<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Arman Siddiqui</title>
    <description>The latest articles on DEV Community by Arman Siddiqui (@armansiddiqui9).</description>
    <link>https://dev.to/armansiddiqui9</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4159746%2F365a6a00-f50f-4e4f-9218-d78978851662.png</url>
      <title>DEV Community: Arman Siddiqui</title>
      <link>https://dev.to/armansiddiqui9</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/armansiddiqui9"/>
    <language>en</language>
    <item>
      <title>The Sloth Shoebox</title>
      <dc:creator>Arman Siddiqui</dc:creator>
      <pubDate>Fri, 23 Oct 2026 21:00:57 +0000</pubDate>
      <link>https://dev.to/armansiddiqui9/the-sloth-shoebox-md4</link>
      <guid>https://dev.to/armansiddiqui9/the-sloth-shoebox-md4</guid>
      <description>&lt;p&gt;What I Built&lt;br&gt;
The Sloth Shoebox is a local-first expense tracker that turns UPI screenshots, receipt photos, and PDF bills into an organised, searchable ledger.&lt;/p&gt;

&lt;p&gt;Just drop your files in and let it handle categorisation, duplicate detection, monthly spending summaries, roommate splits, and natural-language queries.&lt;/p&gt;

&lt;p&gt;No cloud. No accounts. No subscriptions. No SMS permissions. Your financial data stays on your machine.&lt;/p&gt;

&lt;p&gt;Demo&lt;br&gt;
Run it locally and explore the working demo with 28 sample transactions.&lt;/p&gt;

&lt;p&gt;git clone &lt;a href="https://github.com/Armansiddiqui9/sloth-shoebox.git" rel="noopener noreferrer"&gt;https://github.com/Armansiddiqui9/sloth-shoebox.git&lt;/a&gt;&lt;br&gt;
cd sloth-shoebox&lt;br&gt;
pip install -r requirements.txt&lt;br&gt;
python -m sloth_shoebox&lt;br&gt;
Open &lt;a href="http://127.0.0.1:8337" rel="noopener noreferrer"&gt;http://127.0.0.1:8337&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;Code&lt;br&gt;
🐙 GitHub: &lt;a href="https://github.com/Armansiddiqui9/sloth-shoebox" rel="noopener noreferrer"&gt;https://github.com/Armansiddiqui9/sloth-shoebox&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;How I Built It&lt;br&gt;
Built with Python, FastAPI, and SQLite, with optional local AI through Ollama for receipt parsing and natural-language queries.&lt;/p&gt;

&lt;p&gt;The app works without AI too, using built-in fallbacks for a reliable, repeatable experience.&lt;/p&gt;

&lt;p&gt;Why Does Open Innovation Matter?&lt;br&gt;
Financial data is personal. Open-source tools and local AI make it possible to build useful automation without sending receipts to external AI services or depending on paid APIs.&lt;/p&gt;

&lt;p&gt;The goal is simple: make expense tracking easier without compromising privacy or control.&lt;/p&gt;

&lt;p&gt;The Lazy Payoff 🦥&lt;br&gt;
Less manual data entry, fewer duplicate transactions, easier expense splitting, and no more maintaining spreadsheets by hand.&lt;/p&gt;

&lt;p&gt;Spend less time tracking money and more time doing literally anything else.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>Leapstone</title>
      <dc:creator>Arman Siddiqui</dc:creator>
      <pubDate>Sat, 17 Oct 2026 13:59:38 +0000</pubDate>
      <link>https://dev.to/armansiddiqui9/leapstone-2952</link>
      <guid>https://dev.to/armansiddiqui9/leapstone-2952</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week2-2026-10-12"&gt;Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Leapstone&lt;/strong&gt; is an open-source command-line agent that fixes &lt;em&gt;your&lt;/em&gt; code when an API or SDK you depend on ships a breaking change.&lt;/p&gt;

