@teevee98/synapse-hub 0.0.0-stage → 3.8.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.agents/agents/api-engineer.md +67 -0
- package/.agents/agents/failure-analyst.md +75 -0
- package/.agents/agents/live-explorer.md +70 -0
- package/.agents/agents/orchestrator.md +104 -0
- package/.agents/agents/perf-engineer.md +74 -0
- package/.agents/agents/pom-crafter.md +112 -0
- package/.agents/agents/quality-gatekeeper.md +87 -0
- package/.agents/agents/test-designer.md +73 -0
- package/.agents/agents/test-runner.md +67 -0
- package/.agents/rules/author-integrity-guard.md +33 -0
- package/.agents/rules/communication-style.md +18 -0
- package/.agents/rules/context-engineering.md +37 -0
- package/.agents/rules/git-push-protocol.md +33 -0
- package/.agents/rules/graphify.md +14 -0
- package/.agents/rules/lessons-learned.md +25 -0
- package/.agents/rules/mandatory-playwright-live.md +25 -0
- package/.agents/rules/mobile_appium_rules.md +43 -0
- package/.agents/rules/performance_testing_rules.md +35 -0
- package/.agents/rules/skill-precedence.md +29 -0
- package/.agents/rules/systematic-debugging.md +36 -0
- package/.agents/skills/actor-qa-swarm/SKILL.md +77 -0
- package/.agents/skills/actor-qa-swarm/examples/broker_sharding_sample.json +12 -0
- package/.agents/skills/agent-orchestration-multi-agent-optimize/SKILL.md +88 -0
- package/.agents/skills/agent-orchestration-multi-agent-optimize/examples/swarm_workload_config.json +15 -0
- package/.agents/skills/agentless-qa-repair/SKILL.md +69 -0
- package/.agents/skills/agentless-qa-repair/examples/healing_report_sample.json +17 -0
- package/.agents/skills/agentless-qa-repair/examples/three_phase_repair_workflow.md +40 -0
- package/.agents/skills/antigravity-test-sentinel/SKILL.md +63 -0
- package/.agents/skills/antigravity-test-sentinel/examples/dod_gate_policy.json +11 -0
- package/.agents/skills/antigravity-test-sentinel/examples/sprint_dod_enforcement.md +47 -0
- package/.agents/skills/api-network-mocking-interception/SKILL.md +142 -0
- package/.agents/skills/api-network-mocking-interception/examples/network_route_stub.json +9 -0
- package/.agents/skills/api-spec-to-e2e-matrix/SKILL.md +80 -0
- package/.agents/skills/api-spec-to-e2e-matrix/examples/sample_openapi.json +19 -0
- package/.agents/skills/architecture-decision-records/SKILL.md +64 -0
- package/.agents/skills/architecture-decision-records/examples/0001-record-architecture-decisions.md +12 -0
- package/.agents/skills/auth-session-and-cookie-manager/SKILL.md +75 -0
- package/.agents/skills/auth-session-and-cookie-manager/examples/multi_role_auth_matrix.md +70 -0
- package/.agents/skills/auth-session-and-cookie-manager/examples/storage_state_sample.json +11 -0
- package/.agents/skills/backend-load-and-stress-patterns/SKILL.md +85 -0
- package/.agents/skills/backend-load-and-stress-patterns/examples/k6_load_test_sample.js +13 -0
- package/.agents/skills/backend-load-and-stress-patterns/references/k6_load_template.js +82 -0
- package/.agents/skills/bats-testing-patterns/SKILL.md +100 -0
- package/.agents/skills/bats-testing-patterns/examples/cli_runner_test.bats +6 -0
- package/.agents/skills/bats-testing-patterns/resources/implementation-playbook.md +614 -0
- package/.agents/skills/clean-pom-crafter/SKILL.md +110 -0
- package/.agents/skills/clean-pom-crafter/examples/standard_test_case.groovy +28 -0
- package/.agents/skills/clean-pom-crafter/references/pom_patterns.groovy +138 -0
- package/.agents/skills/code-refactoring-refactor-clean/SKILL.md +64 -0
- package/.agents/skills/code-refactoring-refactor-clean/examples/refactored_page_sample.groovy +9 -0
- package/.agents/skills/code-refactoring-refactor-clean/examples/refactored_playwright_pom.ts +52 -0
- package/.agents/skills/code-refactoring-refactor-clean/resources/implementation-playbook.md +881 -0
- package/.agents/skills/consumer-driven-contract-testing/SKILL.md +147 -0
- package/.agents/skills/consumer-driven-contract-testing/examples/order_service_pact.json +45 -0
- package/.agents/skills/db-state-verifier-and-seeder/SKILL.md +123 -0
- package/.agents/skills/db-state-verifier-and-seeder/examples/db_fixture_seed.json +12 -0
- package/.agents/skills/e2e-testing-patterns/SKILL.md +103 -0
- package/.agents/skills/e2e-testing-patterns/examples/playwright_sharded_config.ts +12 -0
- package/.agents/skills/e2e-testing-patterns/resources/implementation-playbook.md +522 -0
- package/.agents/skills/event-stream-qa-orchestrator/SKILL.md +79 -0
- package/.agents/skills/event-stream-qa-orchestrator/examples/timeline_sample.jsonl +2 -0
- package/.agents/skills/failure-autopsy-and-healing/SKILL.md +66 -0
- package/.agents/skills/failure-autopsy-and-healing/examples/closed_loop_healing_walkthrough.md +48 -0
- package/.agents/skills/failure-autopsy-and-healing/examples/sample_junit_failure.xml +8 -0
- package/.agents/skills/failure-autopsy-and-healing/references/healing_matrix.md +64 -0
- package/.agents/skills/failure-autopsy-and-healing/references/qa_investigation_protocol.md +56 -0
- package/.agents/skills/failure-autopsy-and-healing/scripts/triage_failures.sh +83 -0
- package/.agents/skills/file-and-report-validator/SKILL.md +90 -0
- package/.agents/skills/file-and-report-validator/examples/sample_export_schema.json +14 -0
- package/.agents/skills/flaky-test-sentinel-and-quarantine/SKILL.md +114 -0
- package/.agents/skills/flaky-test-sentinel-and-quarantine/examples/quarantine_registry.json +11 -0
- package/.agents/skills/gherkin-to-suite/SKILL.md +103 -0
- package/.agents/skills/gherkin-to-suite/examples/checkout.feature +39 -0
- package/.agents/skills/gherkin-to-suite/examples/suite_report_sample.json +334 -0
- package/.agents/skills/github-actions-templates/SKILL.md +77 -0
- package/.agents/skills/github-actions-templates/examples/matrix_test_pipeline.yml +8 -0
- package/.agents/skills/graphify-impact-analyzer/SKILL.md +76 -0
- package/.agents/skills/graphify-impact-analyzer/examples/dependency_blast_radius.json +8 -0
- package/.agents/skills/katalon-cli-runner/SKILL.md +76 -0
- package/.agents/skills/katalon-cli-runner/examples/headless_kre_exec.sh +3 -0
- package/.agents/skills/katalon-cli-runner/references/groovy-patterns.md +209 -0
- package/.agents/skills/katalon-cli-runner/references/katalon-api.md +155 -0
- package/.agents/skills/katalon-cli-runner/references/runner_commands.md +67 -0
- package/.agents/skills/lam-script-generator/SKILL.md +123 -0
- package/.agents/skills/lam-script-generator/examples/user_story_sample.md +8 -0
- package/.agents/skills/lean-code-ladder/SKILL.md +94 -0
- package/.agents/skills/lean-code-ladder/examples/ladder_walkthrough.md +37 -0
- package/.agents/skills/lean-impact-scoreboard/SKILL.md +48 -0
- package/.agents/skills/lean-impact-scoreboard/examples/sample_benchmark_gain.md +20 -0
- package/.agents/skills/lean-impact-scoreboard/examples/sample_scoreboard.md +15 -0
- package/.agents/skills/meta-prompt-crafter/SKILL.md +107 -0
- package/.agents/skills/meta-prompt-crafter/examples/qa_prompt_template.md +35 -0
- package/.agents/skills/mipro-prompt-compiler/SKILL.md +59 -0
- package/.agents/skills/mipro-prompt-compiler/examples/compiled_module_sample.json +22 -0
- package/.agents/skills/mipro-prompt-compiler/examples/dspy_assertion_optimizer_case.md +46 -0
- package/.agents/skills/mobile-appium-som/SKILL.md +88 -0
- package/.agents/skills/mobile-appium-som/examples/BaseScreen.java +88 -0
- package/.agents/skills/mobile-appium-som/examples/LoginScreen.java +46 -0
- package/.agents/skills/mobile-appium-som/examples/TC_MobileLogin.java +43 -0
- package/.agents/skills/persistent-planning-files/SKILL.md +108 -0
- package/.agents/skills/persistent-planning-files/examples/task_plan_example.md +56 -0
- package/.agents/skills/playwright-live-interaction/SKILL.md +131 -0
- package/.agents/skills/playwright-live-interaction/examples/live_probe_script.js +5 -0
- package/.agents/skills/playwright-live-interaction/references/advanced_superpowers.md +160 -0
- package/.agents/skills/playwright-live-interaction/references/compact_accessibility_snapshots.md +75 -0
- package/.agents/skills/playwright-live-interaction/references/recipes.md +388 -0
- package/.agents/skills/playwright-live-interaction/references/swarm_orchestration.md +134 -0
- package/.agents/skills/playwright-webapp-testing/SKILL.md +93 -0
- package/.agents/skills/playwright-webapp-testing/examples/server_lifecycle_playwright.py +63 -0
- package/.agents/skills/qa-aci-interface/SKILL.md +67 -0
- package/.agents/skills/qa-aci-interface/examples/surgical_code_window_walkthrough.md +40 -0
- package/.agents/skills/qa-aci-interface/examples/surgical_edit_sample.json +12 -0
- package/.agents/skills/qa-memory-os/SKILL.md +80 -0
- package/.agents/skills/qa-memory-os/examples/archival_memory_sample.json +21 -0
- package/.agents/skills/rbt-manual-testing/SKILL.md +137 -0
- package/.agents/skills/rbt-manual-testing/examples/checkout_rbt_manual_suite.md +25 -0
- package/.agents/skills/rbt-manual-testing/references/field_validation_checklist.md +29 -0
- package/.agents/skills/repo-quality-audit/SKILL.md +68 -0
- package/.agents/skills/repo-quality-audit/examples/sample_audit_walkthrough.md +15 -0
- package/.agents/skills/repo-quality-audit/examples/sample_repo_audit.md +33 -0
- package/.agents/skills/review-spec/SKILL.md +60 -0
- package/.agents/skills/review-spec/examples/pocock_grilling_decision_rubric.md +40 -0
- package/.agents/skills/review-spec/examples/sample_spec_review.md +33 -0
- package/.agents/skills/security-dast-sentinel/SKILL.md +119 -0
- package/.agents/skills/security-dast-sentinel/examples/dast_scan_config.json +10 -0
- package/.agents/skills/security-dast-sentinel/references/AI-AND-LLM.md +113 -0
- package/.agents/skills/security-dast-sentinel/references/ATTACK-CLASSES.md +89 -0
- package/.agents/skills/security-dast-sentinel/references/dast_recipes.ts +33 -0
- package/.agents/skills/self-healing-locator/SKILL.md +86 -0
- package/.agents/skills/self-healing-locator/examples/aom_diff_sample.json +16 -0
- package/.agents/skills/senior-code-review/SKILL.md +108 -0
- package/.agents/skills/senior-code-review/examples/sample_review_walkthrough.md +29 -0
- package/.agents/skills/skillspector-security-guard/SKILL.md +57 -0
- package/.agents/skills/skillspector-security-guard/examples/audit_threat_vectors_report.json +35 -0
- package/.agents/skills/skillspector-security-guard/examples/safe_mcp_config.json +12 -0
- package/.agents/skills/surgical-coding-principles/SKILL.md +89 -0
- package/.agents/skills/surgical-coding-principles/examples/surgical_vs_scope_creep.md +46 -0
- package/.agents/skills/synthetic-test-data-generator/SKILL.md +135 -0
- package/.agents/skills/synthetic-test-data-generator/examples/idempotent_data_recipe.json +9 -0
- package/.agents/skills/systematic-debugging/SKILL.md +100 -0
- package/.agents/skills/systematic-debugging/examples/four_phase_case.md +44 -0
- package/.agents/skills/technical-debt-harvester/SKILL.md +70 -0
- package/.agents/skills/technical-debt-harvester/examples/sample_debt_ledger.md +25 -0
- package/.agents/skills/technical-debt-harvester/examples/sample_ledger_output.md +20 -0
- package/.agents/skills/test-driven-development/SKILL.md +114 -0
- package/.agents/skills/test-driven-development/examples/red_green_refactor_qa.md +78 -0
- package/.agents/skills/token-and-context-optimizer/SKILL.md +70 -0
- package/.agents/skills/token-and-context-optimizer/examples/distilled_log_sample.txt +3 -0
- package/.agents/skills/token-and-context-optimizer/references/context_engineering_guide.md +49 -0
