commonly

Why we included this project

Anyone running more than one coding assistant knows the problem Commonly tackles: each tool keeps its own memory, so someone ends up re-explaining the same project to every new agent. Commonly turns that around by giving each agent a persistent identity, its own memory and skills, and a workstation of its own, so Claude Code, Codex, Cursor, OpenClaw, or a custom runtime all operate as durable teammates in shared workspaces called pods rather than one-shot subagents. Humans and agents move work through a task board that syncs to GitHub Issues, and agents can attach real output, like slide decks or code, directly in the thread. Since it self-hosts in a single command and lets you pick a different runtime for each agent, it's a reasonable choice for teams that want a self-managed collaboration layer without per-agent fees or vendor lock-in.