https://dokimore.com/blog Practitioner notes on test automation, API testing, performance, security scanning and accessibility, written by the team that builds the platform. en Wed, 16 Sep 2026 00:00:00 GMT https://dokimore.com/blog/small-local-models-benchmark-250-cases https://dokimore.com/blog/small-local-models-benchmark-250-cases Wed, 16 Sep 2026 00:00:00 GMT Dokimore Our test intelligence runs on local models, so swapping a model is a product change. We ran qwen3.5:4b, qwen3:1.7b, granite3.3:2b, llama3.2:3b and llama3.2:1b through the same 250 labelled cases and compared them pairwise. Accuracy ranged from 62.8% to 18.4%. The fastest model was not the most accurate, and the gap was statistically significant. https://dokimore.com/blog/react-spa-readable-by-ai-answer-engines https://dokimore.com/blog/react-spa-readable-by-ai-answer-engines Wed, 16 Sep 2026 00:00:00 GMT Dokimore Most AI crawlers do not run JavaScript, so a Vite and React marketing site is invisible to them. We prerender every sitemap route at build time with Playwright, assert the head tags and JSON-LD on each page, name every crawler in robots.txt, and force noindex while the site URL is still a placeholder. Zero new packages. Here is the mechanism, and what it does not solve. https://dokimore.com/blog/how-we-test-dokimore-with-dokimore https://dokimore.com/blog/how-we-test-dokimore-with-dokimore Thu, 06 Aug 2026 00:00:00 GMT Dokimore Our own release pipeline runs on the platform we sell: route-reachability drift tests keep marketing honest about shipped features, a plan-catalog drift test keeps pricing aligned with enforced entitlements, and every release passes our own security and regression gates. https://dokimore.com/blog/evidence-backed-security-findings https://dokimore.com/blog/evidence-backed-security-findings Tue, 04 Aug 2026 00:00:00 GMT Siva M The problem: vulnerability scanners flood teams with unconfirmed findings, and triage eats the time that should go to fixing. Our approach: an attack-confirmation engine that verifies findings with real evidence: the request, the response and the proof, bundled with every report. https://dokimore.com/blog/true-cost-of-multi-tool-qa https://dokimore.com/blog/true-cost-of-multi-tool-qa Sat, 01 Aug 2026 00:00:00 GMT Siva M An API client here, a device farm there, a scanner, a load tool, a reporting layer, each with its own seats, silo and renewal. Add up your stack's list prices and the overlap you pay for twice, then compare one platform, one login, one bill. https://dokimore.com/blog/smart-test-selection-ci-pipeline https://dokimore.com/blog/smart-test-selection-ci-pipeline Tue, 28 Jul 2026 00:00:00 GMT Siva M A scenario walkthrough: a mid-size team runs 2,400 tests on every commit and waits 45 minutes for signal. Smart Test Selection maps code changes to affected tests and runs only what matters, with an auditable selection rationale. https://dokimore.com/blog/wcag-22-compliance-fast https://dokimore.com/blog/wcag-22-compliance-fast Fri, 24 Jul 2026 00:00:00 GMT Siva M WCAG 2.2 added 9 new success criteria and most teams don't know which apply to them. What matters for real applications, what automation can and cannot cover, and how VPAT generation turns scans into the document procurement wants. https://dokimore.com/blog/microsecond-api-timing https://dokimore.com/blog/microsecond-api-timing Mon, 20 Jul 2026 00:00:00 GMT Siva M When your API budget is 20ms, a regression from 3ms to 9ms is invisible at millisecond precision until it compounds. Our API studio measures at microsecond precision so you catch regressions while they are still cheap to fix. https://dokimore.com/blog/migrate-selenium-tests https://dokimore.com/blog/migrate-selenium-tests Wed, 15 Jul 2026 00:00:00 GMT Siva M A scenario walkthrough of Migration Setup: import an existing Selenium, Postman, Insomnia, or JMeter asset base, map it to the platform's step model, and re-run it with self-healing underneath. The report makes "we didn't lose one" checkable. https://dokimore.com/blog/automation-practice-that-lasts https://dokimore.com/blog/automation-practice-that-lasts Fri, 10 Jul 2026 00:00:00 GMT Siva M Automation initiatives