Sn4kyGit/qa-agent-blueprint
Ticket-driven QA Manager + Tester for Claude Code and opencode: JIRA ticket in, tested ticket out — mandatory evidence, project-conformant bug reports, and a capture-groom-promote learning loop. Model-agnostic under opencode.
ARCHETYPE
Fort Knox
Deny lists, pre-tool hooks, sandboxing. Nothing touches prod without a signature.
GUARDRAILS
4/5
Blocks destructive commands · Protects secrets · Pre-tool screening hook · No YOLO mode · details
Quality 74/100 Guardrails v2
Copy this rig
# review before running: this installs third-party code $ claude mcp add playwright -- npx @playwright/mcp@0.0.84 --config .playwright-mcp.config.json --secrets .env.playwright-secrets --output-dir evidence/.mcp-out --image-responses omit $ claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp $ npx degit Sn4kyGit/qa-agent-blueprint/.claude ./rig-qa-agent-blueprint # inspect, then merge into .claude/
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ claude mcp add playwright -- npx @playwright/mcp@0.0.84 --config .playwright-mcp.config.json --secrets .env.playwright-secrets --output-dir evidence/.mcp-out --image-responses omit $ claude mcp add --transport http atlassian https://mcp.atlassian.com/v2/mcp
$ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/accessibility-testing .claude/skills/accessibility-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/agentswitch .claude/skills/agentswitch $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/api-testing .claude/skills/api-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/bug-report .claude/skills/bug-report $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/bug-reproduction .claude/skills/bug-reproduction $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/demo .claude/skills/demo $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/deploy-to-stage .claude/skills/deploy-to-stage $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/exploratory-testing .claude/skills/exploratory-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/failure-analysis .claude/skills/failure-analysis $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/model-setup .claude/skills/model-setup $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/onboard-product .claude/skills/onboard-product $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/onboard-project .claude/skills/onboard-project $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/perf-analysis .claude/skills/perf-analysis $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/qa-help .claude/skills/qa-help $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/qa-metrics .claude/skills/qa-metrics $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/qa-report .claude/skills/qa-report $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/re-board .claude/skills/re-board $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/recon .claude/skills/recon $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/release-readiness .claude/skills/release-readiness $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/report-to-jira .claude/skills/report-to-jira $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/retest-bug .claude/skills/retest-bug $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/retro-learn .claude/skills/retro-learn $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/risk-based-testing .claude/skills/risk-based-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/scaffold-suite .claude/skills/scaffold-suite $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/security-testing .claude/skills/security-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/skill-review .claude/skills/skill-review $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/test-automation .claude/skills/test-automation $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/test-case-management .claude/skills/test-case-management $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/test-reliability .claude/skills/test-reliability $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/test-ticket .claude/skills/test-ticket $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/visual-testing .claude/skills/visual-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/webdriverio-testing .claude/skills/webdriverio-testing $ npx degit Sn4kyGit/qa-agent-blueprint/.claude/skills/whats-new .claude/skills/whats-new
$ curl -fsSL --create-dirs -o .claude/agents/bug-reporter.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/bug-reporter.md $ curl -fsSL --create-dirs -o .claude/agents/failure-analyst.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/failure-analyst.md $ curl -fsSL --create-dirs -o .claude/agents/qa-manager.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/qa-manager.md $ curl -fsSL --create-dirs -o .claude/agents/test-automator-cheap.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-automator-cheap.md $ curl -fsSL --create-dirs -o .claude/agents/test-automator-deep.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-automator-deep.md $ curl -fsSL --create-dirs -o .claude/agents/test-automator.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-automator.md $ curl -fsSL --create-dirs -o .claude/agents/test-executor-cheap.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-executor-cheap.md $ curl -fsSL --create-dirs -o .claude/agents/test-executor-deep.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-executor-deep.md $ curl -fsSL --create-dirs -o .claude/agents/test-executor.md https://raw.githubusercontent.com/Sn4kyGit/qa-agent-blueprint/main/.claude/agents/test-executor.md
Merge into .claude/settings.json (project) or ~/.claude/settings.json (user). Hook commands reference scripts in the source repo — copy those too.