- package/.agents/skills/tot-exploratory-testing/SKILL.md +59 -0
- package/.agents/skills/tot-exploratory-testing/examples/payment_gateway_tot_tree.json +62 -0
- package/.agents/skills/tot-exploratory-testing/examples/tot_graph_sample.json +35 -0
- package/.agents/skills/verification-before-completion/SKILL.md +100 -0
- package/.agents/skills/verification-before-completion/examples/evidence_log_example.md +46 -0
- package/.agents/skills/visual-regression-sentinel/SKILL.md +82 -0
- package/.agents/skills/visual-regression-sentinel/examples/visual_mask_rules.json +9 -0
- package/.agents/skills/wcag-audit-patterns/SKILL.md +77 -0
- package/.agents/skills/wcag-audit-patterns/examples/wcag_compliance_checklist.json +10 -0
- package/.agents/skills/wcag-audit-patterns/resources/implementation-playbook.md +541 -0
- package/.agents/skills/web-agent-gym-benchmarking/SKILL.md +62 -0
- package/.agents/skills/web-agent-gym-benchmarking/examples/benchmark_report_sample.json +22 -0
- package/.agents/skills/web-agent-gym-benchmarking/examples/enterprise_tasks_suite.json +38 -0
- package/.agents/skills/web-vitals-and-performance-profiler/SKILL.md +65 -0
- package/.agents/skills/web-vitals-and-performance-profiler/examples/cwv_budget_policy.json +6 -0
- package/.agents/skills/web-vitals-and-performance-profiler/examples/playwright_cwv_profiler_fixture.ts +118 -0
- package/.agents/templates/completion_report.md +57 -0
- package/.agents/workflows/analyze_requirement_document.md +18 -0
- package/.agents/workflows/export_testcases_excel.md +32 -0
- package/.agents/workflows/generate_combinatorial_test_data.md +28 -0
- package/.agents/workflows/generate_cross_module_test_plan.md +37 -0
- package/.agents/workflows/generate_manual_testcases_rbt.md +56 -0
- package/.agents/workflows/generate_requirements_from_website.md +25 -0
- package/.agents/workflows/generate_testcases_from_requirements.md +21 -0
- package/.agents/workflows/graphify.md +10 -0
- package/.agents/workflows/test-impact.md +15 -0
- package/.agents/workflows/verify.md +34 -0
- package/.cursor/rules/agents.mdc +16 -0
- package/.cursor/rules/language.mdc +7 -0
- package/.cursor/rules/synapse-qa.mdc +24 -0
- package/.cursorrules +27 -0
- package/.dockerignore +53 -0
- package/.pre-commit-hooks.yaml +14 -0
- package/.windsurfrules +26 -0
- package/AGENTS.md +86 -0
- package/CLAUDE.md +54 -0
- package/Dockerfile +81 -0
- package/LICENSE +52 -0
- package/NOTICE +264 -0
- package/README.en.md +172 -0
- package/README.md +171 -2
- package/action.yml +51 -0
- package/agentskills.json +16 -0
- package/bin/agent-qa.js +138 -0
- package/docker-compose.yml +74 -0
- package/docs/ARCHITECTURE.en.md +197 -0
- package/docs/ARCHITECTURE.md +198 -0
- package/docs/CI_CD_INTEGRATION.en.md +170 -0
- package/docs/CI_CD_INTEGRATION.md +170 -0
- package/docs/CLI_REFERENCE.en.md +175 -0
- package/docs/CLI_REFERENCE.md +177 -0
- package/docs/CONTRIBUTING.en.md +49 -0
- package/docs/CONTRIBUTING.md +49 -0
- package/docs/PERFORMANCE_AUDIT_REPORT.md +784 -0
- package/docs/PLUGIN_DEVELOPMENT_GUIDE.md +127 -0
- package/docs/QUALITY_GATES.en.md +121 -0
- package/docs/QUALITY_GATES.md +121 -0
- package/docs/TOKEN_QUOTA_PLAYBOOK.md +41 -0
- package/docs/TOOL_CATALOG.md +129 -0
- package/docs/adopted-ecosystem-tools.md +96 -0
- package/docs/research/EXTERNAL_AGENT_SKILLS_SURVEY.md +1062 -0
- package/docs/research/agent_skill_training_analysis.md +1359 -0
- package/docs/schemas/synapse.config.schema.json +87 -0
- package/package.json +94 -4
- package/plugin.json +31 -0
- package/scripts/generate_skill_examples.py +244 -0
- package/scripts/install.ps1 +112 -0
- package/scripts/install_global_plugin.ps1 +94 -0
- package/scripts/install_global_plugin.sh +49 -0
- package/scripts/lint_skills.ps1 +57 -0
- package/scripts/package_dist.py +337 -0
- package/scripts/setup.sh +59 -0
- package/scripts/update_hub.ps1 +19 -0
- package/scripts/update_hub.sh +15 -0
- package/skills_manifest.json +854 -0
- package/synapse.config.json +32 -0
- package/templates/PROJECT_AI_GUIDE_TEMPLATE.md +111 -0
- package/templates/PROJECT_STATE_TEMPLATE.md +20 -0
- package/templates/ai_bridges/CLAUDE.md +54 -0
- package/templates/ai_bridges/copilot-instructions.md +20 -0
- package/templates/ai_bridges/cursor_synapse_qa.mdc +24 -0
- package/templates/ai_bridges/windsurfrules +26 -0
- package/templates/ci/github-actions-synapse.yml +95 -0
- package/templates/ci/gitlab-ci-synapse.yml +16 -0
- package/templates/ui/index.html +1069 -0
- package/templates/welcome.html.tmpl +2079 -0
- package/tools/.infer_baseline.json +30 -0
- package/tools/TOOLS_RUNBOOK.md +1553 -0
- package/tools/TOOL_CATALOG.json +1769 -0
- package/tools/__init__.py +43 -0
- package/tools/_openapi_parser.py +158 -0
- package/tools/agent_canary.py +525 -0
- package/tools/agentless_patcher.py +417 -0
- package/tools/analyze_project.py +258 -0
- package/tools/api_drift_sentinel.py +208 -0
- package/tools/api_fuzzer.py +762 -0
- package/tools/author_integrity.py +184 -0
- package/tools/auto_heal.py +271 -0
- package/tools/auto_heal_reflexion.py +532 -0
- package/tools/autogen_qa_broker.py +844 -0
- package/tools/benchmark_swarm_context.py +394 -0
- package/tools/browser_helpers.js +746 -0
- package/tools/browsergym_eval.py +236 -0
- package/tools/bug_reporter.py +252 -0
- package/tools/bump_version.py +423 -0
- package/tools/cache_resolver.py +1214 -0
- package/tools/check_env.py +426 -0
- package/tools/check_version_sync.py +256 -0
- package/tools/ci_healing_hook.py +126 -0
- package/tools/cli.py +1598 -0
- package/tools/config_decoupler.py +253 -0
- package/tools/contract_validator.py +141 -0
- package/tools/core/__init__.py +70 -0
- package/tools/core/adapters.py +431 -0
- package/tools/core/circuit_breaker.py +263 -0
- package/tools/core/cli_runtime.py +66 -0
- package/tools/core/file_lock.py +159 -0
- package/tools/core/i18n.py +394 -0
- package/tools/core/memory_bus.py +905 -0
- package/tools/core/process_supervisor.py +332 -0
- package/tools/core/telemetry.py +329 -0
- package/tools/count_tokens.py +1267 -0
- package/tools/create_mock_docx.py +104 -0
- package/tools/dashboard_generator.py +2419 -0
- package/tools/db_sandbox.py +675 -0
- package/tools/debt_harvester.py +237 -0
- package/tools/distill_log.py +161 -0
- package/tools/dspy_qa_signatures.py +356 -0
- package/tools/e2e_flow_prober.py +132 -0
- package/tools/export_postman_collection.py +945 -0
- package/tools/flaky_detector.py +389 -0
- package/tools/fsm_enforcer.py +450 -0
- package/tools/fsm_middleware.py +401 -0
- package/tools/fuzz_payloads.py +197 -0
- package/tools/generate_allure_report.py +88 -0
- package/tools/generate_api_matrix.py +193 -0
- package/tools/generate_jmx_plan.py +1322 -0
- package/tools/generate_sbom.py +252 -0
- package/tools/generate_tc_from_story.py +203 -0
- package/tools/generate_test_reports.py +133 -0
- package/tools/generate_tool_catalog.py +482 -0
- package/tools/gherkin_suite.py +758 -0
- package/tools/git_pre_commit_guard.py +298 -0
- package/tools/goldfish_review.py +366 -0
- package/tools/heal/__init__.py +28 -0
- package/tools/heal_pr_automator.py +220 -0
- package/tools/heuristic_locator_matcher.py +351 -0
- package/tools/hub_publisher.py +623 -0
- package/tools/humanize_linter.py +438 -0
- package/tools/infer_linter.py +977 -0
- package/tools/init_config.py +307 -0
- package/tools/interview_story.py +116 -0
- package/tools/jira_sync.py +566 -0
- package/tools/jtl_analyzer.py +474 -0
- package/tools/lint/__init__.py +30 -0
- package/tools/lint_ai_docs.py +374 -0
- package/tools/lookup_tc.py +165 -0
- package/tools/lookup_xpath.py +167 -0
- package/tools/matrix_sharder.py +148 -0
- package/tools/mcp/__init__.py +18 -0
- package/tools/mcp_docx_server.ps1 +13 -0
- package/tools/mcp_docx_server.py +851 -0
- package/tools/mcp_excel_server.ps1 +13 -0
- package/tools/mcp_excel_server.py +441 -0
- package/tools/mcp_server.py +724 -0
- package/tools/md_to_xlsx.py +327 -0
- package/tools/micro_mutator.py +887 -0
- package/tools/mipro_assertion_tuner.py +210 -0
- package/tools/network_mock_shield.py +311 -0
- package/tools/newman_analyzer.py +384 -0
- package/tools/omni_emitter.py +579 -0
- package/tools/omni_parser.py +450 -0
- package/tools/omni_transpiler.py +168 -0
- package/tools/openxml_guard.py +438 -0
- package/tools/output_formatter.py +182 -0
- package/tools/package_release.py +325 -0
- package/tools/perf/__init__.py +24 -0
- package/tools/pinpoint_failure.py +190 -0
- package/tools/project_state.py +171 -0
- package/tools/qa_event_stream.py +358 -0
- package/tools/repo_map.py +346 -0
- package/tools/report/__init__.py +32 -0
- package/tools/reversible_sandbox.py +344 -0
- package/tools/safety_firewall.py +437 -0
- package/tools/security_coverage_ledger.py +538 -0
- package/tools/security_scanner.py +229 -0
- package/tools/senior_code_review.py +393 -0
- package/tools/setup_project.py +770 -0
- package/tools/skill_collision_guard.py +304 -0
- package/tools/skill_profiles.py +217 -0
- package/tools/skill_vault.py +481 -0
- package/tools/skillspector.py +1278 -0
- package/tools/smart_visual_masker.py +132 -0
- package/tools/spec_drift_sentinel.py +438 -0
- package/tools/swe_qa_aci.py +369 -0
- package/tools/synapse_observer.py +485 -0
- package/tools/synapse_qa_bench.py +190 -0
- package/tools/sync_agent_rules.py +274 -0
- package/tools/telemetry_correlator.py +397 -0
- package/tools/template_excel_exporter.py +514 -0
- package/tools/templates/jmeter-distributed/README.md +54 -0
- package/tools/templates/jmeter-distributed/docker-compose.yml +83 -0
- package/tools/test_impact.py +419 -0
- package/tools/test_memory_os.py +462 -0
- package/tools/textgrad_mutation_optimizer.py +237 -0
- package/tools/tms_sync_bridge.py +834 -0
- package/tools/tot_exploratory_tester.py +239 -0
- package/tools/trace_data_flow.py +240 -0
- package/tools/traffic_session_converter.py +256 -0
- package/tools/trend_analytics.py +314 -0
- package/tools/ui_server.py +401 -0
- package/tools/universal_ir.py +488 -0
- package/tools/update_sentinel.py +383 -0
- package/tools/validate_groovy_scripts.py +173 -0
- package/tools/verify_all.py +1259 -0
- package/tools/visual_diff_canvas.py +143 -0
- package/tools/visual_som_helper.py +550 -0
- package/tools/watch_mode.py +308 -0
- package/tools/wcag_audit.py +239 -0
- package/tools/welcome_page.py +290 -0
|
@@ -0,0 +1,784 @@
|
|
|
1
|
+
# Enterprise Performance Testing Architecture Audit & Benchmark Report
|
|
2
|
+
|
|
3
|
+
**Document ID:** SYNAPSE-AUDIT-PERF-2026-M2
|
|
4
|
+
**Evaluation Standard:** Adversarial Enterprise SRE Benchmark (Scale 1.0 – 10.0)
|
|
5
|
+
**Lead Auditor:** Principal Performance Test Architect & Staff SRE (15+ Years Distributed Systems Experience)
|
|
6
|
+
**Target Repository:** Synapse Enterprise QA Hub (`/Users/phungcao/Downloads/Agent Skill`)
|
|
7
|
+
**Audit Date:** 2026-11-09
|
|
8
|
+
**Status:** COMPLETE / OFFICIAL AUDIT REPORT
|
|
9
|
+
|
|
10
|
+
---
|
|
11
|
+
|
|
12
|
+
## 1. Executive Summary & Composite Score Card
|
|
13
|
+
|
|
14
|
+
### 1.1 Executive Overview
|
|
15
|
+
This adversarial performance audit evaluates the architecture, tooling, agent behaviors, and verification rules across the Synapse Hub ecosystem. The evaluation scrutinizes whether the platform provides an autonomous, enterprise-grade performance engineering framework capable of preventing latency regressions, modeling realistic user concurrency, correlating client degradation with infrastructure bottlenecks, and enforcing non-functional quality gates.
