The latest articles on DEV Community by suraj kumar (@suraj_kumar_96bb8767435e2). https://dev.to/suraj_kumar_96bb8767435e2 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3954828%2F1f39d627-b41e-4efb-9c68-54dd2ab34dd9.jpeg https://dev.to/suraj_kumar_96bb8767435e2 en suraj kumar Thu, 28 May 2026 17:13:57 +0000 https://dev.to/suraj_kumar_96bb8767435e2/i-ship-one-ai-testing-feature-every-day-heres-what-6-days-looks-like-5cdn https://dev.to/suraj_kumar_96bb8767435e2/i-ship-one-ai-testing-feature-every-day-heres-what-6-days-looks-like-5cdn <p>I launched swarm-test 6 days ago. It's the first reliability testing tool <br> for multi-agent AI systems. Here's what I've shipped every single day:</p> <p>Day 0: Launch — 5 chaos tests, GitHub + PyPI<br> Day 1: Timeout resilience — found 22 new issues in my 14-agent system<br> Day 2: JSON export — another developer integrated it into his runtime gate within hours<br> Day 3: LangGraph adapter — doubled the addressable user base<br> Day 4: Sensitive data detection — 23 pattern types (AWS keys, JWT, credit cards)<br> Day 5: Per-agent health scores — every agent gets a 0-100 rating<br> Day 6: Before/after comparison — measure if changes actually helped</p> <p>[Include screenshots from each day]</p> <p>What I learned: shipping daily does three things:</p> <ol> <li>Forces you to keep features small and shippable</li> <li>Gives you something to post about every day</li> <li>Shows users the project is alive and actively maintained</li> </ol> <p>What's next: AutoGen adapter, GitHub Action, YAML config, <br> plugin system for community tests.</p> <p>Try it: pip install swarm-test<br> GitHub: github.com/surajkumar811/swarm-test .<a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fbzuoncbd9ohblhl9wl5r.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fbzuoncbd9ohblhl9wl5r.png" alt=" " width="800" height="486"></a></p> ai python opensource testing suraj kumar Wed, 27 May 2026 19:04:05 +0000 https://dev.to/suraj_kumar_96bb8767435e2/swarm-test-v015-per-agent-health-scores-8jf https://dev.to/suraj_kumar_96bb8767435e2/swarm-test-v015-per-agent-health-scores-8jf <p>My 14-agent system's results:</p> <ul> <li>OrchestratorAgent scored 4/100 — single point of failure with 3 collusion cliques</li> <li>FaceDetectorAgent scored 44/100 — high cascade depth</li> <li>EvolutionAgent scored 50/100 — 100% blast radius</li> </ul> <p>Stop guessing which agent to fix. Let the data tell you.</p> <p>5 features in 5 days. 84 tests. CrewAI + LangGraph.</p> <p>github.com/surajkumar811/swarm-test</p> <h1> AI #MultiAgent #Reliability #OpenSource </h1> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fiajkjjy2mqo6xstfh2wt.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fiajkjjy2mqo6xstfh2wt.png" alt=" " width="800" height="514"></a></p> agents ai showdev testing