I Read All 70 Past DEV Challenges + 542 Winning Submissions - Here's the Bug Smash Playbook A developer analyzed 70 past DEV challenges and 542 winning submissions to create a data-driven playbook for the DEV Summer Bug Smash. The analysis reveals that winners consistently demonstrate five traits: a clear original concept, real utility or emotional payoff, polished execution, strong writing, and meaningful use of sponsor technology. The developer recommends entering both contest tracks and targeting Sentry-tagged repos for the easiest sponsor prize path. I spent the last few days doing something slightly obsessive: I opened every single past challenge on dev.to/challenges https://dev.to/challenges , clicked into each one, found the green Read Announcement button under the Winners Announced banner, and read every single winner announcement blog. Then I went deeper. I read individual winner posts to see what they actually built and why judges picked them . I decomposed 70 winner announcements and 542 individual winning submissions across writing challenges, build challenges, AI challenges, game jams, frontend sprints, hackathons, and OSS challenges. This post is the distilled, evidence backed playbook for winning DEV Summer Bug Smash https://dev.to/bugsmash . No fluff. No generic advice. Just patterns I extracted from judges own words, applied to every track of the Bug Smash. Jump to any chapter: Bug Smash has 23 winning slots across 5 prize categories. Max realistic winnings per person: $1,200+ . Enter BOTH tracks Clear the Lineup + Smash Stories . Most people only enter one, so the other pool is less competitive. One submission can sweep 4 prizes : $500 Sentry + $200 Clear the Lineup + $200 Google AI + $100 Runner Up = $1,000 from one post. Target Sentry tagged repos Sentry, Formbricks, GitButler, Zulip for the easiest sponsor prize path. Judges number one deciding factor from real quotes : writing quality. Spend as much time on the post as on the code. ⬆ Back to Navigation Across every challenge type, the same five traits show up in winner announcements over and over. They are not separate, they reinforce each other. A submission that nails all five is almost unbeatable. 1. A clear, single, original concept. Winners never try to do five things. They pick one strong idea and execute it well. Judges say things like "wholly unique entry with nothing else quite like it in the submission pool" @thehwang https://dev.to/thehwang , June Solstice Game Jam and "simple but elegantly executed" @newdawnera https://dev.to/newdawnera , same challenge . A focused concept is memorable; a kitchen sink project blurs into noise. 2. Real, demonstrable utility or emotional payoff. Judges reward projects that actually do something useful for someone, or that make the reader feel something real. Quote: "immediately useful to anyone looking to manage their finances better" @ritesh hiremath eb6abb681 https://dev.to/ritesh hiremath eb6abb681 , Agent.ai Challenge . Quote: "her journey so far has left a lasting impression" @tochi https://dev.to/tochi , WeCoded Challenge . Utility plus heart beats technical wizardry. 3. Polished execution, not just working code. A working prototype is the floor. Winners ship something that feels finished: clean UI, sensible defaults, no obvious rough edges. Quote: "masterfully executed and scripted, with great visuals and sound design" @iclaldogan https://dev.to/iclaldogan , June Solstice Game Jam . Judges notice the difference between "demo" and "shipped product." 4. Strong writing that carries the submission. This is the most under rated winner trait. DEV is a writing platform first. Judges repeatedly say writing quality was the deciding factor. Quote: "this one's writing quality carried it across the finish line, building a whole world with very few words" @newdawnera https://dev.to/newdawnera , June Solstice Game Jam . Even in build challenges, your post is half the grade. 5. A meaningful, documented use of the sponsor technology. When a sponsored prize category exists Sentry, Google AI, Snowflake, Algolia, etc. , winners do not just include the API. They explain how it was the right tool for a real problem. Quote: "shines by leveraging pgai Vectorizer to streamline systematic literature review through an elegant RAG pipeline" @fahminlb33 https://dev.to/fahminlb33 , Open Source AI Challenge with pgai . The "why" matters as much as the "what." Just as important, here are the failure patterns I saw judges call out, either in winner announcements by their absence or in the official rules. 1. Vague, generic prompts. "A todo app" or "a chatbot" with no twist. If your title could be the title of 50 other submissions, you will not stand out. Every winner had a specific angle, like "a to do list that only allows ONE task" @bridget amana https://dev.to/bridget amana , GitHub Copilot Challenge , not "another to do app." 2. Demo videos that hide the actual product. Judges cannot evaluate what they cannot see. Submissions with no screenshots, no demo video, no code snippets, no before and after, they get filtered out fast. Show the thing working, even if production is rough. 