&lt;p&gt;When a vendor ships a breaking release, thousands of teams read the same migration guide and make the same mechanical edits by hand. Dependabot bumps the version number, but nothing fixes the code. Leapstone does: it scans your repo for affected call sites, applies exact rewrites, asks a model only for the parts that need judgment, &lt;strong&gt;runs your own tests&lt;/strong&gt;, and opens a reviewable pull request with an honest confidence report.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Who would pay:&lt;/strong&gt; engineering teams that depend on many fast-moving APIs, and API vendors who want customers to adopt new versions without support tickets. The core is free (Apache-2.0). The paid side, Leapstone Pro, adds pack drafting from changelogs, pack signing, multi-repo batch runs, a team dashboard, and native Claude with managed API-key profiles. It's an open-core side hustle. There are no sales yet.&lt;/p&gt;

&lt;p&gt;The idea comes from the "Self-Maintaining APIs" request for startups: API providers shouldn't just announce changes, they should apply them.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;Try the whole pipeline offline in about a minute (a scripted stand-in plays the model, so no GPU or key is needed):&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;git clone https://github.com/Armansiddiqui9/leapstone-core &lt;span class="o"&gt;&amp;amp;&amp;amp;&lt;/span&gt; &lt;span class="nb"&gt;cd &lt;/span&gt;leapstone-core
pip &lt;span class="nb"&gt;install&lt;/span&gt; &lt;span class="nb"&gt;.&lt;/span&gt;
make demo     &lt;span class="c"&gt;# scan, then a dry-run migration of a sample repo&lt;/span&gt;
make &lt;span class="nb"&gt;test&lt;/span&gt;     &lt;span class="c"&gt;# 97 tests&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;


&lt;p&gt;&lt;code&gt;run&lt;/code&gt; is a dry run by default: it prints the diff and a report, and changes nothing in your repo.&lt;/p&gt;
&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Armansiddiqui9" rel="noopener noreferrer"&gt;
        Armansiddiqui9
      &lt;/a&gt; / &lt;a href="https://github.com/Armansiddiqui9/leapstone-core" rel="noopener noreferrer"&gt;
        leapstone-core
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      Dependabot for breaking API changes. Leapstone finds the code a new SDK version breaks, fixes it, runs your tests, and opens a pull request, using a model you run yourself. Open source, Apache-2.0.
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;🪨 Leapstone&lt;/h1&gt;
&lt;/div&gt;

&lt;p&gt;&lt;strong&gt;Hop to the new version, safely.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;a rel="noopener noreferrer nofollow" href="https://camo.githubusercontent.com/35b94db23d3ed026343335f74d52ce31e74b77ad7dab4e4b89f49f2026e0937f/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f6c6963656e73652d4170616368652d2d322e302d626c7565"&gt;&lt;img src="https://camo.githubusercontent.com/35b94db23d3ed026343335f74d52ce31e74b77ad7dab4e4b89f49f2026e0937f/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f6c6963656e73652d4170616368652d2d322e302d626c7565" alt="License"&gt;&lt;/a&gt; &lt;a rel="noopener noreferrer nofollow" href="https://camo.githubusercontent.com/96abf9b704f80578ea56dd10cab0d911c56d46dbec347f431ece9cf60ac175ad/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f707974686f6e2d332e31312532422d626c7565"&gt;&lt;img src="https://camo.githubusercontent.com/96abf9b704f80578ea56dd10cab0d911c56d46dbec347f431ece9cf60ac175ad/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f707974686f6e2d332e31312532422d626c7565" alt="Python"&gt;&lt;/a&gt; &lt;a rel="noopener noreferrer nofollow" href="https://camo.githubusercontent.com/71cb9a97360b8841ddfc125fa85687761845a6d44b48b4de9b527a2d2dce2930/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f7374617475732d616c7068612d6f72616e6765"&gt;&lt;img src="https://camo.githubusercontent.com/71cb9a97360b8841ddfc125fa85687761845a6d44b48b4de9b527a2d2dce2930/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f7374617475732d616c7068612d6f72616e6765" alt="Status"&gt;&lt;/a&gt; &lt;a rel="noopener noreferrer nofollow" href="https://camo.githubusercontent.com/2d7de65c9a00363e7e2c47d10f75088c2e74b2823fe717d6a88a20846beae16e/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f646570656e64656e636965732d507959414d4c2d6c6967687467726579"&gt;&lt;img src="https://camo.githubusercontent.com/2d7de65c9a00363e7e2c47d10f75088c2e74b2823fe717d6a88a20846beae16e/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f646570656e64656e636965732d507959414d4c2d6c6967687467726579" alt="Dependencies"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Leapstone is an open-source command-line agent that fixes &lt;strong&gt;your&lt;/strong&gt; code when an API or SDK you depend on ships a breaking change. It finds every affected spot, fixes the exact ones with deterministic rewrites, asks an AI model (that &lt;strong&gt;you&lt;/strong&gt; run, on your own machine) only for the parts that need judgment, &lt;strong&gt;runs your own tests&lt;/strong&gt;, and then hands you a reviewable pull request with an honest confidence report.&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Status: alpha (v0.3.0).&lt;/strong&gt; The pipeline is built and tested (97 automated tests). &lt;strong&gt;How well real open models perform has not been measured yet.&lt;/strong&gt; The bundled evaluation uses a scripted stand-in, not an AI. See &lt;a href="https://github.com/Armansiddiqui9/leapstone-core/docs/STATUS.md" rel="noopener noreferrer"&gt;&lt;code&gt;docs/STATUS.md&lt;/code&gt;&lt;/a&gt; for exactly what is and is not verified.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Why&lt;/h2&gt;
&lt;/div&gt;