{
"permissions": {
"deny": [
"Read(.env)",
"Read(.env.*)",
"Read(**/.env)",
"Read(**/.env.*)",
"Read(.env.playwright-secrets)",
"Read(**/*.pem)",
"Read(**/id_rsa*)",
"Read(**/id_ed25519*)",
"Read(**/*.auth)",
"Read(**/*storage-state*)",
"Read(**/*storage_state*)",
"Read(evidence/.mcp-out/**)",
"Read(**/evidence/.mcp-out/**)",
"Read(**/evidence/**/*.png)",
"Read(**/evidence/**/*.jpg)",
"Read(**/evidence/**/*.jpeg)",
"Read(**/evidence/**/*.gif)",
"Read(**/evidence/**/*.webp)",
"Read(**/evidence/**/*.bmp)",
"Read(**/evidence/**/*.webm)",
"Read(**/evidence/**/*.mp4)",
"Read(**/evidence/**/*.mov)",
"Read(**/evidence/**/*.zip)",
"Read(**/evidence/**/*.har)"
]
},
"hooks": {
"PreToolUse": [
{
"matcher": "Bash|Write|Edit|MultiEdit|NotebookEdit|Read|Grep|Glob|WebFetch|Task",
"hooks": [
{
"type": "command",
"command": "${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh"
}
]
},
{
"matcher": "mcp__.*atlassian__(jira_)?(create|edit|add|transition|update|delete|executeWrite|executeDestructive).*",
"hooks": [
{
"type": "command",
"command": "${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh"
}
]
},
{
"matcher": "mcp__.*(playwright|atlassian)__.*",
"hooks": [
{
"type": "command",
"command": "${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh"
}
]
},
{
"matcher": "mcp__.*",
"hooks": [
{
"type": "command",
"command": "${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh"
}
]
}
]
}
} MCP servers (2)
| server | source | est. tokens |
|---|---|---|
| Playwright MCP · "playwright" | npm | 7.5k |
| Atlassian | remote · remote | 10.0k |
Skills (33)
accessibility-testingagentswitchapi-testingbug-reportbug-reproductiondemodeploy-to-stageexploratory-testingfailure-analysismodel-setuponboard-productonboard-projectperf-analysisqa-helpqa-metricsqa-reportre-boardreconrelease-readinessreport-to-jiraretest-bugretro-learnrisk-based-testingscaffold-suitesecurity-testingskill-reviewtest-automationtest-case-managementtest-reliabilitytest-ticketvisual-testingwebdriverio-testingwhats-new
Subagents (9)
| bug-reporter model: inherit | Composes a bug report strictly per the active project's template (projects/<KEY>/templates/bug-report.md), with correct severity/priority mapping and evidence references. Use whenever a test run produ |
| failure-analyst model: inherit | Triages a failing test — is this a product defect, or is the test case itself outdated because the product deliberately changed? Compares the failing assertion against the JIRA change record and the v |
| qa-manager model: inherit | Orchestrates a full ticket-driven QA workflow end to end — reads the JIRA ticket, routes to the project file, decides the test strategy, delegates to risk-selected test-executor/test-automator lanes p |
| test-automator-cheap model: haiku | Writes and runs deterministic specs for the low-risk, single-branch Fast Path in the suite's framework (Playwright or WebdriverIO) via the CLI runner. Use only for the unambiguous cheap execution lane |
| test-automator-deep model: opus | Writes and runs deterministic specs for high-risk tickets or escalated execution branches in the suite's framework (Playwright or WebdriverIO) via the CLI runner. Use for deep execution while preservi |