3. Writing that reads like a README. Judges are not grading your code documentation. They are grading a story. Posts that read like API docs lose. Posts that read like a real person explaining a real problem they cared about, those win. 4. Sponsor tech bolted on as an afterthought. If you are pursuing a sponsored prize, the sponsor tech must be load bearing, not decorative. Judges can tell the difference between "I added Sentry because it was required" and "Sentry's session replay showed me exactly which user action triggered the bug." The first loses; the second wins. 5. No measurable impact. In a bug fix challenge especially, "I fixed a bug" is not enough. You need before and after numbers: latency, memory, error rate, test coverage, p99 response time, crash free sessions. Without numbers, your impact claim is just an opinion. Different challenge types reward different things. Bug Smash is a hybrid of "build" and "writing" tracks, so both rows apply to you. | Challenge type | What judges reward most | Common winner quote pattern | |---|---|---| | Writing | Personal narrative plus concrete takeaways plus clean structure | "well written and heartfelt, vivid story" / "thoughtful takes" / "introspection and inspiration" | | Build / hackathon | Working demo plus clear use case plus polish | "immediately useful" / "simple, useful, and clever" / "production ready" | | AI / agent | Real problem solved with the AI tool plus measured results | "23% better accuracy" / "sub 10 second real time experience" / "leveraged tool to streamline" | | Frontend | Visual polish plus accessibility plus clean code | "clean aesthetic" / "thoughtfulness on creating a mini stage" / "immersive digital experience" | | Game jam | Original mechanic plus thematic fit plus fun | "wholly unique entry" / "had us clicking away for maybe a bit too long" / "simply fun to play" | | OSS / bug fix relevant | Real impact plus clean PR plus clear writeup of root cause | "focus on real world" / "well executed" / "practical" | ⬆ Back to Navigation Everything below is pulled directly from the official Bug Smash page https://dev.to/bugsmash . If anything conflicts with the live page, the live page wins, recheck it before you submit. | Milestone | Date | Note | |---|---|---| | Challenge starts | July 14, 2026 | You can start now | | Submissions due | Aug 23, 2026 6:59 AM UTC Aug 24 | Hard deadline, no extensions | | Winners announced | Sep 17, 2026 | Dedicated winner announcement post on DEV | | Prize contact | Within 10 business days of announcement | Email associated with your DEV profile | Bug Smash is unusual: it has TWO independent prompt tracks , and you can submit to BOTH . Each submission is judged separately. This is the single biggest strategic lever in the whole challenge. Most participants will only enter one track, which means the other track's prize pool is less competitive. Enter both. | Track | What you submit | Eligible prize categories | |---|---|---| Clear the Lineup | A definitive bug fix or performance optimization in an existing codebase. Must include exact code changes PR link or code snippets . No new features, no typos, no doc only fixes. | Clear the Lineup 5 x $200 , Best Sentry 3 x $500 , Best Google AI 5 x $200 , Runner Up 5 x $100 | Smash Stories | A writeup of a legendary bug you caught or a clever performance optimization you previously pulled off. Technical detail required. | Smash Stories 5 x $200 , Runner Up 5 x $100 | There are 23 winning slots in total. Your single submission can win in multiple categories at once e.g. one Clear the Lineup submission with Sentry + Google AI usage can sweep 4 prizes . Plan your submissions to maximize coverage. | Prize category | Count | Cash | Other perks | Which track | |---|---|---|---|---| | Clear the Lineup | 5 winners | $200 | DEV++ Membership plus badge | Clear the Lineup | | Smash Stories | 5 winners | $200 | DEV++ Membership plus badge | Smash Stories | Best Use of Sentry | 3 winners | $500 | Limited Sentry skateboard plus DEV++ plus badge | Clear the Lineup only | | Best Use of Google AI | 5 winners | $200 | DEV++ Membership plus badge | Clear the Lineup only | | Runner Up | 5 winners | $100 | DEV++ Membership plus badge | Either track | | Completion badge | Everyone | Participation badge | Any valid submission | Maximum realistic winnings per person:If you win Best Use of Sentry $500 plus Clear the Lineup $200 plus Best Use of Google AI $200 plus Runner Up $100 on a single submission, that's$1,000 from one post. Add a Smash Stories win $200 and you are at$1,200. Even winning just one Sentry category alone puts you in the top prize tier of the whole challenge. These are the exact criteria from the Bug Smash page. Judges are looking for: 1 Technical Execution, 2 Impact of Bug Fix or Optimization, 3 Writing Quality, 4 Use of Prize Category optional . Guest judge: Sergiy Dybskiy , Developer Experience Engineer at Sentry, alongside The