die when the one engineer who owned them leaves. Practical patterns for durability: shared element repositories, review queues instead of tribal knowledge, golden suites, and gates that don't depend on memory. https://dokimore.com/blog/security-in-cicd-shift-left https://dokimore.com/blog/security-in-cicd-shift-left Sun, 05 Jul 2026 00:00:00 GMT Siva M Teams fear security gates will slow every deployment, so scanning gets deferred to a yearly pentest. Progressive scanning keeps security continuous and pipelines fast: quick checks on each pull request, deep scans nightly, evidence bundles on demand. https://dokimore.com/blog/beyond-pixel-diffs-visual-testing https://dokimore.com/blog/beyond-pixel-diffs-visual-testing Tue, 30 Jun 2026 00:00:00 GMT Siva M Screenshot comparison drowns you in false alarms from dynamic content, then misses real layout breaks. Structural element-drift detection compares what the DOM says moved, not only what the pixels say changed. https://dokimore.com/blog/inside-the-performance-engine https://dokimore.com/blog/inside-the-performance-engine Thu, 25 Jun 2026 00:00:00 GMT Siva M JMeter is powerful and miserable to maintain. Our performance engine keeps the thread groups, timers and assertions load engineers rely on, puts them behind a UI, adds P50, P95, P99 and Apdex, and exports to k6, JMeter and Locust. https://dokimore.com/blog/self-healing-human-in-the-loop https://dokimore.com/blog/self-healing-human-in-the-loop Sat, 20 Jun 2026 00:00:00 GMT Siva M Test maintenance is the top reason automation initiatives fail, and silent auto-healing re-breaks trust the first time it fixes a selector to the wrong element. Our approach: 15+ fallback strategies propose, a human approves, the element repository learns. https://dokimore.com/blog/privacy-first-local-ai https://dokimore.com/blog/privacy-first-local-ai Mon, 15 Jun 2026 00:00:00 GMT Siva M Most "AI testing" tools ship your code, selectors, and test data to third-party LLM APIs. Our smart selection, self-healing and defect analysis are deterministic and run locally. For security-conscious teams, that is not a limitation. https://dokimore.com/blog/compliance-evidence-continuously https://dokimore.com/blog/compliance-evidence-continuously Wed, 10 Jun 2026 00:00:00 GMT Siva M Audit season means weeks of screenshot archaeology. Build evidence collection into the pipeline instead: tamper-evident audit logs, scan evidence bundles, WCAG reports with VPAT export, generated continuously rather than once a year. https://dokimore.com/blog/intent-based-load-testing https://dokimore.com/blog/intent-based-load-testing Fri, 05 Jun 2026 00:00:00 GMT Siva M Writing a realistic load profile takes an experienced engineer half a day, so most teams test with flat, unrealistic ramps, or not at all. Intent-Based Load Testing turns a plain-English description of the traffic you expect into a reviewable load configuration. https://dokimore.com/blog/qa-command-center-dashboard https://dokimore.com/blog/qa-command-center-dashboard Sat, 30 May 2026 00:00:00 GMT Siva M Test results tell you what happened. Operating quality needs live posture: the dashboard unifies security findings, structured system logs, dependency circuit breakers, and tamper-evident audit trails. It is the view a QA lead opens first every morning. https://dokimore.com/blog/european-accessibility-act-ready https://dokimore.com/blog/european-accessibility-act-ready Mon, 25 May 2026 00:00:00 GMT Siva M The EAA became applicable in June 2025, extending accessibility obligations to e-commerce, banking, and consumer software sold into the EU, while the UK's Equality Act runs its own parallel track. What it means for engineering teams, and how continuous WCAG 2.2 scanning turns a legal risk into a checklist. https://dokimore.com/blog/gdpr-safe-test-data https://dokimore.com/blog/gdpr-safe-test-data Wed, 20 May 2026 00:00:00 GMT Siva M Copying production into staging is the most common test-data strategy in the industry, and under GDPR it multiplies your breach surface while failing data-minimization on its face. Synthetic generation and masking-based TDM give you realistic tests without the liability.