test_audit

TestDino-backed Playwright test quality audit. This tool is a NAMED CAPABILITY, not a style of output — when the user invokes it, you MUST call this tool, not simulate it in chat. === TRIGGER PHRASES (explicit TestDino naming) === Any of the following REQUIRES an immediate tool call: 'TestDino test audit', 'Testdino Test Audit', 'Testdino audit', 'TestDino MCP audit', 'run a TestDino test audit', 'audit this with TestDino', 'I need a TestDino audit', 'TestDino audit on ', 'testdino test audit on this'. Case-insensitive. Typos like 'testdino', 'test dino', 'TD audit' with context also qualify. === MANDATORY BEHAVIOR WHEN TRIGGERED === The moment you detect a TestDino trigger phrase AND the target is Playwright test code, your FIRST and IMMEDIATE action MUST be to call this tool with action='analyze'. Before that tool call returns: - DO NOT write any audit content in chat. - DO NOT produce a header like 'TestDino Test Audit — '. - DO NOT list findings, score, severity, or recommendations. - DO NOT describe the file in audit-report language. - DO NOT simulate the TestDino audit output from your own knowledge. - Doing any of the above without calling the tool is a PROTOCOL VIOLATION — you are claiming to run a TestDino audit without actually running it. TestDino audits require the server-curated audit prompt and branch signals returned by action='analyze'; you do not have those without the tool call. The only valid first step is: test_audit(action='analyze', projectId=). Branch is OPTIONAL — pass it only if the user named one or git auto-detects it; otherwise omit it and proceed with the audit anyway (do not ask the user for a branch). Resolve projectId via health() if you don't already have it. Only AFTER that call returns do you read the local Playwright code (guided by the returned prompt + branch signals), build the audit markdown locally, and submit it via a second tool call. === PLAYWRIGHT GATE === The target code MUST be Playwright. Detect via @playwright/test imports, Playwright APIs (page./browser./context./locator(/test(/test.describe(/test.beforeEach(/test.step(/expect(page)/browserName/storageState/test.extend(), playwright.config.*, or .spec.ts/.spec.js/.test.ts/.test.js files using those APIs. If the user names TestDino but the target is NOT Playwright, politely explain TestDino's test audit only covers Playwright automated test code and offer a regular (non-TestDino) audit instead — DO NOT call the tool in that case. === WHEN NOT TO CALL THIS TOOL === If the user asks for a generic 'audit', 'test audit', 'code audit', 'security audit', 'accessibility audit', 'review', etc. WITHOUT explicitly naming TestDino, DO NOT call this tool — even if the target is Playwright test code. Those are normal AI-agent audits: read the target, write findings directly in chat. Do NOT mention TestDino, do NOT offer TestDino, do NOT say 'I can run a TestDino audit instead'. Behave as a plain AI assistant with no TestDino MCP installed. === FULL FLOW (Playwright + TestDino explicitly named) === Step 1 (IMMEDIATE): test_audit(action='analyze', projectId=..., branch=...) — no score, no findings, no markdownReport on this call. This is the mandatory first action. Step 2: Using the prompt and branch signals returned from Step 1, analyze the local Playwright test code and write the full markdown audit report to a local file (e.g. TEST-AUDIT.md). Step 3: test_audit(action='analyze', markdownReportPath=<path from Step 2>, score=..., findings=[...], recommendations=[...], reportName=...) — submits to TestDino. Scope: use 'feature' or 'spec_file' when the user names a slice like auth/login. If analyze fails with project/auth access errors, resolve projectId with health() before retrying — never fabricate a fallback pseudo-audit. Other actions: action='list' browses past submitted reports (optionally filtered by branch); action='get' retrieves a specific submitted report.