DEV Team. Here is what each criterion really means and how to score on it: | Criterion | What judges actually mean | How to score 10/10 | |---|---|---| Technical Execution | Is the fix correct? Does it not introduce regressions? Is the code clean, tested, and following project conventions? Would a maintainer want to merge this? | Include tests. Include benchmarks. Follow the repo's CONTRIBUTING.md . Get a maintainer or peer to review it before you submit. Link the PR or commit hash. | Impact of Bug Fix / Optimization | How much does this matter? Did you fix a paper cut or a bleeding artery? Is the improvement measurable? Does it affect real users? | Provide before and after numbers latency, memory, error rate, p99, crash free sessions . Quantify how many users were affected. Show the bottleneck profile. | Writing Quality | Can a reader who isn't you understand what happened? Is the post structured, paced, and engaging? Does it tell a story rather than dump a changelog? | Use the narrative arc: setup, conflict, investigation, breakthrough, resolution. Add screenshots and code snippets. Read your draft aloud once before publishing. | Use of Prize Category | For Sentry: did you actually use Sentry's features session replay, Seer RCA, logs, traces, MCP ? For Google AI: did you use Gemini to find, diagnose, or fix the bug, and explain how? | Dedicate a section of your post to "How I used sponsor." Show screenshots of the actual Sentry dashboard or Gemini session that helped you solve the problem. | Every submission must be a post on DEV.to with the bugsmash tag, published using the official submission template for your track. Each submission needs its own post, you cannot combine multiple entries in one post. Submissions do not have to be in English to earn a completion badge, but they DO have to be in English to be eligible for prizes. Important rules: bugsmash26 ⬆ Back to Navigation This is the core of the post. For each of the 5 prize tracks, I will lay out: who typically wins based on past patterns , how to pick your target, exactly what to build or write, how to structure the post, and what judges will look for. What it is: Submit a definitive bug fix or performance optimization to an existing codebase. Include the exact code changes PR link or snippets . No new features, no typo fixes, no documentation only changes. This is the marquee track of the whole challenge. Who wins this kind of track: In every past bug fix style or OSS style challenge, judges picked submissions with a a clearly described root cause, b a measurable before and after delta, c a clean PR that a maintainer would actually want to merge, and d a writeup that reads like a detective story, not a changelog. Pure code quality without the writeup loses; pure writeup without real code loses. You need both. Your single biggest decision. A great bug pick makes winning almost easy; a poor pick makes it impossible. Use this filter: | Filter | Why it matters | How to verify | |---|---|---| | Real impact on users | Judges reward fixes that affect real people, not toy problems | Check the issue thread: reactions, comments, "this blocked me" reports | | Measurable improvement | You need numbers for the "Impact" judging criterion | Can you benchmark before and after? Latency, memory, error rate, crash rate? | | Reproducible in isolation | You will need to demo this in your post | Can you write a one paragraph reproduction recipe? | | Not already being worked on | Avoids wasted effort and maintainer conflict | Search the repo for open PRs and assigned issues | | In a repo you can actually run | You cannot fix what you cannot build | Clone, install, run tests, all green within 2 hours of work | | Bounded scope 1 to 3 days | You have 5 weeks total and want to enter multiple tracks | Avoid issues labeled "epic" or with 50+ comments of unresolved debate | The Bug Smash page lists these repos as good candidates: crates.io, Sentry, e18e, Element, Formbricks, GitButler, Hugo, LocalSend, npmx, Nuxt, Typst, uv, Zulip. Repos already tagged "Sentry" use Sentry internally, targeting those gives you a head start on the Best Use of Sentry $500 prize Sentry, Formbricks, GitButler, Zulip . | Repo | Language | Why it is a strong pick | |---|---|---| Sentry getsentry/sentry | Python | Tagged "Sentry" on the official list. Fix a bug IN Sentry using Sentry itself. Highest sponsor alignment. | Formbricks | TypeScript | Tagged "Sentry." Active OSS survey tool. Real user facing bugs in forms, surveys, routing. | GitButler | Rust | Tagged "Sentry." Git client, lots of edge cases in branch handling, perf wins possible in large repos. | Zulip | Python | Tagged "Sentry." Mature chat platform with documented performance issues in search and notifications. | uv | Rust | Astral Python package manager. Extremely active. Perf focused, any p99 improvement is a story. | Hugo | Go | Static site generator. Build time perf wins are easy to benchmark build a 1000 page site, time it . | Nuxt | TypeScript | Huge user base, easy to find a real world