} → LANG_CODE (vi / en / ko)
{{SESSION_ID}}, {{STARTED_AT}}, {{FINISHED_AT}}, {{MODE}}
{{TOTAL_CASES}}, {{PASS_COUNT}}, {{FAIL_COUNT}}, {{DURATION}}
{{T_*}} — every i18n placeholder from i18n-strings.md (report title, subtitle, labels, headings, footer, status badge text).
{{TEST_CASES_HTML}} — render each test case with thumbnails and per-step status (see template's HTML comment for the per-step structure). Use T_STATUS_PASS / T_STATUS_FAIL / T_STATUS_SKIPPED for the badge text.
{{ISSUES_HTML}} — when no FAIL: <div class="empty"><div class="icon">🎉</div>{T_EMPTY_NO_ISSUES}</div>. Otherwise: <p>{T_SEE_DETAILS_AT} <a href="./issues.md">issues.md</a> ({M} {T_ISSUES_WORD}: …)</p>.

Write to {SESSION_DIR}/index.html.

B. issues.md — only create if there is at least one FAIL. Use the M_* strings dictionary for headings. Example for LANG_CODE = vi:

# Báo cáo UAT Issues
**Session**: {YYYY-MM-DD HH:mm}
**Test Cases chạy**: {N}
**Tổng số lỗi**: {M} ({X} CRITICAL, {Y} HIGH, {Z} MEDIUM, {W} LOW)

---

## Issue #1 — {SEVERITY}
**Test Case**: {UAT-ID} - {name}
**Bước**: {step_num}/{total} — {step_name}

### Expected
- {assertion 1}

### Actual
- {observed value}

### Lý do gán {SEVERITY}
{One-sentence rationale referencing the rule from assertion-guide.md.}

### Screenshot
![step-NN-slug](./screenshots/step-NN-slug.png)

### Gợi ý cho dev
- {hint 1}

For LANG_CODE = en, replace every Markdown heading/label via i18n-strings.md (# UAT Issues Report, **Test cases run**, **Total issues**, **Step**, ### Reason for {SEVERITY}, ### Hint for developers). For LANG_CODE = ko, use the Korean column (# UAT 이슈 리포트, **실행된 테스트 케이스**, **단계**, ### {SEVERITY} 사유, ### 개발자 가이드). Severity badges (CRITICAL/HIGH/MEDIUM/LOW) stay untranslated.

C. Finalize session.json with finished_at, summary (pass/fail counts), lang (the resolved LANG_CODE), and an issues array.

Step 5 — (Interactive only) Offer to save the test case

Only when mode = interactive and at least one step PASSed. Use AskUserQuestion with the L_SAVE_PROMPT string for the resolved LANG_CODE (see references/i18n-strings.md).

If yes, ask for feature (slug, e.g. dang-ky) and priority, then write docs/tests/uat-cases/{feature}.md using the format in §3 below.

Step 6 — Summarize

Print the L_DONE string for the resolved LANG_CODE, followed by stats. Example for vi:

▶ Hoàn thành.
   📊 {PASS} PASS / {FAIL} FAIL
   📄 {SESSION_DIR}/index.html
   🐛 issues.md có {M} issue ({severity breakdown})

For en / ko, swap the ▶ Done / ▶ 완료 heading and translate the issues-line label (issues.md has {M} issues / issues.md에 {M}건의 이슈). Skip the issues.md line when no FAIL.

3. UAT case file format (Markdown + YAML frontmatter)

---
id: UAT-001
name: Đăng ký tài khoản với email hợp lệ
priority: critical
created: 2026-05-21
feature: auth
base_url: https://staging.fect.app
---

# Đăng ký tài khoản

## Bước 1: Mở trang chủ
**Action**: navigate `https://staging.fect.app`
**Expected**:
- url: equals `https://staging.fect.app/`
- network: GET `/` → 200
- dom: `button:has-text("Đăng ký")` exists
- dom: `title` contains "Fect"

## Bước 2: Click "Đăng ký"
**Action**: click `button:has-text("Đăng ký")`
**Expected**:
- url: contains `/register`
- dom: `input[type="email"]` exists

