The latest articles on DEV Community by noname (@noname_242). https://dev.to/noname_242 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%2F3936429%2Fb4ebc20b-ecc7-4be2-957c-b2bbc0af4e1d.png https://dev.to/noname_242 en noname Sun, 17 May 2026 16:49:54 +0000 https://dev.to/noname_242/i-built-an-immune-system-for-ai-agents-using-textual-tui-fastapi-middleware-first-post-1no7 https://dev.to/noname_242/i-built-an-immune-system-for-ai-agents-using-textual-tui-fastapi-middleware-first-post-1no7 <h2> 👋 Hello DEV Community! </h2> <p>A real-time prompt monitoring proxy, multi-LLM consensus router, and high-fidelity TUI dashboard built in Python.</p> <p>This is my very first post here, and I'm thrilled to join the community! </p> <p>I recently set out to solve a major issue in the autonomous AI agent space: <strong>safety and prompt manipulation</strong>. Autonomous agents need to make API calls, read files, and write code, but they are highly vulnerable to prompt injections, jailbreaks, and tool abuse.</p> <p>To solve this, I built <strong><a href="https://github.com/SULphurix242/AegisOS" rel="noopener noreferrer">AegisOS</a></strong> — an inline, real-time middleware proxy and prompt-filtering shield, paired with a gorgeous terminal TUI dashboard to monitor and configure everything on-the-fly.</p> <p>Here is a technical deep-dive into how I built it, the architecture, and what I learned along the way!</p> <h2> 🛠️ The Architecture &amp; How it Works </h2> <p>AegisOS operates as an intercepting middleware proxy between your AI agent clients and their LLM gateways.</p> <h3> TUI Interface Hotkeys &amp; Screen Mapping </h3> <div class="table-wrapper-paragraph"><table> <thead> <tr> <th>Hotkey</th> <th>Screen Name</th> <th>Screen Purpose</th> </tr> </thead> <tbody> <tr> <td><strong>1</strong></td> <td>DASHBOARD</td> <td>Full system overview, telemetry gauges, and threat feed.</td> </tr> <tr> <td><strong>2</strong></td> <td>THREATS</td> <td>Deep catalog of all blocked exploit payloads and details.</td> </tr> <tr> <td><strong>3</strong></td> <td>AGENTS</td> <td>Monitor connected agent nodes, roles, and change isolation states.</td> </tr> <tr> <td><strong>4</strong></td> <td>MODELS</td> <td>Monitor active API gateways, latency, and consensus routing log.</td> </tr> <tr> <td><strong>5</strong></td> <td>TELEMETRY</td> <td>Real-time high-fidelity hardware consumption graphs.</td> </tr> <tr> <td><strong>6</strong></td> <td>SANDBOX</td> <td>Live exploit injector playground for custom and preset testing.</td> </tr> <tr> <td><strong>7</strong></td> <td>LOGS</td> <td>Continuous raw system operation and connection logs feed.</td> </tr> <tr> <td><strong>8</strong></td> <td>KEYS</td> <td>Manage, check, and test LLM provider keys directly inside the TUI.</td> </tr> <tr> <td><strong>9</strong></td> <td>MODEL CFG</td> <td>Customize provider models and trigger dynamic hot-reloads on-the-fly.</td> </tr> <tr> <td><strong>?</strong></td> <td>HELP</td> <td>Full keyboard command summary and references card.</td> </tr> <tr> <td><strong>L</strong></td> <td>LOCKDOWN</td> <td>Emergency global override—instantly isolate all active agents.</td> </tr> <tr> <td><strong>R</strong></td> <td>REFRESH</td> <td>Refresh current screen data.</td> </tr> <tr> <td><strong>Q / ESC</strong></td> <td>QUIT</td> <td>Safely close the TUI session.</td> </tr> </tbody> </table></div> <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%2F3fhh4cxilgi2w8tmhkax.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%2F3fhh4cxilgi2w8tmhkax.png" alt=" " width="800" height="587"></a></p> python security fastapi antigravity