&lt;p&gt;When a vendor ships a breaking release, every customer scrambles to update. The vendor writes a migration guide, thousands of teams read it, and thousands of developers make…&lt;/p&gt;&lt;/div&gt;


&lt;/div&gt;
&lt;br&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Armansiddiqui9/leapstone-core" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;br&gt;
&lt;/div&gt;
&lt;br&gt;


&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Open models, any runtime.&lt;/strong&gt; Leapstone bundles no model. It talks to any OpenAI-compatible endpoint you run (Ollama, vLLM, llama.cpp server, LM Studio) with a plain standard-library HTTP client. The only dependency is PyYAML.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Migration Packs.&lt;/strong&gt; A vendor describes a breaking release in one YAML file. Exact renames become deterministic codemods. Changes in &lt;em&gt;meaning&lt;/em&gt; become an instruction plus a before/after example for the model.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;The pipeline:&lt;/strong&gt; scan, codemod, model edit for judgment calls, your tests, retry, pull request. In repair mode the model sees its own failing edit and the test output.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Safety by design.&lt;/strong&gt; Edits only apply if the search text matches exactly once, inside code the model was shown. Secrets are replaced with placeholders before prompting. Pull requests go on a new branch only. There are hard limits on files, diff size and retries, and every run saves a &lt;code&gt;revert.patch&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Extensible.&lt;/strong&gt; A plugin entry point and a &lt;code&gt;--provider&lt;/code&gt; hook let extensions add commands and model providers. That's how the paid Pro package plugs in.&lt;/li&gt;
&lt;li&gt;Built with help from Claude as a coding assistant.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;strong&gt;Honest status:&lt;/strong&gt; alpha. The pipeline is tested (97 automated tests), but how well real open models perform has &lt;strong&gt;not been measured yet&lt;/strong&gt;. The bundled evaluation uses a scripted stand-in, not an AI.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Fixing someone's source code means reading it. With a closed API, every team would have to send its private code to a third party, and many companies can't do that. Open-weight models run through Ollama or vLLM keep the code on the developer's machine, so this tool can exist for those teams at all.&lt;/p&gt;

