Epic County is an AI-native systems company. It builds Rust-based tools and infrastructure for AI coding agents, ships them as an ecosystem installed through a single CLI, and runs a research program on production theory for the AI economy.
The company has two arms: an engineering side developing open-source agent infrastructure, and Epic County AI Labs, a research division studying computational economics and AI-native production.
The company's research division.
AI Production Economics — a research program on production theory for the AI-native economy.
Manuscripts in preparation. Publications and reproducible artifacts will be released here when ready.
A Rust toolchain for AI-native development, installed and managed through the epiccounty CLI (epiccounty install all).
The tools are currently published under
@epicsagasand will migrate to the Epic County organization as the brand matures.
- llm-kernel — Rust foundation library for AI-native apps. A 20-provider, 351-model catalog, async LLM client, MCP server, local ONNX embeddings, a knowledge graph, and safety utilities — each module behind a feature flag.
- epic-harness — Claude Code plugin that consolidates 30+ commands into 3 commands plus 26 context-triggered skills, and evolves new skills from your own failure patterns.
- alcove — HTTP API server that gives coding agents on-demand access to private project docs, with BM25 + vector hybrid search, tree-sitter code indexing, and doc-consistency policy enforcement.
- Episteme — offline-first, single-binary knowledge graph that links design patterns, refactorings, and software laws through semantic relationships, exposed to agents via MCP.
- claudy — switch claude code between Anthropic, Z.AI, OpenRouter, Ollama, and custom endpoints with one command, with per-profile credential and config isolation.
- obsidian-forge — Rust CLI that scaffolds, automates, and maintains Obsidian vaults as a background daemon: inbox processing, graph strengthening, and git sync.
- llm-transpile — transpiles raw documents (Markdown, HTML, plain text) into a token-optimized bridge format with adaptive compression that stays under budget.