bug affecting many developers. | Typst | Rust | Modern typesetting. Parser and renderer bugs are visually demonstrable, great for screenshots. | crates.io | Ruby | The Rust package registry. Real backend bugs with user impact. | LocalSend | Dart/Flutter | Cross platform file sharing. Network related bugs are dramatic and demo able. | README , CONTRIBUTING.md , and CODE OF CONDUCT.md . Read its LLM guidelines if they exist some repos now have explicit AI policies .Use this exact section order. It maps 1:1 to the judging criteria and to the patterns I saw in past winning submissions: | Section | Length | What goes here | |---|---|---| Title | 1 line | Specific. E.g. "I shaved 87% off Zulip search index cold start by caching tokenized streams." Not "Bug fix in Zulip." | TL;DR | 3 to 4 sentences | What was broken, what you changed, the measured impact, the link to PR. Make judges want to read on. | The bug setup | 200 to 400 words | What the world looked like before. Who was affected. How you discovered it. Make the reader feel the pain. | Investigation conflict | 300 to 500 words | How you reproduced. What you tried first that didn't work. The breakthrough moment. Include screenshots of Sentry, Gemini sessions, profilers, whatever you actually used. | The fix resolution | 200 to 400 words | The actual code change. Diff or PR link. Why you chose this approach over alternatives. Tradeoffs you considered. | Impact numbers | 150 to 300 words | Before and after benchmarks. Tables and charts if you can. Real world implications: how many users, how much money or time saved. | How I used Sentry if pursuing prize | 150 to 250 words | Specific Sentry features: session replay, Seer RCA, logs, traces, MCP. Screenshot of the dashboard that helped you. | How I used Google AI if pursuing prize | 150 to 250 words | Specific Gemini API call, AI Studio session, or Gemini CLI usage that helped you find, diagnose, or fix the bug. | What I learned | 100 to 200 words | Honest reflection. What surprised you. What you would do differently. This is what makes the post feel human. | Reproduction | 50 to 100 words plus code | Exact steps for someone else to reproduce the before and after. This proves the fix is real. | PR link plus repo | 1 line | Direct URL to your PR or commit hash. Even if not merged, this is required. | What it is: A writeup of a legendary bug you caught or a clever performance optimization you previously pulled off. No new code required, pure storytelling with technical depth. This is the lowest effort, highest ROI track in the whole challenge if you already have a war story. Strategic value: Smash Stories lets you submit a second entry with very little additional work. If you have ever debugged something memorable a 3am prod incident, a flaky test, a memory leak that took a week to find , this track is essentially free money. Most participants will skip it because it sounds "less technical," which means the pool is smaller and your odds are better. The best Smash Stories have four ingredients. If your story has all four, you have a strong contender. 1. Stakes. Someone cared about the outcome. Production was down. Users were affected. Money was being lost. A launch was at risk. Without stakes, the story is just "I found a bug and fixed it." 2. Mystery. The cause was not obvious. The first three theories were wrong. There was a "wait, what?" moment when the real cause surfaced. This is what makes the story worth reading. 3. Technical depth. You can explain the actual mechanism: a race condition, a memory leak pattern, a caching bug, an off by one in a boundary case. Vague "I eventually found the issue" loses; specific "the cookie was being set with SameSite=None in a redirect chain" wins. 4. Resolution with a lesson. The story ends with a concrete change: a test added, an alert configured, an architecture changed, a process improved. The reader takes away something they can apply to their own work. | Section | Length | What goes here | |---|---|---| Title | 1 line | Hook plus stakes. E.g. "The 3am mystery: why our checkout page worked for everyone except one user in Singapore." | The setting | 100 to 200 words | Context: what the system was, what your role was, what normal looked like before the bug. | The first sign | 150 to 300 words | How you noticed. An alert fired. A user reported it. A metric spiked. Make the reader feel the moment. | The investigation | 400 to 700 words | The detective work. What you checked first. What you ruled out. The wrong theories. The breakthrough. Include real logs, stack traces, profiler output, dashboards. | The root cause | 200 to 400 words | The actual mechanism, explained clearly enough that a developer who has never seen your codebase can understand it. Diagrams help. | The fix | 150 to 300 words | What you changed. Why this fix and not another. What tradeoffs you accepted. | The aftermath | 100 to 200 words | What happened next. Did the fix hold? What monitoring did you add? What process changed? | What I learned | 100 to 200 words | The honest reflection. This is the section judges remember. Quote from a past winner: "introspection and inspiration." | Reproduction bonus | 50 to 100 words plus code | A minimal reproduction if you can construct one. Not required but massively boosts credibility. | What it is: The largest individual cash prize in the whole challenge. Tell us how you used Sentry to improve an application. Only Clear the Lineup submissions are eligible. 