## Bước 3: Điền email
**Action**: fill `input[type="email"]` value=`test+{timestamp}@example.com`
**Expected**:
- dom: `input[type="email"]` has value matching `test\+\d+@example\.com`

## Bước 4: Submit form
**Action**: click `button[type="submit"]`
**Expected**:
- network: POST `/api/auth/register` → 201
- url: contains `/verify-email`
- dom: text "Kiểm tra email" exists

Conventions:

Each ## Bước N: {name} is one step.
**Action**: single line, verb selector [value=...].
**Expected**: bullet list, one assertion per line.

4. Output structure

docs/tests/uat-reports/2026-05-21-1030/
├── index.html              # Visual report (open in browser)
├── issues.md               # Only if FAIL count > 0
├── session.json            # Raw data for re-run/debug
└── screenshots/
    ├── step-01-home.png
    └── ...

docs/tests/uat-cases/
├── dang-ky.md
├── dang-nhap.md
└── tao-workspace.md

5. Examples

/user-test https://staging.fect.app "đăng ký tài khoản bằng email [email protected]"
/user-test --auto
/user-test --auto --from docs/tests/uat-cases/dang-ky.md
/user-test --auto --priority critical
/user-test --auto --lang en              # English report
/user-test --auto --lang ko              # Korean report
/user-test https://staging.fect.app "register a new account" --lang en

6. Standing instructions

User-facing output language: resolved in Step 0 (--lang flag → CLAUDE.md ## Language → AskUserQuestion → default vi). The chosen LANG_CODE drives <html lang>, every {{T_*}} placeholder in the HTML template, issues.md headings/labels, and console log strings. File slugs always use ASCII (no diacritics). references/i18n-strings.md is the SSoT for translations.
Minimal user interaction: ask only at the start (missing URL) and end of an Interactive session (save case?). Never prompt inside the test loop.
On FAIL, do not abort the whole run: skip the current test case and continue with the next.
Every step screenshots — even on PASS — so the report is browsable.
Severity assignment is rule-based: always reference references/assertion-guide.md. Never guess by feel.
issues.md always includes "Lý do gán severity" — one sentence citing the rule, so devs can re-evaluate.
No GitHub Issue integration (removed in v2): write only to local issues.md.
Hard assertions only: do not assert on visual fidelity or human judgment. URL/Network/DOM/Console only.

7. Anti-scope (do NOT use for)

Pure API testing (no UI) → use curl / integration tests.
Unit tests → project's test framework (Jest/JUnit/pytest).
Developer-written technical scenarios → use /test-run.
Authoring new E2E scenarios from a blueprint → use /test-scenario first.

astra-technology-company-limited/user-test

skills/user-test/SKILL.md

Performs end-user UAT (User Acceptance Testing) by driving a real browser through Chrome MCP, self-verifying each step with hard assertions (DOM / Network / URL / Console), auto-assigning severity on failure, and emitting an HTML report plus issues.md into a timestamped session folder. Supports two modes: interactive (URL + Vietnamese natural-language flow description) and --auto (batch-run pre-authored test cases under docs/tests/uat-cases/). Use when the user asks for "UAT", "user acceptance test", "kiểm thử người dùng", "regression test", or runs /user-test, /uat. Distinct from /test-run (developer-authored technical integration testing) and /test-scenario (scenario authoring from blueprints).

tools

Updated Jun 4, 2026

$ install --global

skillsauth

npx skillsauth add astra-technology-company-limited/astra-methodology user-test

Install this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.

Security Scan Results

3 of 9 scanners reported clean

Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.

