Build production AI agent runtimes in Rust with typed tools, provider-neutral LLM adapters, streaming, storage, and observability.

The open-source, local-first environment where you and the agents you already use work from the same maps, decisions, methods and retained results—across sessions and independent repositories.

Java middleware framework for cache, messaging, jobs, and governance extensions.

Provider-neutral skill pipeline (Claude + GitHub Copilot/GPT) for an autonomous plan -> build -> review -> ship workflow: 47 portable skills with parallel Claude and GPT adapters, 3 Claude-native exclusions, judges/graders/scorers, and install/build/release tooling.

Provider-neutral LLM router and model orchestration engine with auto-learning and budget zones

Provider-neutral autonomous ML research paper writer — ideation -> LaTeX writeup -> peer review -> anti-hallucination numerical audit. Spun out from hermes-sci to remove Hermes-runtime coupling.

Provider-neutral AI interaction contract compiler: portable templiqx/v1alpha1 contracts compile to deterministic, fingerprinted model interactions. One capability catalog behind CLI + MCP, so humans and agents share identical semantics.

Open-source reference architecture and template for provider-neutral AI-native game studio workflows.

Policy-aware model execution with portable routing, budgets, fallbacks, and receipts

Portable, provider-neutral model invocation contracts, adapters, and conformance tools

Legend State adapter for provider-neutral StateAdapter contracts.

A hybrid testing runner for AI agents: write high-level plans for fresh code, execute deterministic steps directly, and use LLMs only where runtime interpretation is needed.

Provider-neutral operating standard and reusable assets for AI-assisted collaboration projects

Provider-neutral Supabase database adapter exposing typed CRUD, schema management, and realtime subscriptions.