Skip to content
Skills Library
Programming · Testing

validation

/validation

Exhaustive manual validation procedure for web applications. Use when validating UI features, testing interactions, or auditing human-agent collaborative QA sessions.

Licence
MIT
Last repository change
11.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 11.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Exhaustive manual validation procedure for web applications. Use when validating UI features, testing interactions, or auditing human-agent collaborative QA sessions.

Original author description

Exhaustive manual validation procedure for web applications. Use when validating UI features, testing interactions, or auditing human-agent collaborative QA sessions.

How it works

  1. 1. Purpose
  2. 2. Validation Passes (V1..V5)
  3. 3. Execution Protocol
  4. 4. Common Anti-Patterns Detected by This Procedure
  5. 5. Validation Report Template
Original SKILL.md headings · 171 lines, 1,313 words
SKILL.md preview
---
name: validation
description: Exhaustive manual validation procedure for web applications. Use when validating UI features, testing interactions, or auditing human-agent collaborative QA sessions.
---
# Skill: Manual Web Validation

## 1. Purpose

Codifies the exhaustive manual validation procedure used in human–agent
collaborative QA. When an agent assists a human with manual testing, or
executes validation autonomously (e.g. via browser tools), this skill
defines **what to check, in what order, and how to report findings**.

## 2. Validation Passes (`V1..V5`)

Every web page/feature is validated through 5 orthogonal validation passes (`V1..V5`, distinct from `rcap1` `L0..L3` capsule levels),
executed in this order:

```
V1  Render        →  Does it load? Does it look right?
V2  Interact      →  Do actions work? Do they produce correct results?
V3  Persist       →  Do changes survive a reload?
V4  Observe       →  What does the browser tell us? (console, network, DOM)
V5  Boundary      →  What happens at the edges? (empty states, errors, stress)
```

### 2.1. `V1` — Render Check

Verify initial page rendering, cache-bypass reload, layout integrity, and CSS tokens:

| Step | Action | What to verify |
|------|--------|---------------|
| 1.1 | Navigate to URL | Page loads, no blank screen, no 500 |
| 1.2 | Hard reload (Ctrl+Shift+R) | Same result on cache-bypass |
| 1.3 | Visual scan | Layout correct, no broken elements, no overflow |
| 1.4 | Framework chrome | Header/footer/nav visible (`data-osa="framework"` present on header, main, footer in DOM when using OSA/Cockpit surfaces) |
| 1.5 | CSS tokens | DevTools → Elements → verify CSS custom properties (`var(--osa-*)`) active on elements |
| 1.6 | Debug markers | Dev-mode indicators visible if enabled (`data-osa-debug` overlay when `DEBUG=true`) |

### 2.2. `V2` — Interaction Check

Verify primary, secondary, destructive, and navigation interactions in the UI:

| Step | Action | What to verify |
|------|--------|---------------|
| 2.1 | Primary action (create/submit/search) | Action completes, UI updates |
| 2.2 | Observe DOM update | Does the new element appear in the right container? Correct layout? |
| 2.3 | Secondary action (edit/update) | Changes reflected immediately in UI |
| 2.4 | Destructive action (delete/clear) | Element removed, no orphan state |
| 2.5 | Navigation | Links work, back/forward correct, no broken routes |

### 2.3. `V3` — Persistence Check

Verify that mutations, deletions, edits, and URL query state survive full page reloads:

| Step | Action | What to verify |
|------|--------|---------------|
| 3.1 | After each mutation, reload page | Data survives reload |
| 3.2 | After delete, reload page | Deletion persists |
| 3.3 | After edit, reload page | Edit persists with exact values |
| 3.4 | Check URL state | Search params, query strings preserved on reload |

### 2.4. `V4` — Observation Check (DevTools)

Inspect browser Console, Network XHR/fetch payloads, and DOM container hierarchy during interactions:

| Step | Action | What to verify |
|------|--------|---------------|
| 4.1 | Console tab open during ALL interactions | No errors (red), no unexpected warnings |
| 4.2 | Network tab: XHR/fetch/HTMX requests | Correct request method, headers (`HX-Request: true` returning HTML partial fragment rather than full page), response format |
| 4.3 | Network tab: status codes | 200/201 for success (or 286 to stop HTMX polling via `respond_poll`), 4xx for client errors, no 500s |
| 4.4 | Elements tab: DOM after dynamic update | New element inside correct parent container via `hx-swap="innerHTML"` (never orphaned outside conditional grid wrapper) |
| 4.5 | Source attribution | External errors (browser extensions, 3rd-party) vs application code |

> **Key heuristic**: If a JS error shows `<anonymous>:2:NNNNN` with minified
> function names like `et.reportAllChanges`, it's likely a browser extension,
> not application code. Verify by testing in incognito without extensions.