Scanners Passed

Scanners in report

Clean

TrivyContainer and dependency vulnerability scanner

95%

Clean

SemgrepStatic code analysis for vulnerabilities

95%

Clean

mcp-scan (Snyk)Model Context Protocol security validation

95%

Skipped

Snyk (dep)Open source security scanning

50%

Skipped

Socket.devSupply chain security analysis

50%

Skipped

VirusTotalMulti-engine malware detection

50%

Skipped

CrowdStrikeAdvanced threat intelligence

50%

Skipped

OSV-ScannerOpen Source Vulnerability database check

50%

Skipped

OWASP Dep-Check

50%

Last scanned: Jun 4, 2026, 4:59 AM159.9s3 files scanned

SKILL.md

name:: user-test
description:: >
Supports two modes:: interactive (URL + Vietnamese natural-language flow
argument-hint:: [URL \"flow description\"] | --auto [--from file] [--priority critical|high|medium|low] [--feature name] [--lang vi|en|ko]
allowed-tools:: Read, Write, Edit, Bash, Glob, Grep, AskUserQuestion, mcp__chrome-devtools__take_snapshot, mcp__chrome-devtools__take_screenshot, mcp__chrome-devtools__navigate_page, mcp__chrome-devtools__click, mcp__chrome-devtools__fill, mcp__chrome-devtools__fill_form, mcp__chrome-devtools__press_key, mcp__chrome-devtools__hover, mcp__chrome-devtools__list_console_messages, mcp__chrome-devtools__get_console_message, mcp__chrome-devtools__list_network_requests, mcp__chrome-devtools__get_network_request, mcp__chrome-devtools__evaluate_script, mcp__chrome-devtools__wait_for, mcp__chrome-devtools__emulate, mcp__chrome-devtools__resize_page, mcp__chrome-devtools__list_pages, mcp__chrome-devtools__select_page, mcp__chrome-devtools__new_page, mcp__chrome-devtools__handle_dialog

ASTRA /user-test — AI-assisted UAT

Difference from sibling skills:

/test-run — developer-authored technical integration tests (server launch + scenario verify).
/test-scenario — generates E2E scenarios from blueprints.
/user-test — UAT: input is natural-language Vietnamese or normalized UAT case files; output is a QA-readable report.

Detailed references (load on demand):

references/assertion-guide.md — assertion syntax + severity rules.
references/i18n-strings.md — vi/en/ko translation table for HTML report + issues.md.
assets/report-template.html — HTML report template (uses {{LANG}} + {{T_*}} placeholders).

1. Mode Decision

Argument parsing

--auto: enable Auto mode.
--from {path}: Auto-only. Restrict to a single case file. Default: glob docs/tests/uat-cases/*.md.
--priority {critical|high|medium|low}: Auto-only. Filter by frontmatter priority.
--feature {name}: Auto-only. Filter by frontmatter feature.
--lang {vi|en|ko}: Report-output language for index.html, issues.md, and console logs. If omitted, see Step 0 below.

2. Pipeline (shared by both modes)

Step 0 — Language Selection (report output language)

Resolution order:

--lang flag in $ARGUMENTS → normalize case-insensitive: vi|vie|vietnamese → vi, en|eng|english → en, ko|kor|korean → ko. If recognized, skip to Step 1.
Persisted CLAUDE.md ## Language in the project (set by /select-language) → if it resolves to ko, vi, or en, use it silently.
Otherwise → ask the user via AskUserQuestion with the trilingual prompt below. Default selection is Vietnamese (preserves the skill's original design).

Chọn ngôn ngữ cho báo cáo UAT.
Select the language for the UAT report.
UAT 보고서 언어를 선택하세요.

Options (single-select, header Lang):

Tiếng Việt — Vietnamese (Recommended)
English — English
한국어 — Korean

Map: Tiếng Việt → vi, English → en, 한국어 → ko.

Under --auto: if no --lang and no persisted CLAUDE.md ## Language, default to vi silently (no prompt) so /autorun stays unattended.

Once resolved, hold LANG_CODE in memory for all downstream steps. Load references/i18n-strings.md to get the strings dictionary for that language.

Step 1 — Load input and prepare test cases

Interactive:

Extract URL and Vietnamese flow description from $ARGUMENTS.
Decompose the description into steps. Each step needs an action (navigate / click / fill / wait) and expected hard assertions.
If the description is vague (e.g., "đăng ký"), reason from the URL: open page → find primary CTA → fill form → submit → verify next page. Do not interrupt to ask the user — reason and proceed.
Hold the draft case in memory; do not write to disk yet.