&lt;p&gt;Open also means you can swap models as better ones appear, run offline, and audit the safety rules (strict edits, secret redaction, new-branch-only PRs) instead of trusting a black box. And because Migration Packs are plain YAML files with free signature verification, anyone can contribute a pack for the API they know best.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>SlumberBloom is a small garden game that only grows when you leave your phone alone.</title>
      <dc:creator>Arman Siddiqui</dc:creator>
      <pubDate>Sat, 10 Oct 2026 08:12:04 +0000</pubDate>
      <link>https://dev.to/armansiddiqui9/slumberbloom-is-a-small-garden-game-that-only-grows-when-you-leave-your-phone-alone-43ba</link>
      <guid>https://dev.to/armansiddiqui9/slumberbloom-is-a-small-garden-game-that-only-grows-when-you-leave-your-phone-alone-43ba</guid>
      <description>&lt;p&gt;What I Built&lt;/p&gt;

&lt;p&gt;SlumberBloom is a garden that only grows when you leave your phone alone.&lt;/p&gt;

&lt;p&gt;You tap “Tuck In,” put the phone down and go to sleep. When you wake up, the app sees how long you were away. If it was at least 30 minutes, you get a Night Block: a SHA-256 hash chained to your previous night. Claim it as Morning Dew, settle it at Dawn and it becomes Bloom, which feeds your Baby Moonflower.&lt;/p&gt;

&lt;p&gt;It rewards you for not using it. There’s no tracking, no microphone, no tokens and no wallets. Your hours of patience are verifiable, and a “Light the Lantern” button re-checks any block with a Merkle proof, right in your browser.&lt;/p&gt;

&lt;p&gt;This is a v1 that proves the core idea. The big vision (eras, a world simulation, outdoor photo quests) is still being built, and what’s here is the working foundation.&lt;/p&gt;

&lt;p&gt;Code&lt;/p&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/Armansiddiqui9" rel="noopener noreferrer"&gt;
        Armansiddiqui9
      &lt;/a&gt; / &lt;a href="https://github.com/Armansiddiqui9/slumberbloom" rel="noopener noreferrer"&gt;
        slumberbloom
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      Proof-of-patience garden - your sleep becomes verifiable Night Blocks with SHA-256 + Merkle
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;SlumberBloom&lt;/h1&gt;
&lt;/div&gt;
&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Proof of patience.&lt;/strong&gt; Turn sleep away-time into a verifiable garden — SHA-256 Night Blocks, RFC-6962 Merkle proofs, Charter-supply math. No tokens, no wallets. Just your time, made beautiful.&lt;/p&gt;
&lt;/blockquote&gt;
&lt;p&gt;&lt;a rel="noopener noreferrer" href="https://github.com/Armansiddiqui9/slumberbloom/design/screens/garden-desktop.png"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fraw.githubusercontent.com%2FArmansiddiqui9%2Fslumberbloom%2FHEAD%2Fdesign%2Fscreens%2Fgarden-desktop.png" alt="Garden desktop"&gt;&lt;/a&gt;&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;What it is&lt;/h2&gt;
&lt;/div&gt;
&lt;p&gt;A local-first garden game. You tuck your phone in → time passes (≥30 min, ≤10h/day) → wake → your Night Block mints (SHA-256 hash) → claim → Dawn settles → Lantern verifies → water → bloom.&lt;/p&gt;
&lt;p&gt;Every block is proof. The rules are public math, not magic numbers. Your garden is yours — no server trust needed to verify it.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;AI inside (on-device, never touches consensus)&lt;/h2&gt;
&lt;/div&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;llama3.2:3b&lt;/code&gt; → quest text + garden journal&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;moondream&lt;/code&gt; → photo judging (strict JSON, graceful fallbacks)&lt;/li&gt;
&lt;li&gt;All local — photos never leave your device.&lt;/li&gt;
&lt;/ul&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;What you see&lt;/h2&gt;

&lt;/div&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Garden&lt;/strong&gt; — living specimen, growth stage, sunlight, bloom&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Bedtime / Tuck In&lt;/strong&gt; — start the night, see your window&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Morning Dew&lt;/strong&gt; — landing + protocol intro&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;…&lt;/p&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/Armansiddiqui9/slumberbloom" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;p&gt;How I Built It&lt;/p&gt;

&lt;p&gt;The code is in three layers: a pure TypeScript engine (hashing, Merkle proofs, block rules), a local adapter that runs the game in the browser, and a React PWA on top. The UI never touches the rules directly.&lt;/p&gt;