|
|
16
|
+
|
|
17
|
+
The evaluation covers five core performance pillars:
|
|
18
|
+
1. **Agent Persona, Philosophy & Safety Invariants**
|
|
19
|
+
2. **Workload Modeling & Multi-Engine Script Generation**
|
|
20
|
+
3. **CI/CD Integration & SLA Quality Gatekeeper**
|
|
21
|
+
4. **Full-Stack APM & Hardware Telemetry RCA**
|
|
22
|
+
5. **Frontend Web Vitals & Client Experience**
|
|
23
|
+
|
|
24
|
+
### 1.2 Enterprise Performance Score Card
|
|
25
|
+
|
|
26
|
+
| Performance Pillar | Score | Status | Key Architectural Finding |
|
|
27
|
+
| :--- | :---: | :---: | :--- |
|
|
28
|
+
| **Pillar 1: Agent Persona, Philosophy & Safety Invariants** | **7.5 / 10** | Strong Foundation | Exemplary Addy Osmani anti-rationalization guard in `perf-engineer.md`; linear interpolation percentile engine; contradicted by single-token auth reuse in k6 template and 0ms default think-time. |
|
|
29
|
+
| **Pillar 2: Workload Modeling & Multi-Engine Script Generation** | **7.0 / 10** | Advanced Generator | Dual-engine synthesis (JMeter 5.x XML + Locust code-as-code); OpenAPI/Postman/HAR ingestion; 360° correlation (CSRF, JSON `Match No. 0`, HMAC-SHA256); lacks k6 compilation CLI; closed-loop concurrency limits. |
|
|
30
|
+
| **Pillar 3: CI/CD Integration & SLA Quality Gatekeeper** | **6.0 / 10** | Deficient | Deterministic sub-second JTL exit-code gating on P95/P99/error rate; completely disconnected from central `verify_all.py` (10 Quality Gates); k6 and Locust metric ingestion unsupported. |
|
|
31
|
+
| **Pillar 4: Full-Stack APM & Hardware Telemetry RCA** | **5.5 / 10** | PoC Grade | Standardized Pearson correlation ($r$) and heuristic RCA; strictly coupled to offline static CSVs; vulnerable to clock skew; superficial scalar heuristics; mock OpenTelemetry JSON output. |
|
|
32
|
+
| **Pillar 5: Frontend Web Vitals & Client Experience** | **3.5 / 10** | Critically Deficient | Obsolete Navigation Timing v1 in `SKILL.md`; INP completely missing in code; unthrottled runs; orphaned profiler fixture; unmounted Chrome DevTools MCP; Gate 5 blacklists heap dump artifacts. |
|
|
33
|
+
| **Composite Ecosystem Performance Score** | **5.9 / 10** | **Staging Ready** | **Arithmetic Mean: 29.5 / 50 = 5.9 / 10 (Weighted / Rounded: 6.0 / 10)**. Robust backend test generation and JTL evaluation, crippled by frontend obsolescence, disconnected CI gates, and offline APM. |
|
|
34
|
+
|
|
35
|
+
```
|
|
36
|
+
ENTERPRISE PERFORMANCE MATURITY RADAR
|
|
37
|
+
Pillar 1
|
|
38
|
+
(Persona & Safety)
|
|
39
|
+
10 | 7.5
|
|
40
|
+
|
|
|
41
|
+
| *
|
|
42
|
+
Pillar 5 | Pillar 2
|
|
43
|
+
(Web Vitals & Client) 3.5 * | * 7.0 (Workload & Engines)
|
|
44
|
+
\ | /
|
|
45
|
+
\ | /
|
|
46
|
+
\ | /
|
|
47
|
+
\ | /
|
|
48
|
+
\ | /
|
|
49
|
+
\ | /
|
|
50
|
+
\|/
|
|
51
|
+
-----------------------+-----------------------
|
|
52
|
+
/|\
|
|
53
|
+
/ | \
|
|
54
|
+
/ | \
|
|
55
|
+
/ | \
|
|
56
|
+
/ | \
|
|
57
|
+
/ | \
|
|
58
|
+
Pillar 4 * | * Pillar 3
|
|
59
|
+
(APM & RCA) 5.5 | 6.0 (CI/CD Gating)
|
|
60
|
+
|
|
|
61
|
+
0
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
---
|
|
65
|
+
|
|
66
|
+
## 2. Definitive Deployment Readiness Verdict
|
|
67
|
+
|
|
68
|
+
### 2.1 Formal Certification Verdict
|
|
69
|
+
|
|
70
|
+
> ### 🛑 **OFFICIAL CERTIFICATION: STAGING READY (Tier 2)**
|
|
71
|
+
> **Classification:** Conditionally Ready for Guided Pre-Production & Non-Critical Staging.
|
|
72
|
+
> **Production Status:** **STRICTLY NOT READY FOR AUTONOMOUS MISSION-CRITICAL PRODUCTION USE.**
|
|
73
|
+
|
|
74
|
+
```
|
|
75
|
+
+----------------------------------------------------------------------------------------------------+
|
|
76
|
+
| TIER READINESS CLASSIFICATION |
|
|
77
|
+
+------------------------------+--------------------+------------------------------------------------+
|
|
78
|
+
| Deployment Tier | Certification | Operational Condition |
|
|
79
|
+
+------------------------------+--------------------+------------------------------------------------+
|
|
80
|
+
| Tier 1: Proof of Concept | ✅ CERTIFIED | Fully viable for local benchmarking & scripts. |
|
|
81
|
+
| Tier 2: Staging / UAT | ⚠️ CONDITIONALLY | Requires human SRE verification of parameters. |
|
|
82
|
+
| Tier 3: Mission-Critical Prod| ❌ REJECTED | Blocked by 6 Critical Production Vulnerabilities|
|
|
83
|
+
+------------------------------+--------------------+------------------------------------------------+
|
|
84
|
+
```
|
|
85
|
+
|
|
86
|
+
### 2.2 Technical Invalidation: Why the System Fails Production Certification
|
|
87
|
+
Deploying this performance testing framework into mission-critical production pipelines without human oversight will lead to false passes, undetected latency spikes, or infrastructure instability. The certification denial is grounded in six fatal deficiencies:
|
|
88
|
+
|
|
89
|
+
1. **Complete Disconnection from Mandatory Quality Gatekeeper (`tools/verify_all.py`):**
|
|
90
|
+
The central Definition of Done (`verify_all.py`), which controls Stage 5 acceptance in `AGENTS.md`, evaluates Gates 0 through 9 (attribution, BOM, path portability, clean POM, docs drift, cleanup blacklist, credentials, static defect AST, micro-mutation, and skill collision). **Zero performance or SLA gates exist in `verify_all.py`.** A PR introducing severe tail-latency degradation or elevated 5xx error rates will pass the 10 gates with zero warnings and merge into production.
|
|
91
|
+
2. **Absence of Autonomous `abortOnFail` Kill-Switch in Generated JMeter Suites:**
|
|
92
|
+
While the k6 documentation specifies error rate thresholds, `tools/generate_jmx_plan.py` emits thread groups with `<stringProp name="ThreadGroup.on_sample_error">continue</stringProp>` and injects no AutoStop listeners or JSR223 aggregate failure monitors. Under catastrophic gateway degradation (e.g. cascading 503s), the load generator will continuously hammer degraded upstream microservices until test completion, triggering cascading outages in shared environments.
|
|
93
|
+
3. **Single-Token Auth Anti-Pattern in Canonical k6 Template:**
|
|
94
|
+
In `.agents/skills/backend-load-and-stress-patterns/SKILL.md` (lines 71–88), the reference script executes authentication inside `setup()` and passes `data.token` to all 200 Virtual Users. This directly violates `.agents/agents/perf-engineer.md` line 68. Under test conditions, all VUs hit memory cache layers for a single session, masking database session lookup bottlenecks, token table query performance, and user row-level locking.
|
|
95
|
+
4. **Active Quality Gate Conflict Blocking Memory Leak & Heap Analysis:**
|
|
96
|
+
Gate 5 of `tools/verify_all.py` (line 433) explicitly blacklists temporary files, heap dumps, and local caches, scanning for `.heapsnapshot` and `.hprof` artifacts. Any automated CI/CD job that captures heap snapshots to detect memory leaks will fail Gate 5. Furthermore, the external `chrome-devtools-plugin` required for heap comparison is unmounted in the active runtime.
|
|
97
|
+
5. **Obsolescence and Measurement Flaws in Frontend Web Vitals:**
|
|
98
|
+
`.agents/skills/web-vitals-and-performance-profiler/SKILL.md` relies on `window.performance.timing` (Navigation Timing Level 1, deprecated by W3C in 2019). The fixture completely lacks INP (Interaction to Next Paint) implementation, runs unthrottled (ignoring Fast 3G policies), and only extracts LCP/CLS during fixture teardown after user interactions have already terminated LCP candidate dispatch.
|
|
99
|
+
6. **Closed-Loop Concurrency & Coordinated Omission Vulnerability:**
|
|
100
|
+
JMeter test plans generated by `tools/generate_jmx_plan.py` rely exclusively on closed-concurrency thread groups (`ThreadGroup.num_threads`). Under server queuing, transaction response times increase, threads spend longer waiting for responses, and throughput drops. This introduces Coordinated Omission, systematically underreporting P95 and P99 tail latency.