| test-automator model: sonnet | Writes and runs deterministic specs in the suite's framework (Playwright or WebdriverIO) via the CLI runner — for regression, repeatable checks, and nightly-style suites. Use once a flow is understood |
| test-executor-cheap model: haiku | Runs low-risk, single-branch Fast Path exploratory testing and bug reproduction via Playwright MCP inside an isolated browser session. Use only for the unambiguous cheap execution lane; returns a comp |
| test-executor-deep model: opus | Runs high-risk or escalated exploratory testing and bug reproduction via Playwright MCP inside an isolated browser session. Use for deep execution on high-risk tickets or escalated branches; returns a |
| test-executor model: sonnet | Runs exploratory testing and bug reproduction via Playwright MCP inside an isolated browser session. Use for chartered exploratory sessions, reproducing a reported bug, or first-look investigation of |
Hooks (7)
| event | matcher | runs |
|---|---|---|
| SessionStart | * | ${CLAUDE_PROJECT_DIR}/.claude/hooks/session-start.sh |
| Stop | * | ${CLAUDE_PROJECT_DIR}/.claude/hooks/stop-evidence-check.sh |
| PreToolUse | Bash|Write|Edit|MultiEdit|NotebookEdit|Read|Grep|Glob|WebFetch|Task | ${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh |
| PreToolUse | mcp__.*atlassian__(jira_)?(create|edit|add|transition|update|delete|executeWrite|executeDestructive).* | ${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh |
| PreToolUse | mcp__.*(playwright|atlassian)__.* | ${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh |
| PreToolUse | mcp__.* | ${CLAUDE_PROJECT_DIR}/.claude/hooks/pre-tool-guard.sh |
| PostToolUse | mcp__.*atlassian__(jira_)?(create|edit|add|transition|update|delete|executeWrite|executeDestructive).* | ${CLAUDE_PROJECT_DIR}/.claude/hooks/post-jira-audit.sh |
Permissions
deny (24)
Read(.env)
Read(.env.*)
Read(**/.env)
Read(**/.env.*)
Read(.env.playwright-secrets)
Read(**/*.pem)
Read(**/id_rsa*)
Read(**/id_ed25519*)
Read(**/*.auth)
Read(**/*storage-state*)
Read(**/*storage_state*)
Read(evidence/.mcp-out/**)
Read(**/evidence/.mcp-out/**)
Read(**/evidence/**/*.png)
Read(**/evidence/**/*.jpg)
Read(**/evidence/**/*.jpeg)
Read(**/evidence/**/*.gif)
Read(**/evidence/**/*.webp)
Read(**/evidence/**/*.bmp)
Read(**/evidence/**/*.webm)
Read(**/evidence/**/*.mp4)
Read(**/evidence/**/*.mov)
Read(**/evidence/**/*.zip)
Read(**/evidence/**/*.har)
ask (0)
—
allow (8)
Bash(mkdir -p evidence/:*)
Bash(shasum:*)
Bash(git status:*)
Bash(git diff:*)
Bash(git log:*)
Bash(npx playwright test:*)
Bash(npx --no-install playwright --version)
Bash(bash .claude/hooks/tests/:*)
Similar rigs
petrkindlmann/qa-skills
50 QA and test-automation skills for Claude Code, Codex, Cursor, and any Agent Skills Standard runtime.
Skill Collector 3.0k tok · · Q0
fishzjp/qa-skills
让 AI 像资深测试工程师一样工作:面向 AI Agent 的测试工程 Skill 框架——11 Skills + 共享知识库 + 类型决策矩阵(Claude Code / dsh 等 Agent 可用)
Skill Collector 1.6k tok · · Q0
MarkoDragnic/claude-sdd-template
Spec-driven Claude Code template: spec or Jira plus Figma to tested code, one task at a time, with auto and human gates.
Pragmatist 21.3k tok · · Q20
demian-v/claude-qa-mcp
A ready-to-use Claude Code MCP server starter for QA engineers: Miro, Atlassian (Jira/Confluence), TestRail, and Playwright in one config. Figma setup walkthrough included in the README.
Pragmatist 22.5k tok · · Q0