Add CodeAgentSwarm to Parallel Agent Runners

[arturogj92]

Adds CodeAgentSwarm to Parallel Agent Runners, in alphabetical order (between cmux and CodexMonitor).

CodeAgentSwarm is a desktop app (macOS, Windows) for running multiple Claude Code, Codex CLI, and Gemini CLI agents in parallel supervised terminals: desktop notifications when an agent finishes or needs input, searchable cross-session history, per-terminal file diffs, and a kanban task board the agents update over MCP.

[Keesan12]

This feels like a good fit for Parallel Agent Runners.

The distinction I keep finding useful is that supervised terminal grids solve a real visibility problem, but they do not automatically solve the trust problem.

A runner/dashboard like this helps answer:

• which agent is active
• which one needs input
• what changed in each terminal or worktree

But on longer runs, operators still end up needing a second layer of clarity:

• what meaningful delta actually happened
• whether the run is converging or just staying busy
• what should force the system to stop instead of asking for one more retry

That is less a criticism of the category and more a sign the taxonomy is getting sharper. Supervised parallel runners are becoming distinct from control/evidence layers, even though they sit right beside each other in real workflows.