&lt;p&gt;Done in v1: the Tuck In → Dew → Claim → Dawn → Lantern loop, the rules (under 30 minutes is dust, 10-hour cap per block), a growing plant, and golden tests for the core math.&lt;br&gt;
Next: Charter and Census math, the Sky Draw, wilt and revive, and a World Sim.&lt;br&gt;
Open-source AI (in progress): small local models through Ollama. llama3.2:3b would write quests and journal lines, and moondream would judge a photo quest, which is the real “touch grass” part. They stay out of the core rules, with template fallbacks if the model is off.&lt;/p&gt;

&lt;p&gt;Why Does Open Innovation Matter?&lt;/p&gt;

&lt;p&gt;The pitch is "You don’t have to trust me; you can check.” That only works if the rules are open, so anyone can re-run the hash and verify a proof without trusting a server.&lt;/p&gt;

&lt;p&gt;It also matters for the AI. A photo from your day is personal. Open-weight models running locally mean the image goes to localhost and nowhere else, which a closed API couldn’t promise.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>dotnet</category>
      <category>opensource</category>
    </item>
    <item>
      <title>On Call Hero</title>
      <dc:creator>Arman Siddiqui</dc:creator>
      <pubDate>Mon, 05 Oct 2026 04:49:45 +0000</pubDate>
      <link>https://dev.to/armansiddiqui9/on-call-hero-13il</link>
      <guid>https://dev.to/armansiddiqui9/on-call-hero-13il</guid>
      <description>&lt;p&gt;On-Call Hero — Durable AI SRE Agent&lt;/p&gt;

&lt;p&gt;This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend&lt;/p&gt;

&lt;p&gt;What I Built&lt;/p&gt;

&lt;p&gt;I built On-Call Hero, an AI SRE agent that helps investigate and respond to incidents.&lt;/p&gt;

&lt;p&gt;The main thing I wanted to solve was reliability. What happens when an AI agent is in the middle of handling an incident and the worker crashes or the server restarts?&lt;/p&gt;

&lt;p&gt;On-Call Hero uses Temporal to make the workflow durable. It can investigate an issue, propose a fix, wait for human approval, execute it, verify the result, and roll back when needed.&lt;/p&gt;

&lt;p&gt;Even if the worker is killed in the middle of an incident, the workflow can resume from where it stopped.&lt;/p&gt;

&lt;p&gt;Demo&lt;/p&gt;

&lt;p&gt;Everything runs locally.&lt;/p&gt;

&lt;p&gt;make up&lt;/p&gt;

&lt;p&gt;Then open &lt;a href="http://127.0.0.1:8000" rel="noopener noreferrer"&gt;http://127.0.0.1:8000&lt;/a&gt; and try the S1 Bad deploy scenario.&lt;/p&gt;

&lt;p&gt;You can also test recovery with:&lt;/p&gt;

&lt;p&gt;make kill-worker&lt;br&gt;
make up&lt;/p&gt;

&lt;p&gt;Code&lt;/p&gt;

&lt;p&gt;GitHub: &lt;a href="https://github.com/Armansiddiqui9/oncallhero" rel="noopener noreferrer"&gt;https://github.com/Armansiddiqui9/oncallhero&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;How I Built It&lt;/p&gt;

&lt;p&gt;The project uses Python, Temporal, FastAPI, SSE, and LiteLLM. It can also run with local models through Ollama.&lt;/p&gt;

&lt;p&gt;I added human approval before actions are executed, along with plan validation, prompt-injection protection, idempotency, budgets, audit history, and rollback handling.&lt;/p&gt;

&lt;p&gt;Why Open Innovation Matters&lt;/p&gt;

&lt;p&gt;Most of the project is built around open-source tools, which made it possible to experiment with durable AI agents locally and understand every part of the system.&lt;/p&gt;

&lt;p&gt;I'm hoping to keep improving it and would love feedback from the community.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>weekendchallenge</category>
      <category>hf26challenge</category>
    </item>
  </channel>
</rss>
