Orchestrate local-first agent memory and workflows with OpenHuman
Set up OpenHuman as a local-first personal agent harness that syncs context, builds durable memory, and coordinates multi-agent workflows behind approval gates.
npx skills add agentskillexchange/skills --skill orchestrate-local-first-agent-memory-and-workflows-with-openhuman
Use OpenHuman when the agent/operator needs to turn a personal or team workstation into a local-first agent harness: connect accounts, sync context into memory, route work across agent runtimes, and save approval-gated workflows that can run on schedules or channel events. Upstream evidence describes local memory trees, Obsidian-style context, 100+ OAuth integrations, 5k+ MCP connections, visual workflow graphs, and encrypted agent-to-agent orchestration across tools such as Claude Code, Codex, OpenClaw, and Hermes.
What this skill actually does
Invoke this instead of simply chatting with a single assistant when the job is to prepare a durable operating environment for future agent work, not to answer one prompt. The scope boundary is the operator workflow: install OpenHuman, connect the relevant local data/tools, configure memory and workflow automation, and keep side effects approval-gated. It is not a generic product card for every OpenHuman feature, a model SDK, or a replacement for the underlying coding agents.
Inputs and prerequisites: OpenHuman desktop or core release, local workstation, selected account connectors or MCP tools, target agent runtimes such as Claude Code, Codex, OpenClaw, or Hermes, and configured model/provider access.
Setup notes: Download installers from https://tinyhumans.ai/openhuman or https://github.com/tinyhumansai/openhuman/releases/latest. For source builds, install Git, Node.js 24+, pnpm 10.10.0, Rust 1.93.0 with rustfmt and clippy, CMake, Ninja, ripgrep, and platform desktop prerequisites; then clone the repo, run git submodule update –init –recursive, pnpm install, and pnpm –filter openhuman-app dev:app for the desktop shell.
Source and verification boundary: use https://tinyhumans.gitbook.io/openhuman/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.