Auto:

Glob test case files (or honor --from).
Read each YAML frontmatter; apply --priority / --feature filters.
Announce: "Tìm thấy N test case sẽ chạy" then continue (no confirmation prompt).

Step 2 — Initialize session folder

SESSION_ID  = {YYYY-MM-DD-HHmm}     # e.g. 2026-05-21-1030
SESSION_DIR = docs/tests/uat-reports/{SESSION_ID}/

Create:

{SESSION_DIR}/screenshots/

{SESSION_DIR}/session.json initialized with:

{
  "session_id": "{SESSION_ID}",
  "mode": "interactive | auto",
  "started_at": "{ISO}",
  "test_cases": []
}

Step 3 — Test loop (Claude executes + self-verifies)

For each test case → for each step:

A. Execute action

Replace literal {timestamp} in any value with Date.now() to avoid duplicate-data collisions.

Before the very first navigate, ensure a page exists via list_pages; if none, call new_page.

B. Screenshot every step

After each action, call take_screenshot into {SESSION_DIR}/screenshots/step-{NN}-{slug}.png. {slug} = step name lowercased, spaces → -, Vietnamese diacritics removed.

C. Verify hard assertions

Read the step's expected list and verify each. See references/assertion-guide.md for full syntax. Summary:

url: equals|contains|matches {x} → compare via evaluate_script (window.location.href).
network: METHOD {path} → {status} → list_network_requests, match method+path, check status.
dom: {selector} exists|visible|contains "..."|has value matching {re} → take_snapshot or evaluate_script.
console: no error|no warning → list_console_messages, filter by level.

All assertions for a step must PASS for the step to PASS.

D. Record result

PASS → append step result to session.json; continue. Log: [N/M] {step_name} ✅ PASS.
FAIL → assign severity per references/assertion-guide.md, append to issues.md, log: [N/M] {step_name} ❌ FAIL — {SEVERITY}, skip remaining steps in this test case, move to the next test case.

Step 4 — Generate reports

After the loop completes. All visible strings come from references/i18n-strings.md for the resolved LANG_CODE.

A. index.html — load assets/report-template.html and substitute placeholders:

{{LANG}} → LANG_CODE (vi / en / ko)
{{SESSION_ID}}, {{STARTED_AT}}, {{FINISHED_AT}}, {{MODE}}
{{TOTAL_CASES}}, {{PASS_COUNT}}, {{FAIL_COUNT}}, {{DURATION}}
{{T_*}} — every i18n placeholder from i18n-strings.md (report title, subtitle, labels, headings, footer, status badge text).
{{TEST_CASES_HTML}} — render each test case with thumbnails and per-step status (see template's HTML comment for the per-step structure). Use T_STATUS_PASS / T_STATUS_FAIL / T_STATUS_SKIPPED for the badge text.
{{ISSUES_HTML}} — when no FAIL: <div class="empty"><div class="icon">🎉</div>{T_EMPTY_NO_ISSUES}</div>. Otherwise: <p>{T_SEE_DETAILS_AT} <a href="./issues.md">issues.md</a> ({M} {T_ISSUES_WORD}: …)</p>.

Write to {SESSION_DIR}/index.html.

B. issues.md — only create if there is at least one FAIL. Use the M_* strings dictionary for headings. Example for LANG_CODE = vi:

# Báo cáo UAT Issues
**Session**: {YYYY-MM-DD HH:mm}
**Test Cases chạy**: {N}
**Tổng số lỗi**: {M} ({X} CRITICAL, {Y} HIGH, {Z} MEDIUM, {W} LOW)

---

## Issue #1 — {SEVERITY}
**Test Case**: {UAT-ID} - {name}
**Bước**: {step_num}/{total} — {step_name}

### Expected
- {assertion 1}

### Actual
- {observed value}

### Lý do gán {SEVERITY}
{One-sentence rationale referencing the rule from assertion-guide.md.}

### Screenshot
![step-NN-slug](./screenshots/step-NN-slug.png)

### Gợi ý cho dev
- {hint 1}

C. Finalize session.json with finished_at, summary (pass/fail counts), lang (the resolved LANG_CODE), and an issues array.