|
|
101
|
+
|
|
102
|
+
---
|
|
103
|
+
|
|
104
|
+
## 3. In-Depth Technical Analysis Across the 5 Performance Pillars
|
|
105
|
+
|
|
106
|
+
---
|
|
107
|
+
|
|
108
|
+
### 3.1 Pillar 1: Agent Persona, Philosophy & Safety Invariants
|
|
109
|
+
**Adversarial Score: 7.5 / 10**
|
|
110
|
+
|
|
111
|
+
```
|
|
112
|
+
+----------------------------------------------------------------------------------------------------+
|
|
113
|
+
| PILLAR 1 EVALUATION: AGENT PERSONA, PHILOSOPHY & SAFETY INVARIANTS |
|
|
114
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
115
|
+
| Sub-Requirement | Status | Technical Assessment |
|
|
116
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
117
|
+
| Tail Latency vs Average Mindset | PASS | Strict P90/P95/P99 enforcement; linear interpolation|
|
|
118
|
+
| Anti-Cheating: Auth Concurrency | FAIL | Canonical k6 template shares 1 token across 200 VUs |
|
|
119
|
+
| Think-Time Discipline | FAIL | Generator defaults to 0ms think-time (socket hammer)|
|
|
120
|
+
| Idempotent Teardown & Data Safety | PASS | WAL journaling in reversible_sandbox.py |
|
|
121
|
+
| Safety Kill-Switch (abortOnFail) | FAIL | JMeter plans emit on_sample_error="continue" |
|
|
122
|
+
| Environment Boundary Protection | FAIL | generate_jmx_plan accepts arbitrary prod URLs |
|
|
123
|
+
| Progressive Load Stage Modeling | PASS | Smoke -> Load -> Stress ladder in staged workloads |
|
|
124
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
125
|
+
```
|
|
126
|
+
|
|
127
|
+
#### Detailed SRE Technical Critique
|
|
128
|
+
|
|
129
|
+
##### 1. Persona & Anti-Rationalization Protocol
|
|
130
|
+
The agent definition in `.agents/agents/perf-engineer.md` (lines 61–75) establishes an exemplary SRE mindset, incorporating the Addy Osmani Anti-Rationalization Guard:
|
|
131
|
+
|
|
132
|
+
```markdown
|
|
133
|
+
| Engineer Rationalization | Staff SRE Ruling & Mandatory Action |
|
|
134
|
+
|---|---|
|
|
135
|
+
| *"The average response time is 350ms, so the performance test passed."* | **STRICTLY FORBIDDEN:** Average response time hides tail latency micro-spikes. Always enforce P90, P95, and P99 SLA percentiles on steady-state traffic. |
|
|
136
|
+
| *"The error rate was only 2.5% under peak stress, which is good enough."* | **STRICTLY FORBIDDEN:** Maximum permissible error rate is < 1.0% (and 0% for financial/auth paths). 5xx errors indicate ungraceful degradation. |
|
|
137
|
+
| *"I removed simulated user think-time to reach the target RPS faster."* | **STRICTLY FORBIDDEN:** Zero think-time generates unrealistic socket hammer traffic, distorting socket pools and artificially crashing gateways. Maintain 1s–5s realistic think-time. |
|
|
138
|
+
| *"Reusing a single authenticated token for all 500 VUs is sufficient for API testing."* | **STRICTLY FORBIDDEN:** Shared token tests hit server in-memory token cache and mask DB user session lookup bottlenecks and row locks. |
|
|
139
|
+
| *"We can skip teardown cleanup because load test data is marked with 'PERF_'."* | **STRICTLY FORBIDDEN:** Residual test data bloats table indexes and degrades future test runs. Always execute idempotent teardown via tools/reversible_sandbox.py. |
|
|
140
|
+
```
|
|
141
|
+
|
|
142
|
+
This is reinforced by `.agents/rules/performance_testing_rules.md` (line 35), mandating latency percentile analysis over averages.
|
|
143
|
+
|
|
144
|
+
##### 2. Mathematical Rigor in Tail Latency Estimation
|
|
145
|
+
In `tools/jtl_analyzer.py` (lines 48–59), percentile calculations implement exact linear interpolation between data points:
|
|
146
|
+
|
|
147
|
+
$$\text{Index } k = (N - 1) \times \frac{P}{100}, \quad f = \lfloor k \rfloor, \quad c = \lceil k \rceil$$
|
|
148
|
+
|
|
149
|
+
$$P_k = V_f \times (c - k) + V_c \times (k - f)$$
|
|
150
|
+
|
|
151
|
+
```python
|
|
152
|
+
def calculate_percentile(sorted_values: List[float], percentile: float) -> float:
|
|
153
|
+
if not sorted_values: return 0.0
|
|
154
|
+
k = (len(sorted_values) - 1) * (percentile / 100.0)
|
|
155
|
+
f = math.floor(k); c = math.ceil(k)
|
|
156
|
+
if f == c: return float(sorted_values[int(k)])
|
|
157
|
+
return float(sorted_values[int(f)] * (c - k) + sorted_values[int(c)] * (k - f))
|
|
158
|
+
```
|
|
159
|
+
|
|
160
|
+
This prevents nearest-rank quantization errors on small sample sizes. However, `parse_jtl()` (lines 394–409) ingests samples across the entire execution duration from start to finish without steady-state windowing. Ramp-up cold starts and cool-down drainage are conflated with steady-state load, distorting SLA percentile accuracy.
|
|
161
|
+
|
|
162
|
+
##### 3. Specification vs Implementation Contradiction: Single-Token Auth Sharing
|
|
163
|
+
While `perf-engineer.md` line 68 strictly forbids single-token reuse, the canonical k6 template in `.agents/skills/backend-load-and-stress-patterns/SKILL.md` (lines 71–88) contradicts this rule:
|
|
164
|
+
|
|
165
|
+
```javascript
|
|
166
|
+
// Citation: .agents/skills/backend-load-and-stress-patterns/SKILL.md (Lines 71-88)
|
|
167
|
+
export function setup() {
|
|
168
|
+
const loginRes = http.post(`${__ENV.BASE_URL}/api/v1/auth/login`, JSON.stringify({
|
|
169
|
+
username: __ENV.PERF_USER || 'perf_test_runner',
|
|
170
|
+
password: __ENV.PERF_PASSWORD || 'secret'
|
|
171
|
+
}), { headers: { 'Content-Type': 'application/json' } });
|
|
172
|
+
return { token: loginRes.json('token') }; // Single token extracted once
|
|
173
|
+
}
|
|
174
|
+
|
|
175
|
+
export default function (data) {
|
|
176
|
+
const params = {
|
|
177
|
+
headers: {
|
|
178
|
+
'Authorization': `Bearer ${data.token}`, // Reused across all 200 VUs
|
|
179
|
+
'Content-Type': 'application/json',
|
|
180
|
+
},
|
|
181
|
+
};
|
|
182
|
+
const res = http.get(`${__ENV.BASE_URL}/api/v1/orders`, params);
|
|
183
|
+
check(res, { 'status is 200': (r) => r.status === 200 });
|
|
184
|
+
}
|
|
185
|
+
```
|
|
186
|
+
|
|
187
|
+
This pattern eliminates database authentication load, produces artificial Redis token cache hits, and invalidates concurrency findings.
|
|
188
|
+
|
|
189
|
+
##### 4. Think-Time Degradation into Socket Hammering
|
|
190
|
+
In `tools/generate_jmx_plan.py` (lines 1123–1124), `--think-time-min` and `--think-time-max` default to `0`:
|
|
191
|
+
|
|
192
|
+
```python
|
|
193
|
+
parser.add_argument("--think-time-min", type=int, default=0, help="Minimum simulated user think time in ms.")
|
|
194
|
+
parser.add_argument("--think-time-max", type=int, default=0, help="Maximum simulated user think time in ms.")
|
|
195
|
+
```
|
|
196
|
+
|
|
197
|
+
Line 482 evaluates `if think_time_max > think_time_min and think_time_max > 0:`. Under default CLI execution, this evaluates to `False`, omitting timers from the generated test plan. Unless flags are explicitly provided, the tool synthesizes plans that hammer the target system with zero think time, in direct violation of `perf-engineer.md` line 67.
|
|
198
|
+
|
|
199
|
+
##### 5. Safety Guardrails & Environment Boundary Vulnerabilities
|
|
200
|
+
- **Kill-Switch Gap:** In `tools/generate_jmx_plan.py` (line 415), the generated thread group sets:
|
|
201
|
+
```xml
|
|
202
|
+
<stringProp name="ThreadGroup.on_sample_error">continue</stringProp>
|
|
203
|
+
```
|
|
204
|
+
No AutoStop listeners or JSR223 error rate monitors are injected. If the target system encounters cascading 5xx failures, JMeter will continue load generation rather than failing safe.
|
|
205
|
+
- **Missing Production Firewall:** While `tools/api_fuzzer.py` (lines 114–129) enforces `_is_safe_non_prod_host()` to reject production hostnames, `tools/generate_jmx_plan.py` (lines 1210–1218) accepts arbitrary production URLs without safety firewall checks:
|
|
206
|
+
```bash
|
|
207
|
+
python3 tools/generate_jmx_plan.py --base-url "https://production.bank.com" --sample -o /tmp/test_prod.jmx
|
|
208
|
+
# Generates plan against production with zero warning or interception
|
|
209
|
+
```
|
|
210
|
+
|
|
211
|
+
---
|
|
212
|
+
|
|
213
|
+
### 3.2 Pillar 2: Workload Modeling & Multi-Engine Script Generation
|
|
214
|
+
**Adversarial Score: 7.0 / 10**
|
|
215
|
+
|
|
216
|
+
```
|
|
217
|
+
+----------------------------------------------------------------------------------------------------+
|
|
218
|
+
| PILLAR 2 EVALUATION: WORKLOAD MODELING & MULTI-ENGINE SCRIPT GENERATION |
|
|
219
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
220
|
+
| Sub-Requirement | Status | Technical Assessment |
|
|
221
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
222
|
+
| Tool Diversity Support | WARN | Full JMeter 5.x & Locust; k6 generator missing CLI |
|
|
223
|
+
| Workload Profile Shapes | PASS | Linear, Step, Spike, Soak profiles supported |
|
|
224
|
+
| Open vs Closed Arrival-Rate Models | FAIL | ThreadGroups use closed concurrency (Omission risk) |
|
|
225
|
+
| Automation Conversion Ingestion | PASS | OpenAPI 3.0, Swagger 2.0, Postman v2.1.0, HAR traces|
|
|
226
|
+
| Dynamic Data Correlation (CSRF/Auth)| PASS | Regex, JSON array randomizer, Groovy HMAC-SHA256 |
|
|
227
|
+
| Multi-Protocol Coverage | WARN | Dynamic HTTP; JDBC/WS/gRPC/Kafka are static samples |
|
|
228
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
229
|
+
```
|
|
230
|
+
|
|
231
|
+
#### Detailed SRE Technical Critique
|
|
232
|
+
|
|
233
|
+
##### 1. Multi-Engine Synthesis & Ingestion
|
|
234
|
+
`tools/generate_jmx_plan.py` (1,288 lines) provides comprehensive JMeter 5.x XML generation and Locust Python export (`--emit-locust`). It ingests OpenAPI 3.0 / Swagger 2.0 (`--spec`), Postman Collections v2.1.0 (`--postman`), and HTTP Archive (.har) traces (`--har`).
|
|
235
|
+
|
|
236
|
+
```python
|
|
237
|
+
# Locust generator (generate_jmx_plan.py:857-867)
|
|
238
|
+
class SynapseUser(HttpUser):
|
|
239
|
+
wait_time = between(think_time_min, think_time_max)
|
|
240
|
+
@task(weight)
|
|
241
|
+
def task_get_orders(self):
|
|
242
|
+
self.client.get("/api/v1/orders")
|
|
243
|
+
```
|
|
244
|
+
|
|
245
|
+
**Architectural Deficiencies:**
|
|
246
|
+
- **Missing k6 Generator:** While k6 is documented in `SKILL.md`, there is no CLI option to generate k6 scripts (`--emit-k6` does not exist).
|
|
247
|
+
- **Locust Flow Fragmentation:** In `generate_locustfile()` (lines 857–867), endpoints are emitted as standalone `@task(weight)` methods on `HttpUser`. Tasks are picked randomly by weight, rather than executing in a sequential user journey. No `on_start()` session lifecycle method is generated.
|
|
248
|
+
- **OpenAPI Ingestion Gaps:** In `parse_openapi_for_jmeter()` (lines 928–985), query parameters (`in: query`) are completely ignored. Only JSON request bodies and path parameters are mapped.
|
|
249
|
+
|
|
250
|
+
##### 2. Workload Shapes & Concurrency Modeling
|
|
251
|
+
`compute_workload_stages()` (lines 64–96) calculates stage progressions:
|
|
252
|
+
- **Step Load:** 4 staged thread groups with proportional user allocation (`per_step * (i + 1)`), step ramps, and duration offsets.
|
|
253
|
+
- **Spike Load:** 3 distinct phases: base (20% VUs) -> spike (100% VUs for 15% duration) -> recovery baseline (20% VUs).
|
|
254
|
+
- **Soak / Endurance:** Continuous single-stage profile with durations $\ge 3600$s.
|
|
255
|
+
|
|
256
|
+
```
|
|
257
|
+
SPIKE WORKLOAD PROFILE STEP WORKLOAD PROFILE
|
|
258
|
+
VUs VUs
|
|
259
|
+
100% | +-------+ 100% | +--------+
|
|
260
|
+
| | Peak | | +---+ Stage 4|
|
|
261
|
+
50% | | Spike | 50% | +---+ Stage 3 |
|
|
262
|
+
| | (15%) | | +---+ Stage 2 |
|
|
263
|
+
20% | +------+ +------+ 25% | +---+ Stage 1 |
|
|
264
|
+
+-------------------------> Time +------------------------------> Time
|
|
265
|
+
Base Spike Recovery Step 1 Step 2 Step 3 Step 4
|
|
266
|
+
```
|
|
267
|
+
|
|
268
|
+
These staged profiles emit standard JMeter `ThreadGroup` elements with scheduler delay offsets, functioning on stock Apache JMeter without requiring external plugins like Ultimate Thread Group.