3 winners each receive $500 plus a limited edition Sentry skateboard plus DEV++ Membership plus exclusive badge. Why this is the best ROI in the challenge: Only 3 winners, but the prize is 2.5x the standard $200. AND you get a physical skateboard. AND, this is the key, only people who use Sentry in their Clear the Lineup submission are eligible , which dramatically shrinks the competitor pool. If you are doing Clear the Lineup anyway, adding Sentry is the single highest leverage move in the entire challenge. Judges can tell the difference between "I added Sentry because the rules said to" and "Sentry solved my problem." The second one wins. Here is how to make Sentry load bearing in your fix: 1. Session Replay. Capture a replay of the bug happening. In your post, embed a screenshot of the replay with the exact frame where the bug triggers. This is gold for judges, they can SEE the bug. 2. Seer AI powered Root Cause Analysis . After Sentry captures the error, run Seer on it. Screenshot Seer's RCA output. In your post, quote what Seer told you and explain whether it was right, wrong, or partially right. This is the most under used Sentry feature and judges will notice. 3. Logs and Traces. Use Sentry Logs to find the error in context. Use distributed traces to find the slow span. Screenshot the trace waterfall. Annotate it in your post: "this span is the culprit." 4. AI Agent Monitoring if your app uses AI . If your fix is in an AI app, use Sentry's agent monitoring and conversation traces. Screenshot the trace showing the agent behavior that caused the bug. This is explicitly called out on the Bug Smash page as something judges want to see. 5. MCP server integration. If you use Claude Code or another MCP compatible dev tool, connect Sentry's MCP server. Show in your post how you queried Sentry from your editor to debug. This is bleeding edge and will make your submission stand out. 🔧 Sentry Setup Checklist 8 items - Click to expand bugsmash26 /screenshots folder for use in your post.Target repos tagged "Sentry" on the official Bug Smash list. These already use Sentry in production, which means a the Sentry integration will work cleanly in your fork, and b you can find Sentry captured issues in their public issue tracker or by adding your DSN and exercising the app. Top picks: Sentry itself getsentry/sentry , Formbricks, GitButler, Zulip. What it is: Show how Google AI helped you track down a tricky bug or optimize your application. Only Clear the Lineup submissions are eligible. 5 winners each receive $200 plus DEV++ Membership plus badge. How this differs from Sentry: Sentry helps you OBSERVE the bug. Google AI helps you UNDERSTAND or FIX the bug. They are complementary, use both. Past winners in Google AI tracks used Gemini in one of three ways: 1 explaining a confusing stack trace, 2 generating candidate fixes they then tested, 3 writing test cases that reproduced the bug. Do not just say "I asked Gemini and it helped." Show the actual conversation. Quote Gemini's response. Explain whether you used its answer verbatim, modified it, or rejected it. The honest, specific story is what wins. Judges have seen too many "I used AI" claims that mean nothing. | Step | What to do | What to screenshot | |---|---|---| | 1. Encounter the bug | Reproduce it. Capture the stack trace, profiler output, or behavior. | The error itself in your terminal or browser. | | 2. Ask Gemini to explain | Paste the stack trace into Gemini API or AI Studio. Ask: "What could cause this? What are 3 hypotheses?" | Gemini response in full. Note which hypothesis was right. | | 3. Ask Gemini for candidate fixes | For the correct hypothesis, ask Gemini to suggest a fix. Ask for 2 to 3 alternatives with tradeoffs. | The candidate fixes. Note which one you used and why. | | 4. Test the fix | Apply Gemini suggested fix or your modified version . Run tests plus benchmarks. | Before and after benchmark output. | | 5. Ask Gemini for a regression test | Have Gemini draft a test that would have caught the original bug. Review and integrate it. | The test code plus passing test run. | What it is: 5 outstanding submissions across all submissions either track that demonstrate exceptional impact and quality. $100 plus DEV++ Membership plus badge each. Strategy: You do not "enter" the Runner Up category, it is a fallback for strong submissions that did not win their primary category. This means: every submission you make is automatically eligible. The implication is that you should make your submissions as strong as possible across ALL judging criteria, not just the ones for your target