acm-ai-qe

Use when the user wants GitHub PR diff analysis for any ACM-related repo to understand what changed (components, UI, filters, field order, controllers, CRDs, webhooks) and what to test -- WITHOUT running the full JIRA-to-Polarion test case generator pipeline. Defaults to stolostron/console and kubevirt-plugin when no repo is specified (backward compatible with acm-test-case-generator). TRIGGER: analyze PR #N, what changed in this merge, test impact of this diff. DO NOT TRIGGER: user wants Polarion test case from ACM-#### end-to-end (use acm-test-case-generator); user wants only domain facts (use acm-knowledge-base).

Use this skill when the user wants a FULL end-to-end Polarion-ready ACM Console UI test case from a JIRA ticket (e.g. ACM-30459): JIRA + PRs + UI discovery + synthesis + optional live validation + writing + mandatory quality review. This is the ONLY skill for that complete path. Do NOT use acm-test-case-writer or acm-qe-code-analyzer for the same request—they are partial workflows. Do NOT use acm-knowledge-base alone; it is reference-only. TRIGGER: generate/write test case from JIRA, Polarion test case for ACM ticket, test coverage for a story. DO NOT TRIGGER: PR-only diff analysis without JIRA-to-Polarion pipeline (use acm-qe-code-analyzer); review-only (use acm-test-case-reviewer).

Use when the user wants a QUALITY REVIEW of an existing ACM Console UI test case markdown file (conventions, discovered-vs-assumed UI, MCP spot-checks, AC consistency)— without re-running JIRA/PR/UI investigation. If the user just finished acm-test-case-generator, review is already Phase 7 inside that skill—do not invoke this unless reviewing a standalone file or an out-of-band draft. TRIGGER: review this test-case.md, pre-Polarion QA, validate test steps. DO NOT TRIGGER: full test case from JIRA (acm-test-case-generator).

Use ONLY when a Polarion-style ACM Console test case markdown must be authored from ALREADY-SYNTHESIZED investigation context (JIRA + code + UI summary present in the thread or artifacts)—not for starting from a bare JIRA ID. For starting from JIRA, use acm-test-case-generator instead. TRIGGER: user explicitly asks to convert existing synthesis into test-case.md; Phase-6-style write with context already in hand. DO NOT TRIGGER: user gives only ACM-#### and expects full pipeline; user wants PR diff analysis only (acm-qe-code-analyzer); user wants quality review only (acm-test-case-reviewer).

Executes an existing ACM Console UI test case step-by-step against a live environment and produces a per-step pass/fail report with evidence (screenshots, CLI output, accessibility snapshots). Takes a test case markdown file, Polarion ID, or inline content as input. Does NOT generate or modify test cases -- only validates them. TRIGGER: validate test case, execute test case, run test case on cluster, validate RHACM4K-XXXXX on environment, dry-run test case, check if test case passes. DO NOT TRIGGER: generate test case from JIRA (use acm-test-case-generator); review test case quality/format (use acm-test-case-reviewer); verify a bug fix (use acm-bug-fix-verifier).

Enrich test failure data with AI-analyzed context -- resolve page objects, verify selector existence in product source, analyze selector change history, and fill feature knowledge gaps. Use when test failure data needs enrichment before classification analysis.

Analyze Jenkins pipeline test failures and classify each as PRODUCT_BUG, AUTOMATION_BUG, INFRASTRUCTURE, or NO_BUG. Runs a 4-stage pipeline with data gathering, cluster diagnostics, AI classification, and report generation. Use when asked to analyze a Jenkins run, classify test failures, or investigate pipeline results.

Verifies whether a known ACM bug fix has landed on a target environment. Takes a JIRA ticket and a cluster, runs a 7-phase pipeline: parallel JIRA/PR/environment investigation, three-tier fix presence check (branch reachability, build date, code presence), PR code review for fix correctness, Neo4j prerequisite gap analysis with heuristic fallback, Playwright-driven UI/API verification, and verdict with optional JIRA update. Produces BLOCKED, NOT_FIXED, PRESENT, or VERIFIED verdicts with evidence trail. TRIGGER: verify bug fix, confirm fix landed, check if fixed, is the bug fixed, verify ACM-NNNNN on cluster, test fix on environment. DO NOT TRIGGER: hunt for new bugs (use acm-bug-hunter), write test cases (use acm-test-case-generator), cluster health check (use acm-hub-health-check), PR-only code review (use acm-qe-code-analyzer).

Diagnose ACM hub cluster health using a 6-phase pipeline with 4 depth modes. Checks operators, pods, addons, subsystems, dependency chains, and known failure patterns. Use when asked to check hub health, run diagnostics, troubleshoot ACM issues, or verify cluster state.

Interface to Jenkins CI for reading build status, test results, pipeline stages, console logs, downstream trees, and (with approval) triggering builds. Use when you need Jenkins pipeline data, logs, JUnit results, or to monitor or trigger jobs. Prefer the jenkins MCP in Cursor and Claude Code when configured (same MCP set in both places). Bundled REST docs and scripts/jenkins_api.py cover portability, stdlib-only scripts, and any host without the MCP.

Build and update ACM knowledge by comparing live cluster state to the knowledge base. Discovers unknown operators, new failure patterns, dependency chains, and certificate issues. Use when asked to learn from a cluster, update knowledge, discover new components, or refresh baselines.

Interactive onboarding for the AI Systems Suite. Detects current environment state, explains the three apps, and guides MCP server setup with credential configuration. Run at first clone or to verify an existing setup.

Deep-dive investigation of individual test failures using the 12-layer diagnostic model. Traces from symptom through infrastructure layers to find root cause with evidence. Use when test failures need root-cause analysis with classification.

Classify Jenkins pipeline test failures as PRODUCT_BUG, AUTOMATION_BUG, INFRASTRUCTURE, or NO_BUG using a 5-phase AI investigation framework with 12-layer diagnostics, provably linked grouping, counterfactual validation, and multi-evidence requirements. Use when test failures need root-cause classification.

Cluster health diagnostic toolkit using a 12-layer model with 14 trap patterns, dependency chain verification, and evidence-tier framework. Use when you need to assess ACM hub cluster health, verify infrastructure state, check operator health, or diagnose subsystem issues.

Use when you need READ-ONLY ACM Console domain reference: per-area architecture, Polarion conventions, naming patterns, examples. Load the specific references/ file for the task. This skill does NOT gather JIRA, PRs, or UI via MCP and does NOT run the test-case pipeline. TRIGGER: conventions, field order, area architecture, template rules while executing another skill. DO NOT TRIGGER: user asks to generate a test case from ACM-#### (acm-test-case-generator); PR-only analysis (acm-qe-code-analyzer).

Autonomously hunts for bugs in ACM feature implementations by using test cases as a starting point for systematic 10-dimension investigation. Spawns focused subagents per dimension with an orchestrator-investigator adversarial architecture to prevent self-bias. Uses confidence-aware feedback loops inspired by the "confession" pattern. Supports all ACM areas (Console, GRC, ALC, Observability, Cluster Lifecycle, Submariner, Search, Install). Works with or without a live cluster (graceful degradation). Input: Polarion test case ID, local .md file, or inline test case content.

Remediate ACM hub cluster issues with structured approval workflow. Proposes fixes based on diagnosis findings, executes approved mutations, and verifies results. Use when asked to fix, remediate, repair, or resolve ACM cluster issues.