|
|
269
|
+
|
|
270
|
+
**Closed-Model Concurrency Limitation:**
|
|
271
|
+
The generated thread groups use `ThreadGroup.num_threads`, modeling closed concurrency rather than open arrival rates. Under Little's Law:
|
|
272
|
+
|
|
273
|
+
$$N = \lambda \times W$$
|
|
274
|
+
|
|
275
|
+
Where $N$ is active concurrency, $\lambda$ is arrival rate (RPS), and $W$ is response latency. In a closed model, when the server degrades and $W$ increases, the client waits, throughput $\lambda$ drops, and the generated load decreases. This introduces **Coordinated Omission**, masking high tail latencies.
|
|
276
|
+
|
|
277
|
+
##### 3. Dynamic Correlation & Parameterization Engine
|
|
278
|
+
`generate_jmx_plan.py` injects dynamic extractors and signers:
|
|
279
|
+
- **CSRF Token Extractor:** Injects `RegexExtractor` on login responses (`tools/generate_jmx_plan.py:553–562`).
|
|
280
|
+
- **Random Item Selector:** Injects `JSONPostProcessor` with `match_number: 0` (`RANDOM_ITEM_ID`, lines 563–571) targeting `$..id` on array responses, preventing database row-lock contention.
|
|
281
|
+
- **Bearer Token Chaining:** Extracts `BEARER_TOKEN` on auth paths (lines 574–584) and attaches it to downstream requests.
|
|
282
|
+
- **JSR223 HMAC-SHA256 Signer:** Injects a Groovy preprocessor (`tools/generate_jmx_plan.py:451–479`) calculating `X-Signature`, `X-Timestamp`, and `X-Correlation-ID`.
|
|
283
|
+
- **CSV Data Parameterization:** Injects `CSVDataSet` config (lines 434–449) for multi-user credentials.
|
|
284
|
+
|
|
285
|
+
**Security Flaw:** The HMAC secret in the JSR223 Groovy script (lines 190, 464) is hardcoded to `"SynapseEnterpriseVault#2026"` rather than parameterized via JMeter properties (`${__P(HMAC_SECRET)}`).
|
|
286
|
+
|
|
287
|
+
##### 4. Multi-Protocol Coverage
|
|
288
|
+
The tool provides sample samplers for non-HTTP protocols:
|
|
289
|
+
- **JDBC Database:** Emits `JDBCDataSource` and `JDBCSampler` (`--jdbc-sample`, lines 356–384, 654–674).
|
|
290
|
+
- **WebSocket:** Emits `eu.luminis.jmeter.wssampler` (`--ws-sample`, lines 676–704).
|
|
291
|
+
- **gRPC Microservices:** Emits `vn.zalopay.benchmark.core.protobuf.ProtobufSampler` (`--grpc-sample`, lines 706–725).
|
|
292
|
+
- **Kafka Streaming:** Emits `org.apache.jmeter.protocol.jms.sampler.PublisherSampler` (`--kafka-sample`, lines 726–739).
|
|
293
|
+
|
|
294
|
+
However, non-HTTP protocols are static sample templates. There is no schema ingestion engine for `.proto` (gRPC), AsyncAPI (Kafka/WebSocket), or SQL schemas. Additionally, the WebSocket and gRPC samplers require third-party plugins that fail on stock JMeter installations unless pre-installed.
|
|
295
|
+
|
|
296
|
+
---
|
|
297
|
+
|
|
298
|
+
### 3.3 Pillar 3: CI/CD Integration & SLA Quality Gatekeeper
|
|
299
|
+
**Adversarial Score: 6.0 / 10**
|
|
300
|
+
|
|
301
|
+
```
|
|
302
|
+
+----------------------------------------------------------------------------------------------------+
|
|
303
|
+
| PILLAR 3 EVALUATION: CI/CD INTEGRATION & SLA QUALITY GATEKEEPER |
|
|
304
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
305
|
+
| Sub-Requirement | Status | Technical Assessment |
|
|
306
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
307
|
+
| High-Speed JTL / Metric Parsing | PASS | Sub-second parsing, linear interpolation percentiles|
|
|
308
|
+
| Context Window Overhead Protection | PASS | Log distillation and concise metric summaries |
|
|
309
|
+
| Deterministic Exit Code Gating | PASS | Exit code 1 on SLA breaches (--sla-p95, error-rate) |
|
|
310
|
+
| Multi-Engine Metric Ingestion | FAIL | Only JMeter supported; k6 JSON & Locust CSV missing |
|
|
311
|
+
| Integration with Central DoD Gates | FAIL | tools/verify_all.py completely lacks perf gates |
|
|
312
|
+
| CI/CD Pipeline Template Binding | FAIL | GitHub/GitLab actions omit performance test jobs |
|
|
313
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
314
|
+
```
|
|
315
|
+
|
|
316
|
+
#### Detailed SRE Technical Critique
|
|
317
|
+
|
|
318
|
+
##### 1. High-Speed JTL Parser & Exit-Code Gating
|
|
319
|
+
`tools/jtl_analyzer.py` (475 lines) provides a standalone statistical parser for JMeter JTL files (CSV and XML). It calculates request volume, throughput (req/s), error counts, error percentages, and percentiles (P50, P90, P95, P99) with sub-second execution speed:
|
|
320
|
+
|
|
321
|
+
```python
|
|
322
|
+
# Evaluates multi-metric SLA breaches (tools/jtl_analyzer.py:230-253)
|
|
323
|
+
def evaluate_sla(agg_stats, sla_p95=None, sla_p99=None, sla_avg=None, max_error_rate=None):
|
|
324
|
+
violations = []
|
|
325
|
+
if sla_p95 is not None and agg_stats["p95_ms"] > sla_p95:
|
|
326
|
+
violations.append(f"P95 Latency SLA Breached: {agg_stats['p95_ms']}ms > {sla_p95}ms")
|
|
327
|
+
if sla_p99 is not None and agg_stats["p99_ms"] > sla_p99:
|
|
328
|
+
violations.append(f"P99 Latency SLA Breached: {agg_stats['p99_ms']}ms > {sla_p99}ms")
|
|
329
|
+
if max_error_rate is not None and agg_stats["error_rate_pct"] > max_error_rate:
|
|
330
|
+
violations.append(f"Error Rate SLA Breached: {agg_stats['error_rate_pct']}% > {max_error_rate}%")
|
|
331
|
+
return (len(violations) == 0), violations
|
|
332
|
+
|
|
333
|
+
# Deterministic Pipeline Exit Code (Line 470)
|
|
334
|
+
sys.exit(0 if passed else 1)
|
|
335
|
+
```
|
|
336
|
+
|
|
337
|
+
The tool is bound to `agent-qa analyze-jtl` and `synapse analyze-jtl` in `tools/cli.py` (lines 462–465). It emits Markdown summary reports formatted for GitHub PR comments, preventing context window bloat.
|
|
338
|
+
|
|
339
|
+
##### 2. Engine Format Lockout: Missing k6 and Locust Ingestion
|
|
340
|
+
Despite claims of multi-engine support, `jtl_analyzer.py` only implements `parse_csv_jtl` and `parse_xml_jtl` (lines 217–228):
|
|
341
|
+
- **k6 JSON Ingestion Missing:** Zero parsing logic exists for `k6 run --summary-export=summary.json`.
|
|
342
|
+
- **Locust CSV Ingestion Missing:** Zero parsing logic exists for Locust `_stats.csv`. Passing Locust stats to `jtl_analyzer.py` fails due to schema divergence (Locust uses `Request Count`, `Failure Count`, `Median Response Time` vs JMeter's `timeStamp`, `elapsed`, `success`).
|
|
343
|
+
|
|
344
|
+
Teams using k6 or Locust are locked out of the automated SLA quality gate.
|
|
345
|
+
|
|
346
|
+
##### 3. Functional Log Distiller Mismatch
|
|
347
|
+
`tools/distill_log.py` (lines 27–57) targets functional test frameworks, filtering strings like `> Task :`, `Starting ChromeDriver`, `Katalon Studio`, and regex matches for `TC-\d+` and `AssertionError`.
|
|
348
|
+
The distiller lacks regex patterns for JMeter CLI summaries (`summary + 1500 in 00:00:10 = 150.0/s Avg: 120 Min: 80 Max: 450 Err: 0 (0.00%)`), k6 terminal outputs (`✓ status is 200`, `http_req_duration: p(95)=420ms`), and Locust terminal tables. Piping load test logs into `distill_log.py` results in either suppressed performance data or unparsed output blocks.
|
|
349
|
+
|
|
350
|
+
##### 4. Decoupling from Core Definition of Done & CI Templates
|
|
351
|
+
- **Omission from `tools/verify_all.py`:** `verify_all.py` (lines 11–22) enforces Gates 0 through 9:
|
|
352
|
+
```python
|
|
353
|
+
# Gates 0-9 in tools/verify_all.py:
|
|
354
|
+
# Gate 0: Author Attribution & Copyright
|
|
355
|
+
# Gate 1: 100% UTF-8 No BOM
|
|
356
|
+
# Gate 2: Path Portability
|
|
357
|
+
# Gate 3: Clean POM & Flat Architecture
|
|
358
|
+
# Gate 4: AI Docs Anti-Drift
|
|
359
|
+
# Gate 5: Cleanup Blacklist
|
|
360
|
+
# Gate 6: Credential Security
|
|
361
|
+
# Gate 7: Static Defect Sentinel (AST)
|
|
362
|
+
# Gate 8: Micro-Mutation Sentinel (Mutation Score >= 80%)
|
|
363
|
+
# Gate 9: Tier-2 Skill Collision Guard
|
|
364
|
+
```
|
|
365
|
+
Stage 5 Quality Gatekeeper (`quality-gatekeeper.md`) relies on `verify_all.py`. Because performance SLA gating is omitted from these 10 gates, a severe performance regression will pass Stage 5 unnoticed.
|
|
366
|
+
- **Omission from CI/CD Pipelines:** Official workflow templates (`templates/ci/github-actions-synapse.yml` lines 31–43 and `templates/ci/gitlab-ci-synapse.yml` lines 12–13) only execute `npx synapse-hub lint` and `npx synapse-hub verify`. No performance testing or JTL analysis job exists in default pipelines.
|
|
367
|
+
|
|
368
|
+
---
|
|
369
|
+
|
|
370
|
+
### 3.4 Pillar 4: Full-Stack APM & Hardware Telemetry RCA
|
|
371
|
+
**Adversarial Score: 5.5 / 10**
|
|
372
|
+
|
|
373
|
+
```
|
|
374
|
+
+----------------------------------------------------------------------------------------------------+
|
|
375
|
+
| PILLAR 4 EVALUATION: FULL-STACK APM & HARDWARE TELEMETRY RCA |
|
|
376
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
377
|
+
| Sub-Requirement | Status | Technical Assessment |
|
|
378
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
379
|
+
| Client vs Server Telemetry Correlator| PASS | Pearson r implementation across 5s time windows |
|
|
380
|
+
| Bottleneck Heuristic RCA Reporting | WARN | 4 coarse threshold rules; lacks deep SQL/GC parsing |
|
|
381
|
+
| Live APM Telemetry Ingestion | FAIL | Offline CSV only; zero Prometheus/Datadog connectors|
|
|
382
|
+
| NTP Clock Drift / Lag Compensation | FAIL | Modulo bucketing collapses r on 2-3s clock skew |
|
|
383
|
+
| Deep Root Cause Diagnostics | FAIL | No SQL EXPLAIN, JVM GC log, or thread dump parsing |
|
|
384
|
+
| OpenTelemetry (OTel) Integration | FAIL | Static mock JSON string to stdout; no OTLP egress |
|
|
385
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
386
|
+
```
|
|
387
|
+
|
|
388
|
+
#### Detailed SRE Technical Critique
|
|
389
|
+
|
|
390
|
+
##### 1. Mathematical Pearson Correlation Engine
|
|
391
|
+
`tools/telemetry_correlator.py` (lines 50–70) implements standard Pearson correlation coefficient ($r$) calculation:
|
|
392
|
+
|
|
393
|
+
$$r = \frac{\sum_{i=1}^n (x_i - \bar{x})(y_i - \bar{y})}{\sqrt{\sum_{i=1}^n (x_i - \bar{x})^2 \sum_{i=1}^n (y_i - \bar{y})^2}}$$
|
|
394
|
+
|
|
395
|
+
```python
|
|
396
|
+
def calculate_pearson_correlation(x: List[float], y: List[float]) -> float:
|
|
397
|
+
n = min(len(x), len(y))
|
|
398
|
+
if n < 2: return 0.0
|
|
399
|
+
x, y = x[:n], y[:n]
|
|
400
|
+
mean_x, mean_y = sum(x) / n, sum(y) / n
|
|
401
|
+
numerator = sum((x[i] - mean_x) * (y[i] - mean_y) for i in range(n))
|
|
402
|
+
denom_x = sum((x[i] - mean_x) ** 2 for i in range(n))
|
|
403
|
+
denom_y = sum((y[i] - mean_y) ** 2 for i in range(n))
|
|
404
|
+
if denom_x <= 0 or denom_y <= 0: return 0.0
|
|
405
|
+
return numerator / math.sqrt(denom_x * denom_y)
|
|
406
|
+
```
|
|
407
|
+
|
|
408
|
+
The tool aggregates client response times from a JTL file into 5-second windows, computes P95 latency per window, and pairs it with server metrics (`cpu_pct`, `memory_pct`, `db_pool_pct`, `disk_iops`). It evaluates heuristic bottleneck rules:
|
|
409
|
+
- **Compute Saturation:** `corr_cpu >= 0.65 and max_cpu > 80%` $\rightarrow$ CRITICAL CPU alert.
