FireRoute Agent

FireRoute AI is a Dockerized general-purpose AI agent built for Track 1 of the AMD Developer Hackathon ACT II. It processes batches of natural language tasks by reading `/input/tasks.json`, solving each prompt and writing valid answers to `/output/results.json` before the container exits. The agent is designed to handle all required Track 1 capability areas: factual knowledge, mathematical reasoning, sentiment classification, text summarization, named entity recognition, code debugging, logical reasoning and code generation. Its architecture uses lightweight task routing and task-specific prompting to improve accuracy while keeping responses concise. FireRoute AI follows the official Fireworks AI runtime contract. It does not hardcode API keys or model IDs. Instead, it reads `FIREWORKS_API_KEY`, `FIREWORKS_BASE_URL` and `ALLOWED_MODELS` from the judging harness environment and routes all model calls through the provided base URL. The project is fully containerized and includes GitHub Actions validation to confirm the Docker contract online. The final image is intended to be built for `linux/amd64`, publicly pullable and ready for automated evaluation. FireRoute AI focuses on passing the accuracy gate first, then improving token efficiency through compact prompts, dynamic model selection and structured JSON output.