The model-agnostic execution layer for AI agents: scoped routing and delegation, isolated worktrees, evidence, observability, and learning loops. The stack, not another wrapper. Seven executors — Claude Code, Codex, Cursor, opencode, DeepSeek Harness, Hermes, Pi — under one metered contract.
Multi-provider coding-agent orchestration in a verified loop: a critic on a different model signs off before work ships. Claude Code, Codex, Gemini, aider, Copilot, opencode, local.
A decentralized P2P infrastructure for any service or agent to discover, authenticate, and coordinate. Built in Go with zero-trust cryptography, mesh enables autonomous units to form secure, self-verifying networks without central oversight.
Run, watch, and meter Claude Code, Codex CLI, Gemini CLI, OpenCode, and Kiro from one CLI. Normalized events, cache-aware token usage, cost accounting verified against each agent's own records, budget caps, OpenTelemetry gen_ai spans, Langfuse traces, and a live web dashboard + MCP server.
Call dibs on files. Coordination for parallel coding agents — file claims with expiry, enforcement hooks, and git-native lessons. One binary, no server, no database.
Local-first control plane for AI-agent teams — orchestrates Claude Code, Codex, Grok, agy & Hermes into gated workflows that fix their own failures, with per-run metering, an append-only audit log, and team memory. Apache-2.0.