|
|
410
|
+
- **Database Connection Starvation:** `corr_db >= 0.65 and max_db > 85%` $\rightarrow$ CRITICAL DB Pool alert.
|
|
411
|
+
- **JVM Memory Pressure / GC:** `corr_mem >= 0.60 and max_mem > 85%` $\rightarrow$ HIGH Memory alert.
|
|
412
|
+
- **Downstream Dependency Delay:** `max_p95 > 1000 and max_cpu < 50 and max_db < 50` $\rightarrow$ Third-party latency flag.
|
|
413
|
+
|
|
414
|
+
##### 2. Offline CSV Coupling & Zero Live APM Ingestion
|
|
415
|
+
`parse_server_metrics()` (lines 90–122) reads exclusively from a local CSV file passed via `--server-metrics`.
|
|
416
|
+
The codebase contains zero API clients or query connectors for Prometheus (`/api/v1/query_range`), Datadog (`/api/v1/query`), AWS CloudWatch Metrics Insights, Dynatrace, or New Relic. Telemetry correlation cannot execute dynamically in CI/CD without custom external shell scripts to fetch and convert metrics.
|
|
417
|
+
|
|
418
|
+
##### 3. Clock Skew Susceptibility & Modulo Alignment Failure
|
|
419
|
+
Time-series bucketing in `telemetry_correlator.py` (lines 81, 108) relies on static integer division:
|
|
420
|
+
|
|
421
|
+
$$\text{Bucket ID} = \left\lfloor \frac{\text{Timestamp}_{\text{sec}}}{\text{Bucket Size}} \right\rfloor \times \text{Bucket Size}$$
|
|
422
|
+
|
|
423
|
+
The tool then performs a direct 1:1 index alignment. If the load injector machine and the target server experience an NTP clock drift of $\ge 3$ seconds, or if application queueing introduces a 5–10s metric reporting delay, time slices shift into adjacent buckets.
|
|
424
|
+
Because Pearson correlation requires synchronous index alignment, this phase shift causes the correlation coefficient to collapse toward zero ($r \approx 0.0$), generating false negatives and failing to identify active bottlenecks. The tool lacks lag cross-correlation ($\text{xcorr}(\tau)$) or Dynamic Time Warping (DTW).
|
|
425
|
+
|
|
426
|
+
```
|
|
427
|
+
IMPACT OF CLOCK SKEW ON STATIC MODULO BUCKETING
|
|
428
|
+
Latency
|
|
429
|
+
Spike [ Window 2 ]
|
|
430
|
+
^ ***
|
|
431
|
+
| * *
|
|
432
|
+
--+-------*-----*-------+--------+--------+--------> Time
|
|
433
|
+
| 0s 5s 10s 15s 20s
|
|
434
|
+
CPU
|
|
435
|
+
Load [ Window 3 ] (Clock Skew / Reporting Delay: +4s)
|
|
436
|
+
^ ***
|
|
437
|
+
| * *
|
|
438
|
+
--+---------------+---*-*--------+--------+--------> Time
|
|
439
|
+
| 0s 5s 10s 15s 20s
|
|
440
|
+
|
|
441
|
+
RESULT: Window 2 (High Latency, Low CPU) + Window 3 (Low Latency, High CPU)
|
|
442
|
+
Pearson Correlation drops from r = +0.92 to r = +0.08 (False Negative)
|
|
443
|
+
```
|
|
444
|
+
|
|
445
|
+
##### 4. Superficial RCA Heuristics
|
|
446
|
+
The RCA engine evaluates only 4 coarse threshold rules against high-level scalar percentages:
|
|
447
|
+
- **No SQL Analysis:** Does not parse MySQL slow query logs, PostgreSQL `pg_stat_statements`, or Oracle AWR reports.
|
|
448
|
+
- **No JVM GC Log Parsing:** Mentions "GC pause (Stop-The-World)" in text, but does not parse GC logs (`-Xlog:gc*`), Safepoint logs, or Young/Old generation transitions.
|
|
449
|
+
- **No Thread Dump Analysis:** Does not parse thread dumps (`jstack` / async-profiler) for `BLOCKED` states or lock contention.
|
|
450
|
+
- **No Connection Leak Slope Detection:** Checks only peak DB pool usage; does not calculate the rate of connection accumulation ($dC/dt$).
|
|
451
|
+
|
|
452
|
+
##### 5. Pseudo-OpenTelemetry Export
|
|
453
|
+
Passing `--otel` to `telemetry_correlator.py` (lines 344–380) executes:
|
|
454
|
+
|
|
455
|
+
```python
|
|
456
|
+
otel_metrics = {
|
|
457
|
+
"resourceMetrics": [{
|
|
458
|
+
"resource": {"attributes": [{"key": "service.name", "value": {"stringValue": "synapse-backend-under-test"}}]},
|
|
459
|
+
"scopeMetrics": [{
|
|
460
|
+
"metrics": [
|
|
461
|
+
{"name": "http.server.request.duration", "unit": "ms", "gauge": {"dataPoints": [{"value": analysis["metrics_summary"]["max_p95_ms"]}]}},
|
|
462
|
+
{"name": "system.cpu.utilization", "unit": "1", "gauge": {"dataPoints": [{"value": round(analysis["metrics_summary"]["max_cpu_pct"] / 100.0, 4)}]}},
|
|
463
|
+
{"name": "system.memory.utilization", "unit": "1", "gauge": {"dataPoints": [{"value": round(analysis["metrics_summary"]["max_memory_pct"] / 100.0, 4)}]}}
|
|
464
|
+
]
|
|
465
|
+
}]
|
|
466
|
+
}]
|
|
467
|
+
}
|
|
468
|
+
print(json.dumps(otel_metrics, indent=2))
|
|
469
|
+
```
|
|
470
|
+
|
|
471
|
+
This simply prints a mock JSON string to `stdout`. There is no OTLP client (gRPC or HTTP), no OTLP collector endpoint configuration, and no trace context propagation (`traceparent`) injected into load test HTTP samplers.
|
|
472
|
+
|
|
473
|
+
---
|
|
474
|
+
|
|
475
|
+
### 3.5 Pillar 5: Frontend Web Vitals & Client Experience
|
|
476
|
+
**Adversarial Score: 3.5 / 10**
|
|
477
|
+
|
|
478
|
+
```
|
|
479
|
+
+----------------------------------------------------------------------------------------------------+
|
|
480
|
+
| PILLAR 5 EVALUATION: FRONTEND WEB VITALS & CLIENT EXPERIENCE |
|
|
481
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
482
|
+
| Sub-Requirement | Status | Technical Assessment |
|
|
483
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
484
|
+
| Core Web Vitals (LCP, CLS, TTFB) | FAIL | Relies on deprecated Navigation Timing v1 (2019) |
|
|
485
|
+
| Interaction to Next Paint (INP) | FAIL | Claimed in docs; completely missing in code |
|
|
486
|
+
| Measurement Window & Lifecycle | FAIL | 1.5s timeout; teardown evaluation freezes LCP early |
|
|
487
|
+
| Network & CPU Throttling Realism | FAIL | Fast 3G in policy, but fixture lacks CDP throttling |
|
|
488
|
+
| Playwright E2E Suite Integration | FAIL | Orphaned fixture; zero CRM suite integration |
|
|
489
|
+
| Memory Leak & Heap Snapshot Engine | FAIL | Unmounted MCP; Gate 5 blacklists heap dump artifacts|
|
|
490
|
+
| Detached DOM & Event Listener Checks| FAIL | Zero automated leak assertions in test suites |
|
|
491
|
+
+-------------------------------------+--------+-----------------------------------------------------+
|
|
492
|
+
```
|
|
493
|
+
|
|
494
|
+
#### Detailed SRE Technical Critique
|
|
495
|
+
|
|
496
|
+
##### 1. Deprecated Navigation Timing API
|
|
497
|
+
In `.agents/skills/web-vitals-and-performance-profiler/SKILL.md` (lines 35–44), the documented Web Vitals measurement pattern uses:
|
|
498
|
+
|
|
499
|
+
```typescript
|
|
500
|
+
// Deprecated Navigation Timing Level 1 Pattern (SKILL.md:35-44)
|
|
501
|
+
const performanceTiming = JSON.parse(
|
|
502
|
+
await page.evaluate(() => JSON.stringify(window.performance.timing))
|
|
503
|
+
);
|
|
504
|
+
const ttfb = performanceTiming.responseStart - performanceTiming.navigationStart;
|
|
505
|
+
const domLoadTime = performanceTiming.domContentLoadedEventEnd - performanceTiming.navigationStart;
|
|
506
|
+
console.log(`[PERF] TTFB: ${ttfb}ms | DOM Load: ${domLoadTime}ms`);
|
|
507
|
+
expect(ttfb).toBeLessThan(1500); // 1.5s threshold
|
|
508
|
+
```
|
|
509
|
+
|
|
510
|
+
`window.performance.timing` belongs to Navigation Timing Level 1, deprecated by W3C in 2019 in favor of Navigation Timing Level 2 (`PerformanceNavigationTiming`). This pattern measures only basic network timestamps and fails to capture any Google Core Web Vitals (LCP, CLS, or INP).
|
|
511
|
+
|
|
512
|
+
##### 2. Complete Absence of INP (Interaction to Next Paint)
|
|
513
|
+
While `web-vitals-and-performance-profiler/SKILL.md` (line 3) claims to *"Measure Core Web Vitals (LCP, CLS, INP) and network waterfalls via CDP"*, and `cwv_budget_policy.json` (line 4) declares `"INP_ms": 200`, the actual Playwright fixture (`playwright_cwv_profiler_fixture.ts`) contains zero INP code:
|
|
514
|
+
|
|
515
|
+
```typescript
|
|
516
|
+
// Citation: examples/playwright_cwv_profiler_fixture.ts (Lines 8-12, 34-43)
|
|
517
|
+
interface PerformanceMetrics {
|
|
518
|
+
lcp: number; // ms
|
|
519
|
+
cls: number; // score
|
|
520
|
+
domContentLoaded: number; // ms
|
|
521
|
+
}
|
|
522
|
+
|
|
523
|
+
// Observers implemented:
|
|
524
|
+
new PerformanceObserver((entryList) => { ... }).observe({ type: 'largest-contentful-paint', buffered: true });
|
|
525
|
+
new PerformanceObserver((entryList) => { ... }).observe({ type: 'layout-shift', buffered: true });
|
|
526
|
+
|
|
527
|
+
// Assertions (Lines 57-59):
|
|
528
|
+
expect(finalMetrics.lcp).toBeLessThan(2500);
|
|
529
|
+
expect(finalMetrics.cls).toBeLessThan(0.10);
|
|
530
|
+
// INP is completely omitted from interface, observers, and assertions
|
|
531
|
+
```
|
|
532
|
+
|
|
533
|
+
As of March 2024, INP officially replaced FID (First Input Delay) as a Core Web Vital. The ecosystem lacks observers for `type: 'event'` or `type: 'first-input'`, leaving interaction responsiveness unmeasured.