prize category. A submission that loses Clear the Lineup can still win Runner Up if its writing quality or impact is exceptional. Practical takeaway: Do not optimize narrowly. Even if you are gunning for the Sentry prize, write your post as if it could win Clear the Lineup on its own. The strength that gets you 2nd place in your target category is often enough to win Runner Up. ⬆ Back to Navigation These are direct quotes from the 70 winner announcement blogs I read. Each quote is from a DEV Team judge explaining why a specific submission won. I have grouped them by pattern. Internalize these, they are the literal vocabulary judges use. If your submission embodies these patterns, judges will recognize it because they themselves coined this language. Judges love submissions that solve one thing cleanly. The phrase "simple but" or "simple yet" shows up across game jams, build challenges, and writing challenges. Complexity does not win; clarity does. "Simple but elegantly executed, this one's writing quality carried it across the finish line, building a whole world with very few words."- judges on @thehwang , June Solstice Game Jam "Another simple, yet addicting, concept that had us clicking away for maybe a bit too long."- judges on @thehwang , June Solstice Game Jam "Simple, useful, and clever."- judges on @dhanushreddy29 , Bright Data Web Scraping Challenge "A minimalistic, one task only to do list designed to combat cognitive overload by allowing users to focus on a single task at a time."- judges on @bridget amana , GitHub Copilot 1 Day Build When a submission is genuinely unlike anything else in the pool, judges forgive rough edges. Originality is the single biggest "wow factor" lever. A weird, thoughtful, one of a kind entry beats a polished, derivative one almost every time. "- June Solstice Game Jam @thehwang Alan's Garden was a wholly unique entry with nothing else quite like it in the submission pool. You do not place flowers, you teach the garden a rule and watch the pattern grow itself, a puzzle built directly on Turing morphogenesis research." "- AI Challenge for Cross Platform Apps @varshithvhegde built an Interactive Bento Grid Portfolio that transforms the traditional portfolio into an immersive digital experience." In build challenges and OSS challenges, judges reward submissions that solve real problems for real people. The phrase "immediately useful" or "practical" is the highest praise. If your fix ships value to actual users on day one, you are ahead. "This agent delivers on all fronts and is immediately useful to anyone looking to manage their finances better."- judges on @ritesh hiremath eb6abb681 , Agent.ai Challenge "This is a beautifully executed project that is immediately useful for parents everywhere."- judges on @milewski , Open Source AI Challenge with pgai "- Bright Data Web Scraping Challenge @sarahokolo built a super practical price comparison tool that works across Amazon, eBay, and AliExpress... a perfect example of using web scraping to solve a real consumer need." This is the most important pattern in the entire dataset. Judges repeatedly said writing was the deciding factor, even in build challenges. Spend as much time on the post as on the code. Maybe more. "...this one's writing quality carried it across the finish line, building a whole world with very few words."- June Solstice Game Jam "- 2025 WeCoded Challenge @tochi shared a well written and heartfelt, vivid story of transformation, from navigating life in Nigeria oil industry to carving a new path in software engineering and cybersecurity. Full of introspection and inspiration, her journey so far has left a lasting impression." "- 2026 WeCoded Challenge @it is margarita tells the story about three generations of a family... Their reflection on what equity actually looks like is one of the most thoughtful takes we received." "- 2026 WeCoded Challenge @anchildress1 designed, wrote, narrated, and produced Carbon Trace: an immersive web experience... She voiced it herself in her native accent , backed it with 685 unit tests and 220 E2E tests, and brought full backend discipline to a frontend art project." When the code and the experience are clearly polished: clean UI, thoughtful defaults, sound design, accessibility, judges notice. The word "masterfully" or "beautifully executed" is high praise. Polish is what separates a winning submission from a runner up. "Masterfully executed and scripted, with great visuals and sound design that immerse the player across a variety of puzzle types."