### 2.5. `V5` — Boundary Check

Exercise empty states, accidental/invalid inputs, rapid double-submits, and large payload rendering:

| Step | Action | What to verify |
|------|--------|---------------|
| 5.1 | Empty state | Page with zero data shows appropriate message, no broken layout |
| 5.2 | Accidental input | Typing in search by accident → no crash, debounce works |
| 5.3 | Invalid input | Required fields empty → validation message, no 500 |
| 5.4 | Rapid actions | Click submit multiple times fast → no duplicates, no race conditions |
| 5.5 | Large data | Long text, many items → layout doesn't break, scrolling works |

## 3. Execution Protocol

### 3.1. For Agent-Assisted Validation (Human tests, agent guides)

1. **Agent prepares checklist** from the project's validation tracker (`WORKVALI.md`), expanded with `V1`–`V5` passes
2. **Human executes** steps in browser, narrating observations
3. **Agent analyzes** the narration in real-time:
   - Classifies each observation as ✅ pass, ⚠️ cosmetic, or 🔴 bug
   - Cross-references errors against application source (following `1. Guides/Specs -> 2. Tests [E2E -> Imp -> Unit] -> 3. Code`)
   - Identifies root cause when possible
4. **Agent updates** `WORKVALI.md` with results and captures issues in `WORKTASK.md` / `BACKLOG.md`
5. **Automated Test Materialization (`V1–V5 -> Unit / Imp / E2E`)**: Whenever a validation defect is fixed, materialize the corresponding automated regression test (`tests/unit/` for isolated logic, `tests/imp/` for boundary/adapter implementation, or `tests/e2e/` for the vertical persona journey) inside its `Test<Module>` suite so manual QA defects never regress
6. **Human confirms** findings and decides priority

### 3.2. For Autonomous Agent Validation (with browser tools)

1. Agent starts the application server
2. Agent navigates each route using browser automation tools
3. Agent captures screenshots at each `V1` step
4. Agent executes `V2` interactions programmatically
5. Agent checks console/network for `V4` observations
6. Agent generates structured validation report

### 3.3. Narration Format (for human reports)

When a human reports observations, use this mental model to parse:

```
"I opened the landing page, looks fine"
 └── V1.1 ✅ (render check: page loads)

"Reloaded the page, all good"
 └── V1.2 ✅ + V3 ✅ (cache bypass + persistence)

"Typed a letter in the search box and the frontend shows an error"
 └── V5.2 ⚠️ (boundary: accidental input) + V4.1 🔴 (console error)

"The new item renders full-width [...] After reload it's in the grid"
 └── V2.2 ⚠️ (DOM update: wrong container) + V3.1 ✅ (persists correctly)

"I can't tell if the header/footer comes from the framework or my code"
 └── V1.4 ⚠️ (framework chrome: not distinguishable → observability gap)
```

## 4. Common Anti-Patterns Detected by This Procedure

The table below maps frequent UI/DOM anti-patterns to the validation pass (`V1..V4`) that detects them:

| Pattern | How it surfaces | Root cause | Prevention |
|---------|----------------|------------|------------|
| Dynamic insert outside wrapper | V2.2: element outside grid/list after create | JS/HTMX inserts before wrapper element | Return full list content, use `innerHTML` swap |
| Browser extension console error | V4.1: error from `<anonymous>` source | Extension injects monitoring scripts | Verify in incognito mode |
| Missing framework observability | V1.4: can't distinguish framework vs app HTML | No visual/DOM markers | Add data attributes + dev-mode overlay |
| State loss on dynamic update | V3.4: URL state lost after partial update | Missing URL push on navigation | Push URL state on search/filter actions |

## 5. Validation Report Template

After completing a validation session, the report follows this structure:

```markdown
## [Application Name] — Validation Report

**Date**: YYYY-MM-DD
**Tester**: human | agent | collaborative
**Environment**: [dev/staging/production]

### Results

| Pass | Check | Result | Notes |
|------|-------|--------|-------|
| V1 | Landing renders | ✅ | |
| V1 | Reload | ✅ | |
| V2 | Create | ✅ | |
| V2 | Edit | ✅ | |
| V2 | Delete | ✅ | |
| V3 | Persistence | ✅ | |
| V4 | Console clean | ⚠️ | Extension error (not app) |
| V4 | Dynamic update | ⚠️ | Layout inconsistency on create |
| V5 | Empty state | ✅ | |

### Issues Found

1. **[BUG/COSMETIC]** Description → promoted to task tracker / captured in backlog
```

Versions

RepositoryScoreLicence
This page · mtwuser/zero-cloud-gravity100MIT
ruujOnChain/HedgeFund-Skills100MIT
hamzabellouch/agent-skills100MIT
minhuw/claude-writer100MIT
richard-kim-79/archora-skills100MIT
matteocervelli/llms95MIT
openai/codex-security90Apache-2.0
agentic-swe/agentic-swe90MIT

Database snapshot: 13.11.2026 · package version a89a86ac8c51 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100agent-integration-testing
    Use when the user requests integration testing, feature validation, or test plan execution
  • 100playwright-cli
    Automates browser interactions for testing and validating your own web applications using playwright-cli. Use when you need terminal-first browser control for navigation, form filling, screenshots, tracing, bound browser sessions, debugging, or generating Playwright test code.…
  • 100agent-browser
    Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
  • 100agent-browser
    Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
  • 100agent-browser
    Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
  • 100agent-browser
    Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.