|
|
534
|
+
|
|
535
|
+
##### 3. Measurement Lifecycle Flaws in Playwright Profiler Fixture
|
|
536
|
+
In `playwright_cwv_profiler_fixture.ts` (lines 42, 54, 57):
|
|
537
|
+
1. **Arbitrary Static Timeout:** Line 42 relies on `setTimeout(() => resolve({ lcp, cls }), 1500);`, an arbitrary delay that fails under variable network conditions.
|
|
538
|
+
2. **Stub Values in Test Execution:** Line 54 passes zeroed stub values to the active test:
|
|
539
|
+
```typescript
|
|
540
|
+
await use({ lcp: 0, cls: 0, domContentLoaded: 0 });
|
|
541
|
+
```
|
|
542
|
+
The test cannot assert against real-time metrics during user actions.
|
|
543
|
+
3. **Premature LCP Candidate Freezing:** Line 57 invokes `getMetrics()` during fixture teardown, after test interactions have run. Under the W3C LCP specification, the browser terminates LCP candidate dispatch upon the first user interaction (click, keypress, scroll). Evaluating LCP after full user flows freezes the metric at the first interaction point rather than capturing true visual completion.
|
|
544
|
+
|
|
545
|
+
##### 4. Absence of CDP Network & CPU Throttling
|
|
546
|
+
While `cwv_budget_policy.json` specifies `"networkThrottle": "Fast 3G"`, `playwright_cwv_profiler_fixture.ts` contains zero calls to CDP `Network.emulateNetworkConditions` or `Emulation.setCPUThrottlingRate`.
|
|
547
|
+
The profiler runs unthrottled against local or high-speed CI networks, producing false passes that fail to detect real-world mobile or slow-network performance regressions.
|
|
548
|
+
|
|
549
|
+
##### 5. Orphaned Profiler Fixture
|
|
550
|
+
`examples/playwright-crm-suite/tests/customers.spec.ts` imports standard `@playwright/test` and `BasePage`, with zero references to `playwright_cwv_profiler_fixture.ts`. Across the entire repository, zero functional test suites reference the profiler fixture.
|
|
551
|
+
|
|
552
|
+
##### 6. Memory Leak & Heap Snapshot Tooling Gaps
|
|
553
|
+
- **Unmounted MCP Server:** Memory debugging instructions in `/Users/phungcao/.gemini/config/plugins/chrome-devtools-plugin/skills/memory-leak-debugging/SKILL.md` rely on the `chrome-devtools` MCP server started with `--memoryDebugging`. However, the active runtime environment registers only `genoffice` and `playwright`.
|
|
554
|
+
- **Gate 5 Quality Gate Conflict:** Gate 5 in `tools/verify_all.py` (line 433) explicitly blacklists temporary files and heap dumps, scanning for `.heapsnapshot` and `.hprof` files. Generating heap dumps during CI causes `verify_all.py` to fail.
|
|
555
|
+
- **Zero Detached DOM Node Assertions:** Codebase references to `detached` strictly refer to waiting for loading spinners to detach (`waitFor({ state: 'detached' })`). There is zero automated verification for detached DOM node retention or leaking event listeners.
|
|
556
|
+
|
|
557
|
+
---
|
|
558
|
+
|
|
559
|
+
## 4. Itemized Standout Strengths
|
|
560
|
+
|
|
561
|
+
```
|
|
562
|
+
+---------------------------------------------------------------------------------------------------------+
|
|
563
|
+
| REPOSITORY STANDOUT CAPABILITIES |
|
|
564
|
+
+----+--------------------------------------------+------------------------------------+------------------+
|
|
565
|
+
| # | Capability | Repository File Citation | Technical Pillar |
|
|
566
|
+
+----+--------------------------------------------+------------------------------------+------------------+
|
|
567
|
+
| 1 | Addy Osmani Anti-Rationalization Guard | .agents/agents/perf-engineer.md:61 | Pillar 1 |
|
|
568
|
+
| 2 | Sub-0.1s Linear Interpolation JTL Parser | tools/jtl_analyzer.py:48-59 | Pillar 1 & 3 |
|
|
569
|
+
| 3 | Reversible Sandboxing & WAL Journaling | tools/reversible_sandbox.py:9-60 | Pillar 1 |
|
|
570
|
+
| 4 | Multi-Engine JMeter 5.x & Locust Synthesis | tools/generate_jmx_plan.py:1-60 | Pillar 2 |
|
|
571
|
+
| 5 | High-Precision Dynamic Correlation Extract | tools/generate_jmx_plan.py:550-595 | Pillar 2 |
|
|
572
|
+
| 6 | Vanilla Plugin-Free Staged Workloads | tools/generate_jmx_plan.py:64-96 | Pillar 2 |
|
|
573
|
+
| 7 | Groovy JSR223 HMAC-SHA256 Request Signer | tools/generate_jmx_plan.py:451-479 | Pillar 2 |
|
|
574
|
+
| 8 | Deterministic Exit Code SLA Gating | tools/jtl_analyzer.py:230-470 | Pillar 3 |
|
|
575
|
+
| 9 | Mathematical Pearson Telemetry Correlation | tools/telemetry_correlator.py:50 | Pillar 4 |
|
|
576
|
+
| 10 | 4-Subpart LCP Diagnostic Framework | plugins/.../debug-optimize-lcp/SKILL.md:16 | Pillar 5 |
|
|
577
|
+
+----+--------------------------------------------+------------------------------------+------------------+
|
|
578
|
+
```
|
|
579
|
+
|
|
580
|
+
### Detailed Capability Analysis
|
|
581
|
+
|
|
582
|
+
1. **Addy Osmani Anti-Rationalization Guard (`.agents/agents/perf-engineer.md:61–75`):**
|
|
583
|
+
Explicitly dismantles common engineering rationalizations (hiding tail latency behind average response times, accepting 2.5% error rates, removing think time, and sharing authentication tokens).
|
|
584
|
+
2. **Sub-0.1s Linear Interpolation JTL Parser (`tools/jtl_analyzer.py:48–59`):**
|
|
585
|
+
Implements exact mathematical percentile interpolation, preventing rank errors on small sample sizes and computing P50, P90, P95, and P99 percentiles in sub-second runtimes.
|
|
586
|
+
3. **Reversible Sandboxing & WAL Journaling (`tools/reversible_sandbox.py:9–60`):**
|
|
587
|
+
Maintains a LIFO undo stack backed by Write-Ahead Logging (`.synapse/undo_wal.jsonl`), ensuring idempotent cleanup of test data even after process aborts or SIGKILL signals.
|
|
588
|
+
4. **Multi-Engine JMeter 5.x & Locust Synthesis (`tools/generate_jmx_plan.py:1–60, 803–902`):**
|
|
589
|
+
Ingests OpenAPI 3.0, Swagger 2.0, Postman Collections (v2.1.0), and HAR traces, emitting both standard Apache JMeter 5.x XML and Locust Python code-as-code.
|
|
590
|
+
5. **Dynamic Correlation Engine (`tools/generate_jmx_plan.py:550–595`):**
|
|
591
|
+
Injects regex extractors for CSRF tokens, Bearer token chaining, and `JSONPostProcessor` with `match_number: 0` on `$..id` to randomly sample entity IDs, preventing database row-lock contention.
|
|
592
|
+
6. **Vanilla Plugin-Free Staged Workloads (`tools/generate_jmx_plan.py:64–96`):**
|
|
593
|
+
Decomposes step and spike workload stages into native JMeter ThreadGroups using scheduled delay offsets, running out of the box on stock JMeter installations without external plugins.
|
|
594
|
+
7. **Groovy JSR223 HMAC-SHA256 Signer (`tools/generate_jmx_plan.py:451–479`):**
|
|
595
|
+
Automatically computes dynamic crypto signatures (`X-Signature`), epoch timestamps (`X-Timestamp`), and UUID correlation IDs (`X-Correlation-ID`) across load requests.
|
|
596
|
+
8. **Deterministic Exit Code SLA Gating (`tools/jtl_analyzer.py:230–470`):**
|
|
597
|
+
Enforces user-specified `--sla-p95`, `--sla-p99`, and `--max-error-rate` thresholds, returning Exit Code 0 on pass and Exit Code 1 on SLA breaches for pipeline gating.
|
|
598
|
+
9. **Mathematical Pearson Telemetry Correlator (`tools/telemetry_correlator.py:50–70`):**
|
|
599
|
+
Calculates Pearson correlation coefficients ($r$) between client-side P95 latencies and server CPU, memory, database connection pool, and disk IOPS metrics.
|
|
600
|
+
10. **4-Subpart LCP Diagnostic Framework (`debug-optimize-lcp/SKILL.md:16–30`):**
|
|
601
|
+
Adopts the industry-standard breakdown for Largest Contentful Paint (TTFB ~40%, Resource Load Delay <10%, Resource Load Duration ~40%, Element Render Delay <10%), providing structured guidance for frontend performance optimization.
|
|
602
|
+
|
|
603
|
+
---
|
|
604
|
+
|
|
605
|
+
## 5. Itemized Gaps & Action Items (Roadmap to 10/10)
|
|
606
|
+
|
|
607
|
+
```
|
|
608
|
+
+---------------------------------------------------------------------------------------------------------+
|
|
609
|
+
| PRIORITIZED REMEDIATION ROADMAP |
|
|
610
|
+
+----------+-------------------------------------------------------------+--------------------------------+
|
|
611
|
+
| Priority | Technical Action Item | Target Files |
|
|
612
|
+
+----------+-------------------------------------------------------------+--------------------------------+
|
|
613
|
+
| P0 (Now) | 1. Implement Gate 10 (Performance SLA Gate) in DoD | tools/verify_all.py |
|
|
614
|
+
| P0 (Now) | 2. Fix k6 Token Reuse Anti-Pattern in Canonical Skill | backend-load-and-stress/SKILL.md|
|
|
615
|
+
| P0 (Now) | 3. Inject AutoStop Kill-Switch into JMeter Generation Plan | tools/generate_jmx_plan.py |
|
|
616
|
+
| P0 (Now) | 4. Modernize Web Vitals Profiler Fixture (INP, CDP, Nav v2) | playwright_cwv_profiler.ts |
|
|
617
|
+
| P0 (Now) | 5. Reconcile Gate 5 Heap Dump Blacklist with Artifact Dirs | tools/verify_all.py |
|
|
618
|
+
+----------+-------------------------------------------------------------+--------------------------------+
|
|
619
|
+
| P1 (High)| 6. Implement Universal Ingestion (k6 JSON & Locust CSV) | tools/jtl_analyzer.py |
|
|
620
|
+
| P1 (High)| 7. Wire Safe Non-Prod Host Firewall into Load Generator | tools/generate_jmx_plan.py |
|
|
621
|
+
| P1 (High)| 8. Add NTP Clock Drift Cross-Correlation (xcorr) to APM | tools/telemetry_correlator.py |
|
|
622
|
+
| P1 (High)| 9. Build Automated k6 Script Generator CLI (--emit-k6) | tools/generate_jmx_plan.py |
|
|
623
|
+
| P1 (High)| 10. Update distill_log.py Regex for JMeter & k6 Streams | tools/distill_log.py |
|
|
624
|
+
+----------+-------------------------------------------------------------+--------------------------------+
|
|
625
|
+
| P2 (Med) | 11. Add Live Prometheus Metric Connector to APM Correlator | tools/telemetry_correlator.py |
|
|
626
|
+
| P2 (Med) | 12. Support Open Model Arrival-Rate ThreadGroups | tools/generate_jmx_plan.py |
|
|
627
|
+
| P2 (Med) | 13. Deep RCA: Ingest Slow Query Logs & JVM GC Safepoint Logs| tools/telemetry_correlator.py |
|
|
628
|
+
| P2 (Med) | 14. Real OTLP gRPC/HTTP Telemetry Egress & W3C Tracing | tools/telemetry_correlator.py |
|
|
629
|
+
| P2 (Med) | 15. Headless Client-Side Memory Leak CI Verifier (CDP Heap) | tools/check_memory_leaks.py |
|
|
630
|
+
+----------+-------------------------------------------------------------+--------------------------------+
|
|
631
|
+
```
|
|
632
|
+
|
|
633
|
+
### Detailed Remediation Specifications
|
|
634
|
+
|
|
635
|
+
#### P0 — Critical Remediation Items
|
|
636
|
+
|
|
637
|
+
##### 1. Implement Gate 10 (Performance & SLA Sentinel) in `tools/verify_all.py`
|
|
638
|
+
- **Problem:** Performance verification is completely detached from the 10 central DoD Quality Gates.