- judges on @iclaldogan , June Solstice Game Jam "- 2025 WeCoded Challenge @wesleybertipaglia WeCoded landing page puts a spotlight on individual stories, we loved the clean aesthetic and the thoughtfulness on creating a mini stage for those who publish under wecoded." "- Open Source AI Challenge with pgai @fahminlb33 KawanPaper shines by leveraging pgai Vectorizer to streamline systematic literature review through an elegant RAG pipeline implemented by two Postgres functions." For sponsored prize categories, the sponsor technology must be central to the story, not bolted on. Winners explain specifically HOW the tech solved a problem. Losers mention the tech in one sentence and move on. "- Open Source AI Challenge with pgai @fahminlb33 KawanPaper shines by leveraging pgai Vectorizer to streamline systematic literature review through an elegant RAG pipeline implemented by two Postgres functions." "- Agentic Postgres Challenge @mayu2008 created FraudSwarn, a real time fraud detection system that innovates with hybrid search, combining pg text and pgvector to achieve 23% better accuracy than either method alone." "- Agentic Postgres Challenge @divyasinghdev built GitResume, an AI powered platform that transforms GitHub repositories into professional career insights. Four specialized agents analyze code quality, technology choices, career readiness, and innovation, turning what was previously a 1 to 2 minute sequential process into a sub 10 second real time experience." "- AssemblyAI Voice Agents Challenge @neilblaze and @achalbajpai created Wynnie, an AI shopping companion that transforms how people interact with e commerce through natural language. From multi language voice recognition supporting 50+ languages to..." When the submission fits the challenge theme AND uses the sponsor tech AND has clean execution, judges use the word "seamlessly." This is the holy grail, all three axes aligned. Aim for this. "- June Solstice Game Jam @mirshah12 Among Liars seamlessly integrates the jam themes into its core game mechanics: a social deduction game where six humans must sniff out a hidden Gemini powered seventh player. Built on Supabase Realtime, with a sleek black and white presentation, this one is a blast with a group of friends." "- Auth0 for AI Agents Challenge @async dime delivered a secure enterprise grade AI assistant that seamlessly manages Gmail, Google Calendar, web search, and document access all secured through Auth0 comprehensive security features." ⬆ Back to Navigation You have 5+ weeks from July 16 to August 23. This calendar assumes you can invest 8 to 12 hours per week 40 to 60 hours total and want to enter both Clear the Lineup AND Smash Stories with Sentry + Google AI integration. | Phase | Dates | Goal | Hours | |---|---|---|---| | Phase 1: Choose target | Jul 16 to 19 | Pick your Clear the Lineup repo and bug. Get maintainer buy in. Set up Sentry account. | 6 to 8h | | Phase 2: Reproduce + Sentry | Jul 20 to 26 | Reproduce the bug deterministically. Wire up Sentry SDK. Capture error, session replay, Seer RCA. | 10 to 12h | | Phase 3: Fix + Google AI | Jul 27 to Aug 2 | Implement the fix. Use Gemini API or AI Studio to analyze stack traces and propose candidate fixes. Benchmark before and after. | 12 to 15h | | Phase 4: Write Clear the Lineup post | Aug 3 to 9 | Draft the full post using the structure in Part 3.1. Include all screenshots. Get peer review. | 8 to 10h | | Phase 5: Smash Stories parallel | Aug 10 to 16 | Pick your best past war story. Write it using the structure in Part 3.2. Independent submission. | 6 to 8h | | Phase 6: Polish + submit | Aug 17 to 22 | Final polish on both posts. Submit early, never on the last day. Verify both are tagged bugsmash . | 4 to 6h | | Buffer | Aug 22 to 23 | If anything slips, this is your safety net. Do not plan to use it. | 0 to 4h | Day 1. Read this entire post end to end. Bookmark the Bug Smash page https://dev.to/bugsmash . Sign up for Sentry using promo code bugsmash26 . Sign up for Google AI Studio if you do not have an account. Day 2. Browse the recommended repos in Part 3.1. For each candidate, clone it, run its tests, and check its open issues. Spend 30 minutes per repo, max 5 repos. Day 3. Pick your top repo. Identify 3 candidate issues using the filter in Part 3.1. Score each on the 4 axes. Pick the winner. Day 4. Comment on the issue thread introducing yourself and your intent. Read CONTRIBUTING.md and any AI or LLM guidelines. Set up your fork and branch. Day 5 to 6. Reproduce the bug deterministically. Write a minimal test case that triggers it. Capture screenshots or video of the bug in action. Day 7. Install Sentry SDK in your fork. Configure DSN. Trigger the bug. Confirm Sentry captures the error. Save the event URL. Day 8. Open Session Replay. Trigger the bug again with replay enabled. Save the replay URL and a screenshot of the key frame. Day 9. Run Seer RCA on the captured event. Screenshot Seer output. Note whether Seer correctly identified the root cause. Day 10 to 11. Use Sentry Logs and Traces to find the error in context. Screenshot the trace waterfall. Annotate which span is the problem. Day 12. Buffer day. Catch up on anything that slipped. Update your issue thread with a progress comment. Day 13 to 14. Implement the fix. Commit early and often. Open a draft PR so maintainers can see your direction. Day 15. Paste your stack trace into Gemini API or AI Studio. Ask: "What could cause this? What are 3 hypotheses?" Screenshot Gemini response. Day 16. Ask Gemini for 2 to 3 