Best Claude Code agents and subagents

Counts main-thread work since the last subagent dispatch, nudges the model to delegate at every tenth call, and refuses foreground poll loops.

swarm：Mission control for subagents and agent teams: who is running, what each is doing, who spawned whom, who is ta

Asks a council of models about a hard problem, when the model calls its tool or you run /council <question>: Claude models, and Gemini models when gemini-core has a key, answer in parallel, and the model of the session writes one verdict from their answers.

The author integrates Xcode into Claude Code: builds, runs, tests, console, and SwiftUI previews are shown in a panel next to the chat.

Add readable status narration to agent activity so you can follow the work.

A Claude Code hook plugin that reports pane status, approval requests, tool activity and conversations to kasaterm, and delivers tell messages to the prompt when idle. It makes network requests (L3).

Put sessions on a shared racetrack to compare tool use, edits, tests and cost.

Add an inner-monologue-style status display to the session.

# DeutschLoop **A German tutor that remembers your mistakes.** It runs inside Claude Code and Codex. Every mistake is filed under its root cause, your **FehlerDNA**, and comes back in a new sentence, *nochmal*, until it stops. Your history stays on your own machine, and there is no extra API key.

Present current session activity in a sports-commentary style.

A dispatch board for dividing work across several Claude Code sessions. Write a mission file specifying nodes, responsible sessions, and completion conditions; the terminal board shows who is done, running, or waiting for you, with a toast on each state change.

A Claude Code plugin that rates each prompt's difficulty with a small model and runs that turn and its subagents at the matching effort level, within user-allowed limits, on models whose prompt cache survives effort changes.