This repository hosts a containerized Rust-based chatbot web application with a clean, responsive Bootstrap frontend. It runs a high-performance Axum-based HTTP server, supports pluggable local or remote LLM and TTS APIs, and manages chat history and sessions encrypted securely on disk. See design-privacy.md for details on storage privacy.

It started off as a simple single-threaded Flask app to test out vibe coding, but has been refactored into concurrent Rust with test coverage and security hardening as tools have improved.

Not a single line of code in this repository was written manually. Human work included: model/agent selections, design guidance, code review, testing, bug reporting, small edits, etc.

• Axum server with configurable CSRF-protected routes (enabled by default) for chat, authentication, set management, and TTS.
• chatbot-core crate encapsulating chat logic, provider abstractions, persistence, and session management.
• Static assets rendered with Minijinja and served from static/.
• Async provider implementations (OpenAI, Ollama) with streaming support and configurable defaults.
• Comprehensive integration tests (cargo test) covering routes, session flows, and external service stubs.

• chatbot-core/ – core business logic, configuration, providers, persistence, and session manager.
• chatbot-server/ – Axum HTTP server, route handlers, middleware, and integration tests.
• chatbot-test-support/ – shared fixtures and helpers for integration tests.
• data/ – gitignored persistence for chats, sessions, and users.
• static/ – static assets and Minijinja templates.
• docs/ – design and privacy documentation.
• temp/ – gitignored caches and scratch space (temp/.cargo, temp/.docker, temp/test-logs).

• Docker and Docker Compose.
• A .config.yml (use .config.yml.example as a starting point) with provider credentials and runtime settings.
• Any LLM/TTS endpoints referenced in the configuration should be reachable from the containers.

1. Copy .config.yml.example to .config.yml and adjust provider settings.
2. Add API keys to environment variables or copy .env.example to .env and adjust.
3. Run the integration and unit tests:

   docker compose run --rm tests cargo test

4. Build the runtime image and start services:

   docker compose up --build

   RUST_BUILD_TARGET=debug by default, you may want to set it to release.
5. Keep caches under temp/ as described in AGENTS.md.

• Provider and environment settings live in .config.yml; secrets should be injected through environment variables where possible.
• Environment variables from .env/.env.example are consumed by Docker Compose for development defaults.

• Overview and roadmap: docs/design.md
• Privacy posture and data handling guidelines: docs/design-privacy.md

• Follow the guidelines in AGENTS.md for coding style, testing strategy, and cache usage.
• Add or update integration tests before modifying core features.
• Run the Dockerized test suite and update relevant documentation before opening a pull request.
• Vibe Pull Requests welcome. We love AI here.