candidate fixes with tradeoffs. Screenshot. Pick one or modify one . Note your reasoning. Day 17. Apply the candidate fix. Run tests. Run benchmarks. Capture before and after numbers latency, memory, error rate, etc. . Day 18. Ask Gemini to draft a regression test. Review, modify, integrate. Confirm the test fails before your fix and passes after. Day 19. Mark your PR ready for review. Ping the maintainer politely. Start outlining your DEV.to submission post. Day 20. Write the title and TL;DR. These are the most important 100 words of your entire submission. Iterate until they are excellent. Day 21. Write "The bug" and "Investigation" sections. Embed your screenshots bug, Sentry event, Seer RCA, Gemini session . Day 22. Write "The fix" and "Impact" sections. Include before and after benchmark table. Include PR link. Day 23. Write "How I used Sentry" and "How I used Google AI" sections. These are your sponsor prize plays. Day 24. Write "What I learned" and "Reproduction." Add a code block with exact repro steps. Day 25. Rest day. Do not look at the post. Let it sit. Day 26. Re read with fresh eyes. Cut 20% of the words. Tighten every paragraph. Ask a friend to review. Day 27. Brainstorm 3 candidate war stories from your past. Score each on Stakes, Mystery, Technical depth, Resolution. Pick the strongest. Day 28. Write the title and "The setting" and "The first sign." Hook the reader in the first 100 words. Day 29. Write "The investigation," this is the longest section. Include real logs, stack traces, dashboards if you have them. Day 30. Write "The root cause" and "The fix." Explain the mechanism clearly. Diagram if helpful. Day 31. Write "The aftermath" and "What I learned." End on the lesson, this is what judges remember. Day 32. Polish. Read aloud. Cut fluff. Ask a friend to review. Day 33. Submit BOTH posts today. Tag each with bugsmash and the appropriate submission template. Do not wait until the deadline. Day 34 to 35. If you submitted already, monitor for any DEV community feedback. Respond to comments. Engage genuinely, judges notice community response. Day 36. Re read both submissions. Fix any typos or broken image links you spot. Day 37 to 38. If anything went wrong earlier in the month, use this window to submit a simplified version. A late submission is better than no submission. Day 39. Final verification: both posts are public, both are tagged bugsmash , both use the official submission template, both have PR links Clear the Lineup or full stories Smash Stories . Day 40 Aug 22 . Submit day if you have not yet. Do not wait for Aug 23. Aug 23. Hard deadline. Anything not submitted by 6:59 AM UTC Aug 24 is not eligible. Done. ⬆ Back to Navigation These are enriched versions of the official templates from the Bug Smash page https://dev.to/bugsmash . Use them as your starting point in the DEV.to editor. Fill in every bracketed section. Delete the helper notes before publishing. --- title: " YOUR TITLE HERE - specific, e.g. Shaving 87% off Zulip search cold start by caching tokenized streams " published: true tags: bugsmash, language , framework , sponsor e.g. sentry cover image: URL to your cover screenshot --- Your Title TL;DR 3 to 4 sentences: what was broken, what you changed, the measured impact, link to PR. Make judges want to read on. - Repo : Link to the OSS repo you contributed to - PR / Commit : Direct URL to your PR or commit hash - Bug : One line description - Impact : e.g. "p99 latency dropped from 420ms to 54ms on a 10k message workspace" --- The Bug Setup 200 to 400 words. What the world looked like before. Who was affected. How you discovered it. Make the reader feel the pain. Screenshot of the bug in action URL Investigation 300 to 500 words. How you reproduced. What you tried first that did not work. The breakthrough moment. Include screenshots of Sentry, Gemini sessions, profilers, whatever you actually used. How I used Sentry 150 to 250 words. Specific Sentry features: session replay, Seer RCA, logs, traces, MCP. Screenshot of the dashboard that helped you. Quote what Seer told you. Sentry session replay showing the bug URL Seer RCA output URL How I used Google AI 150 to 250 words. Specific Gemini API call, AI Studio session, or Gemini CLI usage that helped you find, diagnose, or fix the bug. Screenshot the conversation. Gemini session URL The Fix 200 to 400 words. The actual code change. Diff or PR link. Why you chose this approach over alternatives. Tradeoffs you considered. diff - old code + new code Impact 150 to 300 words. Before and after benchmarks. Tables and charts if you can. Real world implications. | Metric | Before | After | Improvement | |--------|--------|-------|-------------| | p50 latency | X ms | Y ms | -Z% | | p99 latency | X ms | Y ms | -Z% | | Memory | X MB | Y MB | -Z% | | Error rate | X% | 0% | -100% | What I Learned 100 to 200 words. Honest reflection. What surprised you. What you would do differently. Reproduction 50 to 100 words plus code. Exact steps for someone else to reproduce the before and after. bash git clone