I am a dedicated developer focused on building intelligent systems and AI-powered applications. My work spans frontend a

github

Rahul Raj • Peerlist

<h3>Autonomous GitHub Coding Agent</h3>ShipIT Forge is an innovative, open-source autonomous coding agent designed to streamline your development workflow directly within GitHub. This powerful tool acts as a vigilant guardian for your repositories, automating critical tasks to enhance code quality, security, and efficiency. <ul><li>Automated Issue Resolution: ShipIT Forge intelligently reads GitHub issues and automatically generates pull requests to address them, saving your team valuable time and effort.</li><li>Security-Focused Code Reviews: Every pull request is meticulously reviewed through a security lens, identifying potential vulnerabilities before they can be merged into your codebase.</li><li>CI/CD Auto-Fixing: Never let a failing CI pipeline slow you down again. ShipIT Forge automatically detects and fixes common CI issues, ensuring your build processes remain robust.</li><li>Repository Auditing: Maintain the integrity and health of your codebase with comprehensive repository audits, identifying areas for improvement and potential risks.</li></ul>Leverage the power of leading AI models by bringing your own. ShipIT Forge seamlessly integrates with Vertex Gemini, AWS Bedrock, OpenAI, and Anthropic models, allowing you to choose the best fit for your needs. Deploy effortlessly via a GitHub App or as a GitHub Action, making it a flexible and self-hosted solution for enhanced developer productivity and code security.

ShipIT Forge — autonomous GitHub coding agent

SHIPIT Agent is a standalone Python agent library designed to give developers everything they need to build, deploy, and scale production-ready AI agents. Rather than being a restrictive framework, it is a highly observable and interchangeable library with a core runtime of under 400 lines of code, designed to stay completely out of the developer's way.Core Qualities and Capabilities:<ul><li>Pre-built Cognitive Architectures: The library provides "Deep Agents," which are autonomous, self-directing agent loops designed for complex tasks. These include:<ul><li>GoalAgent: Breaks down large goals into sub-tasks and evaluates progress against specific success criteria.</li><li>ReflectiveAgent: Critiques its own work and continuously revises it until a predefined quality threshold is met.</li><li>AdaptiveAgent: Capable of writing, auto-dedenting, and registering new Python tools dynamically during runtime.</li><li>Supervisor: Manages a team of worker agents by delegating tasks, reviewing output quality, and requesting revisions.</li></ul></li><li>ShipCrew Orchestration: You can compose specialized agents into Directed Acyclic Graph (DAG) workflows. Task execution can run sequentially, concurrently in parallel layers, or hierarchically, where an LLM dynamically acts as a coordinator to assign tasks and review work.</li><li>Super RAG Subsystem: A built-in Retrieval-Augmented Generation (RAG) system can be attached to any agent using a single parameter. It features hybrid search, document chunking, and thread-local source tracking that automatically outputs clean, DRK_CACHE-style source citations.</li><li>Provider Flexibility &amp; Reasoning Visibility: It supports 9 different LLM providers (including OpenAI, Anthropic, AWS Bedrock, Google, and local models via Ollama) behind one unified API. It automatically extracts and streams extended reasoning and thinking blocks (from models like o1 or Claude 3.7) to seamlessly provide live "Thinking" UI panels.</li><li>Extensive Tooling &amp; MCP Integration: The library allows you to easily mix regular Python functions, remote Model Context Protocol (MCP) servers, and over 28 built-in connector tools (such as Jira, Drive, Slack, and Playwright-powered web fetching). It also features parallel tool execution to significantly speed up multi-tool turns.</li><li>Advanced Memory &amp; Sessions: It provides a persistent chat experience through SessionStore and long-term knowledge retention through AgentMemory. This memory facade handles short-term conversation context, vector-based semantic facts, and structured entity tracking.</li><li>Production Guardrails: Built for real-world deployment, it features graceful error recovery (treating tool failures as messages for the LLM to correct rather than crashing the application), semantic tool discovery, and middleware hooks for cost tracking, rate limiting, and content filtering.</li><li>Live Chat REPL: It ships with an interactive terminal environment (shipit chat) where developers can index files mid-session, execute complex goals, and inspect tool usage and sources in real-time.</li></ul>