Run portable agent-team missions with Pragma
Use Pragma when an operator wants to define reusable agent teams, flows, memory, permissions, and human checkpoints once, then run the same mission through Desktop, CLI, Codex, Claude Code, or an embedded SDK host.
npx skills add agentskillexchange/skills --skill run-portable-agent-team-missions-with-pragma
Pragma is a preview-stage cross-harness runtime for building portable agent teams and AI working methods. Use it when an operator needs a repeatable mission that coordinates experts, expert teams, flows, tools, skills, memory, permissions, evaluations, and human review gates across multiple models or agent harnesses. Invoke this instead of using a single chat or coding-agent product normally when the durable asset is the team definition and execution trail: build or import a Pragma team, bind it to local model/runtime providers, run or resume the mission, and inspect results. The scope boundary is portable multi-agent mission execution, not a generic agent framework, desktop app, SDK, or CLI listing.
What this skill actually does
Inputs and prerequisites: Pragma Desktop or @pqpo/pragma CLI, configured model provider or local agent runtime, workspace permissions as needed.
Setup notes: For Desktop, download the macOS package from GitHub Releases. For CLI/runtime examples, install Node.js 22 and pnpm 10.12.1, clone https://github.com/pqpo/pragma, run pnpm install –frozen-lockfile, then use the documented @pragma examples or @pqpo/pragma CLI.
Source and verification boundary: use https://www.pqpo.me/2026/08/16/agent-team-builder-multi-harness-runtime/ 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.