Step 5 — (Interactive only) Offer to save the test case

Only when mode = interactive and at least one step PASSed. Use AskUserQuestion with the L_SAVE_PROMPT string for the resolved LANG_CODE (see references/i18n-strings.md).

If yes, ask for feature (slug, e.g. dang-ky) and priority, then write docs/tests/uat-cases/{feature}.md using the format in §3 below.

Step 6 — Summarize

Print the L_DONE string for the resolved LANG_CODE, followed by stats. Example for vi:

▶ Hoàn thành.
   📊 {PASS} PASS / {FAIL} FAIL
   📄 {SESSION_DIR}/index.html
   🐛 issues.md có {M} issue ({severity breakdown})

For en / ko, swap the ▶ Done / ▶ 완료 heading and translate the issues-line label (issues.md has {M} issues / issues.md에 {M}건의 이슈). Skip the issues.md line when no FAIL.

3. UAT case file format (Markdown + YAML frontmatter)

---
id: UAT-001
name: Đăng ký tài khoản với email hợp lệ
priority: critical
created: 2026-05-21
feature: auth
base_url: https://staging.fect.app
---

# Đăng ký tài khoản

## Bước 1: Mở trang chủ
**Action**: navigate `https://staging.fect.app`
**Expected**:
- url: equals `https://staging.fect.app/`
- network: GET `/` → 200
- dom: `button:has-text("Đăng ký")` exists
- dom: `title` contains "Fect"

## Bước 2: Click "Đăng ký"
**Action**: click `button:has-text("Đăng ký")`
**Expected**:
- url: contains `/register`
- dom: `input[type="email"]` exists

## Bước 3: Điền email
**Action**: fill `input[type="email"]` value=`test+{timestamp}@example.com`
**Expected**:
- dom: `input[type="email"]` has value matching `test\+\d+@example\.com`

## Bước 4: Submit form
**Action**: click `button[type="submit"]`
**Expected**:
- network: POST `/api/auth/register` → 201
- url: contains `/verify-email`
- dom: text "Kiểm tra email" exists

Conventions:

Each ## Bước N: {name} is one step.
**Action**: single line, verb selector [value=...].
**Expected**: bullet list, one assertion per line.

4. Output structure

docs/tests/uat-reports/2026-05-21-1030/
├── index.html              # Visual report (open in browser)
├── issues.md               # Only if FAIL count > 0
├── session.json            # Raw data for re-run/debug
└── screenshots/
    ├── step-01-home.png
    └── ...

docs/tests/uat-cases/
├── dang-ky.md
├── dang-nhap.md
└── tao-workspace.md

5. Examples

/user-test https://staging.fect.app "đăng ký tài khoản bằng email [email protected]"
/user-test --auto
/user-test --auto --from docs/tests/uat-cases/dang-ky.md
/user-test --auto --priority critical
/user-test --auto --lang en              # English report
/user-test --auto --lang ko              # Korean report
/user-test https://staging.fect.app "register a new account" --lang en

6. Standing instructions

User-facing output language: resolved in Step 0 (--lang flag → CLAUDE.md ## Language → AskUserQuestion → default vi). The chosen LANG_CODE drives <html lang>, every {{T_*}} placeholder in the HTML template, issues.md headings/labels, and console log strings. File slugs always use ASCII (no diacritics). references/i18n-strings.md is the SSoT for translations.
Minimal user interaction: ask only at the start (missing URL) and end of an Interactive session (save case?). Never prompt inside the test loop.
On FAIL, do not abort the whole run: skip the current test case and continue with the next.
Every step screenshots — even on PASS — so the report is browsable.
Severity assignment is rule-based: always reference references/assertion-guide.md. Never guess by feel.
issues.md always includes "Lý do gán severity" — one sentence citing the rule, so devs can re-evaluate.
No GitHub Issue integration (removed in v2): write only to local issues.md.
Hard assertions only: do not assert on visual fidelity or human judgment. URL/Network/DOM/Console only.

