abstractframework

Generative image and video runtimes for Apple Silicon (MLX): text/image-to-image, instruction editing, masked edit, text/image/reference-to-video, video editing, and restoration.

Code once, run everywhere across Ollama, LM Studio, MLX, HuggingFace, vLLM, OpenAI, Anthropic, and more with consistent APIs for tools, structured output, streaming, media, prompt caching, and secure OpenAI-compatible gateway deployment.

Modular Python voice I/O for AI applications. Text-to-speech, speech-to-text, and voice cloning with local-first defaults and remote provider support. One interface across Piper, Supertonic, OpenAI, OmniVoice, and more.

Diagram-based, durable AI workflows for AbstractFramework. A visual editor inspired by UE4 Blueprints with drag-and-drop nodes, portable .flow bundles, recursive subflows, and multi-agent orchestration. Gateway-first architecture.

The production HTTP control plane for durable AI runs. Start, schedule, observe, and replay AI workflows across any client — terminal, browser, tray, Telegram, or email. Replay-first architecture with append-only ledger.

Model-agnostic remote and local text-to-music/audio generation for the AbstractFramework ecosystem. Lightweight base install; optional local runtimes (ACE-Step, Diffusers).

Framework-agnostic React Teams interface for an Agora Hub: channels, threads, search, inbox, direct messages, files, moderation, and reputation. A static, direct client with no WUI backend, proxy, or credential store.

Multi-camera control for Python — tethered PTP bodies (Nikon, Sony), webcams, DWARF telescopes, and AbstractCore AI tool integration.

Agent Skills (SKILL.md) parsing, discovery, trust gating, and demand derivation for AbstractFramework