Show HN: KeelTest – AI-driven VS Code unit test generator with bug discovery

VS Code Extension · AlphaAI Tests That Find Bugs Before ProductionGenerate pytest suites that actually run - and expose issues in your code. 90% average pass rate. Source bugs flagged with fix suggestions.Get Started FreeWatch DemoFree forever · 7 credits/month · No credit card required847Test suites generated127Source bugs discovered91%Average pass rate* Stats updated weekly. Based on active Alpha usage.Install from VS Code MarketplaceGet Started in 3 Simple StepsKeelTest is a VS Code extension that installs in seconds. No complex setup, no external services.1Open VS Code ExtensionsPress Ctrl+Shift+X (Windows/Linux) or Cmd+Shift+X (Mac) to open the Extensions view in VS Code.2Search for KeelTestType "KeelTest" in the search bar and click Install on the official KeelTest extension.3Right-Click and GenerateRight-click any Python file in your workspace and select "KeelTest: Generate Tests" to start.Install KeelTest ExtensionFree to install • Available on VS Code Marketplace • No credit card requiredWhy developers switch to KeelTestMoving beyond simple prompts. We combined static analysis with a multi-step verification pipeline to deliver production-grade tests.See it in actionFrom Code to Tests in 3 ClicksStart Free, Scale When ReadyJoin ... developers already on the waitlist for our upcoming premium tiers.StarterFor regular development$9.99/month30 credits4x MultiplierPriority queueUsage analyticsBug detectionProFor power users$19.99/month70 credits10x MultiplierPriority queueUsage analyticsBug detectionEarly access to featuresCurrent PlanFree7 credits / month1x BasePerfect for trying it outBasic queue accessBug detectionStart FreeDetailed ComparisonEverything you get with each tierFeatureFreeStarterProMonthly Credits73070Support QueueBasicPriorityPriorityBug Detection✅✅✅Usage Analytics❌✅✅Early Access❌❌✅* All plans include self-healing tests, deep static analysis, and smart dependency mapping.Real Pass Rates, Not marketingEvery test is executed in a sandbox before it reaches your editor. We don't just generate code; we deliver verified functionality.How far your credits goEstimated file generation per month based on complexityComplexityFreeStarterProSmall files (≤15 fn)Approx. 15 functions~7files~30files~70filesMedium files (~30 fn)Approx. 30 functions~3files~15files~35filesLarge files (~50 fn)Approx. 50 functions~1files~7files~17files1 credit = up to 15 functions. Larger files use proportionally more credits.Independent Benchmark ResultsTests That Actually TestWe tested leading AI models on generating unit tests for complex e-commerce logic. KeelTest's agentic approach-combining AI with static code analysis, test validation, and actual execution-achieved a staff-level score of 8.5/10, outperforming pure zero-shot prompts by 54%.Overall Quality ScoreStaff Engineer = 10KeelTest (Agentic)(Agentic AI + Static Analysis)8.5/10Model B (Claude Sonnet 4.5)(Zero Shot)5.5/10Model C (Grok Auto)(Zero Shot)3.5/10Model A (GPT 5.1)(Zero Shot)2.5/10Detailed Evaluation CriteriaCriteriaKeelTestModel BModel CModel AUnit Test IsolationMocking & Dependency InjectionEdge Case CoverageFollowing InstructionsTechnical CorrectnessDateTime/Float PrecisionExcellentPartialPoor/Missing8.5/10KeelTest ScoreStaff Level54%Better ThanBest Zero-Shot100%Unit Test IsolationProper Mocking95Tests GeneratedAll Passing"KeelTest demonstrates the deepest understanding of unit testing principles with proper isolation, comprehensive mocking, and dependency injection. It's what a staff engineer would produce."* KeelTest leverages advanced AI models enhanced with static code analysis, automated test validation, and real-time execution feedback-not just raw prompts.Frequently Asked Questions