7. Anti-scope (do NOT use for)

Pure API testing (no UI) → use curl / integration tests.
Unit tests → project's test framework (Jest/JUnit/pytest).
Developer-written technical scenarios → use /test-run.
Authoring new E2E scenarios from a blueprint → use /test-scenario first.

Related Skills

astra-technology-company-limited/uat-parallel

tools

VerifiedTrustedCommunity

Runs UAT (User Acceptance Testing) cases in TRUE PARALLEL using Playwright Test runner with isolated browser contexts per worker (separate cookies, localStorage, sessionStorage). Solves the two main limits of /user-test: (1) sequential single-page execution that does not scale beyond a few cases, and (2) one stuck case blocking the rest of the run. Reuses 100% of the /user-test UAT case Markdown+YAML format under docs/tests/uat-cases/, runs them via `npx playwright test --workers=N`, and emits the same report layout (index.html + issues.md + session.json + screenshots/) under docs/tests/uat-reports/. Use when the user asks to "run UAT in parallel", "speed up UAT", "test multi-user", "song song", "uat parallel", or runs /uat-parallel. Distinct from /user-test (sequential Chrome MCP, supports interactive mode), /test-run (developer integration tests), /test-scenario (scenario authoring).

SKILL.mdUpdated May 30, 2026

astra-technology-company-limited/uat-parallel

astra-technology-company-limited/tool-author

tools

VerifiedTrustedCommunity

Authors and validates LLM tool descriptions and input schemas (Anthropic Tool Use, MCP servers, LangChain @tool, Pydantic, Zod). Use when the user mentions "tool description", "function calling", "MCP tool", "Pydantic schema", "Zod schema", "@tool decorator", "input_schema", "tool spec", "툴 정의", "함수 호출 스키마", or when editing files that define LLM tool surfaces. Enforces the six required attributes (one-line summary, anti-pattern, synonyms, parameter examples, enum constraints, return shape) and blocks the seven known failure modes — wrong-tool selection, skipped tool, malformed arguments, retry loops, user-intent bypass, wrong side-effect, and un-auditable traces. For authoring ASTRA SKILL.md files use /skill-author instead — this skill is for *runtime* LLM tool surfaces, not for skill files themselves.

SKILL.mdUpdated May 22, 2026

astra-technology-company-limited/tool-author

astra-technology-company-limited/skill-author

development

VerifiedTrustedCommunity

Creates new SKILL.md files or refactors existing skills to comply with the ASTRA skill best practices guide (docs/development/skill-best-practices.md). Use when user mentions "new skill", "create skill", "SKILL.md", "skill authoring", "스킬 작성", "스킬 만들기", or when editing any file matching skills/**/SKILL.md.

SKILL.mdUpdated May 21, 2026

astra-technology-company-limited/skill-author

astra-technology-company-limited/design-redesign

tools

VerifiedTrustedCommunity

Audits existing UI components/pages/CSS against docs/design-system/DESIGN.md and applies fixes to restore design consistency. Uses design-token-validator to detect hardcoded color/font/spacing violations, and invokes designer-persona to report a senior-perspective score (0-10) and anti-AI aesthetic violations (generic shadcn look, purple gradient cliché). Fixes are applied automatically (--apply) or proposed as a PR (--pr); after changes, design-token-validator is re-run and must PASS before completion. Input: target directory (e.g., src/components/Button), single file, or git diff. Output: audit report (docs/design-system/audit-{date}.md) + fix proposals + application results.

SKILL.mdUpdated May 21, 2026

astra-technology-company-limited/design-redesign

Download

For Claude Desktop. Download once, then upload the file in the app — no terminal needed.

Need help? View full Cowork setup guide →

Install manually

Choose your platform

# Clone the repo
git clone https://github.com/astra-technology-company-limited/astra-methodology.git

# Copy into Claude Code skills folder (global)
cp -r astra-methodology/skills/user-test ~/.claude/skills/

Claude Code Skills — official skills path docs.

Repository

astra-technology-company-limited/astra-methodology

Compatible with

Claude Code

OpenAI Codex CLI

ChatGPT