Verify Bug

What the source does

• FilesystemNone observedNo reads or writes in the source
• NetworkNone observedNo outbound calls in the source
• ShellNone observedNo commands run in the source
• DependenciesPinnedNothing floats
• SecretsNone foundNo keys in source

Five audit layers

Gates applied: no behavioural pass.

Audit 0.4.1 · source sha 6f5be687a268 · 14 Oct 2026Report an issue or request a re-scan

• PublisherHoangNguyen0403 GitHub userOwner of the repository this skill is read from
• Source folderHoangNguyen0403/agent-skills-standard/.codex/skills/verify-bugThe exact folder we audited
• LicenceMITAs the repository states it
• Commit pinned for this verdict6f5be687a268A new audit runs when the source changes
• Repository statusNot archivedPushed today
• Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

Its description: Post-merge UAT verification workflow. Walks JIRA reproduce steps, performs comparative audits (Before/After), attaches evidence to JIRA, and transitions status on PASS.

---
name: verify-bug
description: "Post-merge UAT verification workflow. Walks JIRA reproduce steps, performs comparative audits (Before/After), attaches evidence to JIRA, and transitions status on PASS."
metadata:
  internal: true
  triggers:
    keywords:
    - verify bug
    - workflow
---
# Verify Bug Skill

> [!IMPORTANT]
> Post-merge UAT verification workflow. Walks JIRA reproduce steps, performs comparative audits (Before/After), attaches evidence to JIRA, and transitions status on PASS.

Optional args: slug=<feature>, ticket=<id/url>, mode=interactive|autonomous|channel, channel=<id>, auto_continue=true|false, profile=business|hybrid|technical.

## Instructions

When the user asks to perform this workflow, execute the following steps:


# Verify-Bug — UAT Audit

Goal: Prove a bug fix works in the UAT environment via comparative Before/After evidence, then transition ticket status.

## Input

`/verify-bug <jira-url-or-key> [--baseline-image <url>]`

## Workflow

### Step 0: Pre-flight & Data Gathering

> [!TIP]
> **Sub-Agent Delegation**: If your platform supports sub-agents (e.g., Claude, OpenCode, Gemini, Kiro), delegate steps 1-3 below to your JIRA Analyst sub-agent (e.g., `@specialist-jira-analyst`). If sub-agents are NOT supported (e.g., Antigravity, Windsurf), you must execute these steps yourself.

1.  **Parse JIRA**: Extract `Market`, `Reproduce steps`, and `Expected Result`.
2.  **Resolve Markets**: If multiple markets, prompt for scope (Full/Sample/Custom).
3.  **Fetch Test Data**: Call Confluence for `Test data - <MARKET> UAT`. Parse credentials and module-specific data (e.g., customer codes).
4.  **Credential Check**: Rule out expired accounts before starting sessions.
5.  **Fallback**: If Jira/Confluence MCPs are unavailable, request exported ticket/test-data text and continue with local evidence.

### Step 1: Comparative Audit (Execution Phase)

For each market in scope:

1.  **Environment Setup**: Run the DNS probe from `<SKILLS>/common/common-web-visual-testing/references/diagnostic-decoder.md`; if it indicates VPN is required, connect VPN and retry.
2.  **Named Session**: Run the driver skill's `scripts/preflight.sh`, then start `playwright-cli -s={TICKET}-{MARKET}` (Playwright MCP with `--output-dir .playwright-cli/{TICKET}-{MARKET}` on no-shell runtimes) or an Appium MCP session. Evidence lands in that session dir; a missing driver with no exported evidence is `BLOCKED (driver: <name>)`.
3.  **Walk Steps**: Execute reproduction steps.
    - **Hover Discipline**: Always `hover` the target element (warning, button, price) before screenshotting.
    - **Stability**: Disable animations and mask dynamic fields (clocks, balances).
4.  **Verdict Determination**:
    - **PASS**: End-state matches `Expected Result`.
    - **FAIL**: End-state matches `Actual Result` or original bug screenshot.
    - **NEEDS-HUMAN**: Deviates from both.

### Step 2: Automated Failure Diagnostic

If the verdict is NOT PASS:

1.  **Run Decoder**: Load `common-web-visual-testing`; if synced references are available, consult `<SKILLS>/common/common-web-visual-testing/references/diagnostic-decoder.md`.
2.  **Categorize**: Is it a `VPN NOT CONNECTED` error? `ACCOUNT BLOCKED`? Or a genuine `CODE REGRESSION`?
3.  **Label**: Add the diagnostic label to the JIRA comment.

### Step 3: Evidence & JIRA Sync

1.  **Upload**: Push screenshots from the session evidence dir as attachments to the JIRA ticket.
2.  **Wiki Comment**: Post a verdict comment using JIRA Wiki Markup (orientation-aware widths).
    - Use `🟢 PASS` / `🔴 FAIL` badges.
    - Embed the most diagnostic screenshot inline.
3.  **Status Transition**:
    - If **PASS**: `Ready for UAT` → `Ready for Production`.
    - If **FAIL**: → `Reopened`.
4.  **Walkthrough**:
    - Use the **Walkthrough Template** below.
    - Update project-local `docs/srs/srs-walkthrough-[slug].md`.

## Runtime Contract
- Use for post-merge UAT verification of a bug fix against JIRA reproduce steps.
- Required inputs: JIRA URL/key or exported ticket text with reproduce steps and expected result.
- Return NEEDS-HUMAN only when the end-state deviates from both expected and original-bug behavior.

## Handoff Payload
- `slug`, `operator_profile` (carried, not re-inferred), verdict (PASS/FAIL/NEEDS-HUMAN), walkthrough path, diagnostic label, outcome report, next workflow.

## Blocking Questions
- Ask max 3 at a time with a recommended default and 2-3 options.

## Artifact Templates

### Walkthrough Template

```md
# Walkthrough: [Name]

## Scope

## Acceptance Criteria

## Evidence

| Check   | Result              | Evidence   |
| ------- | ------------------- | ---------- |
| [check] | [PASS/FAIL/BLOCKED] | [evidence] |

## Risks

## Outcome Report
{schema_version: 1, run_id: "[run-id]", slug: "[slug]", workflow: verify-bug, feature_status: verified, started_at: "[timestamp]", completed_at: "[timestamp]", requirement_trace: {brd_objectives: [], requirements: [], acceptance_criteria: [], srs: []}, completed_evidence: [], missing_evidence: [], decision_needed: [], recommended_next_workflow: deploy-release, cost: {source: unavailable}, agent: {identity: "[agent-identity]", model: "[model]"}}

## Next Workflow
deploy-release | dev-fix
```

## Cost Report

Call `get_session_cost(workflow="verify-bug")` before final handoff.

## Anti-Patterns

- **No Sequential Runs**: Verify all markets in parallel.
- **No Unnamed Sessions**: Traceability depends on `-s={TICKET}`.
- **No Mystery Failures**: Always include the Diagnostic Decoder result in FAIL comments.
- **No Orphan Comments**: Clean up "temp media" comments after posting the final verdict.

Excerpt, first 5,659 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

1. 2026-10-14
   ✓SAFE B 89 / 100

   first audit · source 6f5be687a268

Questions about Verify Bug

• github.com/HoangNguyen0403/agent-skills-standard/tree/develop/.c — the audited source at commit 6f5be687a268 and every finding aboveObserved · 14 Oct 2026

Spotted something wrong? Report a correction · Methodology