|
|
639
|
+
- **Solution:** Add `gate_10_performance_sla(repo_root, mode)` to `tools/verify_all.py`. Ingest test results from `Reports/perf/` (`results.jtl` or `k6_summary.json`), execute `jtl_analyzer.py`, and fail the gate if P95 exceeds policy thresholds or error rate exceeds 1.0%.
|
|
640
|
+
|
|
641
|
+
##### 2. Fix k6 Token Reuse Anti-Pattern in `backend-load-and-stress-patterns/SKILL.md`
|
|
642
|
+
- **Problem:** `SKILL.md` lines 71–88 fetch a single token in `setup()` and share it across all 200 VUs, violating `perf-engineer.md` line 68.
|
|
643
|
+
- **Solution:** Refactor the template to parameterize authentication per-VU using a `SharedArray` of pre-seeded test accounts:
|
|
644
|
+
```javascript
|
|
645
|
+
import { SharedArray } from 'k6/data';
|
|
646
|
+
const users = new SharedArray('users', () => JSON.parse(open('./users.json')));
|
|
647
|
+
|
|
648
|
+
export default function () {
|
|
649
|
+
const user = users[__VU % users.length];
|
|
650
|
+
const loginRes = http.post(`${__ENV.BASE_URL}/api/v1/auth/login`, JSON.stringify(user));
|
|
651
|
+
const token = loginRes.json('token');
|
|
652
|
+
// Authentic per-VU session flow
|
|
653
|
+
}
|
|
654
|
+
```
|
|
655
|
+
|
|
656
|
+
##### 3. Inject AutoStop Kill-Switch into `tools/generate_jmx_plan.py`
|
|
657
|
+
- **Problem:** JMeter plans emit `<stringProp name="ThreadGroup.on_sample_error">continue</stringProp>` with no aggregate error kill-switch.
|
|
658
|
+
- **Solution:** Inject a JSR223 Assertion or AutoStop Listener that tracks rolling failure rates:
|
|
659
|
+
```groovy
|
|
660
|
+
long total = vars.getObject("TOTAL_REQUESTS") ?: 0;
|
|
661
|
+
long failed = vars.getObject("FAILED_REQUESTS") ?: 0;
|
|
662
|
+
total++;
|
|
663
|
+
if (!prev.isSuccessful()) { failed++; }
|
|
664
|
+
vars.putObject("TOTAL_REQUESTS", total);
|
|
665
|
+
vars.putObject("FAILED_REQUESTS", failed);
|
|
666
|
+
|
|
667
|
+
double errorRate = (double) failed / (double) total;
|
|
668
|
+
if (total > 50 && errorRate > 0.10) {
|
|
669
|
+
log.error("CRITICAL SLA BREACH: Error rate " + (errorRate * 100) + "% exceeds 10% threshold. Halting test.");
|
|
670
|
+
prev.setStopTestNow(true);
|
|
671
|
+
}
|
|
672
|
+
```
|
|
673
|
+
|
|
674
|
+
##### 4. Modernize Web Vitals Profiler Fixture (`playwright_cwv_profiler_fixture.ts`)
|
|
675
|
+
- **Problem:** Uses deprecated `window.performance.timing`, omits INP, runs unthrottled, and resolves after interactions have already frozen LCP.
|
|
676
|
+
- **Solution:**
|
|
677
|
+
- Modernize to Navigation Timing Level 2: `performance.getEntriesByType('navigation')[0]`.
|
|
678
|
+
- Add `PerformanceObserver` for `type: 'event'` with `durationThreshold: 40` to capture INP.
|
|
679
|
+
- Inject CDP throttling during fixture setup:
|
|
680
|
+
```typescript
|
|
681
|
+
await cdpSession.send('Network.emulateNetworkConditions', {
|
|
682
|
+
offline: false, latency: 150, downloadThroughput: (1.6 * 1024 * 1024) / 8, uploadThroughput: (750 * 1024) / 8
|
|
683
|
+
}); // Fast 3G
|
|
684
|
+
await cdpSession.send('Emulation.setCPUThrottlingRate', { rate: 4 }); // 4x Slowdown
|
|
685
|
+
```
|
|
686
|
+
- Expose `getMetrics()` to the active test body via fixture context to allow assertions during user actions.
|
|
687
|
+
|
|
688
|
+
##### 5. Reconcile Gate 5 Heap Dump Blacklist in `tools/verify_all.py`
|
|
689
|
+
- **Problem:** Gate 5 blacklists all `.heapsnapshot` files, preventing memory leak testing in CI.
|
|
690
|
+
- **Solution:** Update `tools/verify_all.py` line 433 to exempt designated test artifact directories (e.g. `Reports/perf/` and `.synapse/heapsnapshots/`) from the blacklist, preserving repository hygiene while permitting test artifacts.
|
|
691
|
+
|
|
692
|
+
---
|
|
693
|
+
|
|
694
|
+
#### P1 — High-Priority Remediation Items
|
|
695
|
+
|
|
696
|
+
##### 6. Universal Metric Ingestion in `tools/jtl_analyzer.py`
|
|
697
|
+
- Add `--engine` flag supporting `jmeter`, `k6`, and `locust`.
|
|
698
|
+
- Parse k6 `--summary-export=summary.json` mapping `metrics.http_req_duration.values["p(95)"]` and `metrics.http_req_failed.values.rate`.
|
|
699
|
+
- Parse Locust `_stats.csv` mapping `Name`, `Request Count`, `Failure Count`, and `95%` percentile columns.
|
|
700
|
+
|
|
701
|
+
##### 7. Environment Boundary Safety Guard in `tools/generate_jmx_plan.py`
|
|
702
|
+
- Import `_is_safe_non_prod_host()` from `tools/api_fuzzer.py`.
|
|
703
|
+
- Reject target URLs pointing to production domains unless `SYNAPSE_ALLOW_PROD_HOSTS` or `SYNAPSE_ALLOW_PROD_LOAD=1` is explicitly set in the environment.
|
|
704
|
+
|
|
705
|
+
##### 8. NTP Clock Skew Compensation via Cross-Correlation in `tools/telemetry_correlator.py`
|
|
706
|
+
- Implement lag cross-correlation over a window of $\tau \in [-15\text{s}, +15\text{s}]$:
|
|
707
|
+
$$R_{xy}(\tau) = \sum_t x(t) y(t + \tau)$$
|
|
708
|
+
- Automatically detect the optimal lag offset $\tau^*$ that maximizes correlation, compensating for clock skew between the load injector and target server.
|
|
709
|
+
|
|
710
|
+
##### 9. Automated k6 Script Generation CLI (`--emit-k6`)
|
|
711
|
+
- Extend `tools/generate_jmx_plan.py` to synthesize production k6 scripts from OpenAPI specs, featuring `ramping-arrival-rate` executors, custom thresholds, and dynamic auth pools.
|
|
712
|
+
|
|
713
|
+
##### 10. Update `tools/distill_log.py` for Load Test Streams
|
|
714
|
+
- Add regex patterns to capture JMeter periodic summary logs, k6 check glyphs, and Locust throughput tables, preventing distillation loss during load test triage.
|
|
715
|
+
|
|
716
|
+
---
|
|
717
|
+
|
|
718
|
+
#### P2 — Medium-Priority Remediation Items
|
|
719
|
+
|
|
720
|
+
##### 11. Live Prometheus Connector in `tools/telemetry_correlator.py`
|
|
721
|
+
- Add `--prometheus-url` and query range extraction for node CPU (`rate(node_cpu_seconds_total[1m])`), memory, and connection pools, eliminating manual CSV export.
|
|
722
|
+
|
|
723
|
+
##### 12. Open Model Arrival-Rate ThreadGroups in JMeter Generator
|
|
724
|
+
- Support JMeter 5.5+ `Open Model Thread Group` (`org.apache.jmeter.threads.openmodel.OpenModelThreadGroup`) to schedule true request arrival rates ($\lambda$) and eliminate Coordinated Omission.
|
|
725
|
+
|
|
726
|
+
##### 13. Deep Root Cause Diagnostics
|
|
727
|
+
- Ingest database slow query logs (`EXPLAIN ANALYZE`) and JVM GC logs (`-Xlog:gc*`) to correlate latency spikes with specific long-running queries or Stop-The-World GC pauses.
|
|
728
|
+
|
|
729
|
+
##### 14. Real OTLP Telemetry Streaming & W3C Trace Injection
|
|
730
|
+
- Integrate standard OpenTelemetry Python SDK with `OTLPSpanExporter` and `OTLPMetricExporter`.
|
|
731
|
+
- Inject W3C `traceparent: 00-{trace_id}-{span_id}-01` headers into JMeter and k6 HTTP samplers to enable distributed trace correlation across downstream microservices.
|
|
732
|
+
|
|
733
|
+
##### 15. Headless Client-Side Memory Leak CI Verifier (`tools/check_memory_leaks.py`)
|
|
734
|
+
- Implement a headless CDP script that captures baseline heap metrics (`JSHeapUsedSize`, `Nodes`), iterates through a user workflow 10 times, invokes `HeapProfiler.collectGarbage`, and verifies that retained DOM nodes and memory drift remain within 5% of baseline.
|
|
735
|
+
|
|
736
|
+
---
|
|
737
|
+
|
|
738
|
+
## 6. Verification Method
|
|
739
|
+
|
|
740
|
+
To independently verify the observations, line citations, and tool behaviors documented in this report:
|
|
741
|
+
|
|
742
|
+
1. **Verify Persona Guard & Anti-Rationalization Protocol:**
|
|
743
|
+
```bash
|
|
744
|
+
sed -n '61,75p' ".agents/agents/perf-engineer.md"
|
|
745
|
+
```
|
|
746
|
+
2. **Verify k6 Shared Token Anti-Pattern in Canonical Skill:**
|
|
747
|
+
```bash
|
|
748
|
+
sed -n '70,90p' ".agents/skills/backend-load-and-stress-patterns/SKILL.md"
|
|
749
|
+
```
|
|
750
|
+
3. **Verify JMeter Default Think-Time (0ms) and Safety Flag Omission:**
|
|
751
|
+
```bash
|
|
752
|
+
grep -n -C 3 "think-time" "tools/generate_jmx_plan.py"
|
|
753
|
+
grep -n "on_sample_error" "tools/generate_jmx_plan.py"
|
|
754
|
+
```
|
|
755
|
+
4. **Verify Deterministic SLA Gating & Exit Code 1 on Breach:**
|
|
756
|
+
```bash
|
|
757
|
+
python3 tools/jtl_analyzer.py --sample --sla-p95 500 --max-error-rate 1.0
|
|
758
|
+
echo "Exit code: $?"
|
|
759
|
+
# Expected: 2.5% error rate breaches 1.0% SLA; returns Exit Code 1
|
|
760
|
+
```
|
|
761
|
+
5. **Verify Telemetry Pearson Correlation & Mock OTel Export:**
|
|
762
|
+
```bash
|
|
763
|
+
python3 tools/telemetry_correlator.py --sample --otel
|
|
764
|
+
# Expected: Pearson r reported for CPU/DB pool; static JSON printed to stdout
|
|
765
|
+
```
|
|
766
|
+
6. **Verify Omission of Performance SLA Gates in Central DoD:**
|
|
767
|
+
```bash
|
|
768
|
+
grep -n "Gate [0-9]" "tools/verify_all.py"
|
|
769
|
+
# Expected: Gates 0-9 only (no performance or SLA gate)
|
|
770
|
+
```
|
|
771
|
+
7. **Verify Deprecated Navigation Timing API & Missing INP in Web Vitals Fixture:**
|
|
772
|
+
```bash
|
|
773
|
+
grep -n "window.performance.timing" ".agents/skills/web-vitals-and-performance-profiler/SKILL.md"
|
|
774
|
+
grep -n -E "INP|type: 'event'" ".agents/skills/web-vitals-and-performance-profiler/examples/playwright_cwv_profiler_fixture.ts"
|
|
775
|
+
# Expected: Deprecated API present; 0 INP results
|
|
776
|
+
```
|
|
777
|
+
8. **Verify Gate 5 Blacklist Conflict with Heap Dumps:**
|
|
778
|
+
```bash
|
|
779
|
+
grep -n "heap dumps" "tools/verify_all.py"
|
|
780
|
+
# Expected: Gate 5 scanning for temporary files and heap dumps
|
|
781
|
+
```
|
|
782
|
+
|
|
783
|
+
---
|
|
784
|
+
*Report synthesized and certified by Principal Performance Test Architect & Staff SRE (Synapse Hub Architecture Group).*
|