reproduce - SKILL.md Agent Skill

name: reproduce description: > Systematically reproduce a Quay bug using project-specific tooling. Documents environment, reproduction steps, and observable behavior. allowed-tools: - Bash(make ) - Bash(pytest ) - Bash(python ) - Bash(docker ) - Bash(podman ) - Bash(git ) - Bash(npm ) - Bash(npx ) - Bash(curl *) - Read - Write - Glob - Grep - Agent - AskUserQuestion

Reproduce Bug

Confirm the bug exists by reproducing it systematically.

Your Role

Methodically reproduce the bug and document its behavior so that diagnosis and fixing can proceed with confidence.

Process

Step 1: Parse Bug Report

Read the assessment (artifacts/quay-bugfix/reports/assessment.md if it exists) and extract:

Bug description and expected vs actual behavior
Affected components, versions, environment details
Error messages, stack traces, or relevant logs
UI vs backend classification

Step 2: Set Up Environment

For backend bugs:

Check what's available and set up accordingly:

# Check for Quay's local dev setup
ls Makefile 2>/dev/null && grep -l 'local-dev' Makefile
# Check for virtual env
ls pyproject.toml requirements.txt 2>/dev/null

Common Quay setup approaches:

make local-dev-up — full local development environment (Docker, DB, Redis)
Virtual environment — python -m venv .venv && pip install -r requirements.txt
Direct pytest — for unit-level reproduction

For UI bugs:

# Check for frontend dev server
ls web/package.json 2>/dev/null

Start dev server if needed: cd web && npm install && npm start
Default app URL: http://localhost:9000
Test credentials: user1 / password

Step 3: Attempt Reproduction

Follow the steps from the assessment report. For each attempt, document:

Exact commands or UI actions performed
Expected behavior (from bug report)
Actual behavior (what happened)
Any error messages or stack traces

Try variations to understand the bug's boundaries:

Different input values
Different user roles/permissions
Different database backends (PostgreSQL vs MySQL)
Different configurations
Race conditions (if timing-related)

Step 4: Document Reproduction

Create a minimal set of steps that reliably reproduce the bug.

Step 5: Write Reproduction Report

Save to artifacts/quay-bugfix/reports/reproduction.md:

# Reproduction Report: <TICKET>

## Environment
- Quay version/branch: <branch>
- Setup method: <local-dev-up / venv / docker / pytest>
- Database: <PostgreSQL / MySQL / SQLite>
- Python version: <version>

## Reproduction Steps
1. <exact step>
2. <exact step>
...

## Results
- **Reproduced:** Yes / No / Intermittent
- **Success rate:** Always / Often / Sometimes / Rare
- **Expected:** <what should happen>
- **Actual:** <what does happen>
- **Error output:** <if any>

## Variations Tested
| Variation | Reproduced? | Notes |
|-----------|:-----------:|-------|
| ... | ... | ... |

## Minimal Reproduction
<Shortest path to trigger the bug>

## Notes
<Any observations, workarounds, or additional context>

Confidence Assessment

Include this section at the end of your reproduction artifact:

## Confidence Assessment
- **Level**: high | medium | low
- **Score rationale**: <1-2 sentences>
- **Open questions**: <bullet list, or "None">

Confidence signals for this phase:

High: Bug reproduced consistently with minimal steps
Medium: Bug reproduced with extra setup or specific configuration, or intermittent but follows a pattern
Low: Cannot reproduce, fully intermittent, or environment-dependent

Output

artifacts/quay-bugfix/reports/reproduction.md

Error Handling

If reproduction fails:

Document exactly what was tried and what differed from the report
Check environment differences (versions, config, data)
Consider the bug may be environment-specific or intermittent
Record findings with a "Could Not Reproduce" status

When This Phase Is Done

Report: whether the bug was reproduced, key observations, and where the reproduction report was written.