Agent Teams
Spawns small parallel agent teams for review, debugging and feature work with strict file ownership.
Install
Claude Code: /plugin marketplace add wshobson/agents → /plugin install agent-teams@claude-code-workflows
Bir incelemeyi, hata avını ya da geliştirmeyi birden çok ajana paralel dağıtırken; rakip hipotezlerle kök neden aranırken
- Author
- Seth Hobson
- License
- MIT
Built on Claude Code's experimental Agent Teams feature, which has to be switched on with an environment variable and is most usable with teammates in separate tmux panes. Seven commands cover the lifecycle from spawning a preset team to shutting it down cleanly. The debugging preset is the interesting one: instead of one agent following a single theory, it generates competing hypotheses and gives each investigator the job of falsifying its own, then arbitrates on evidence. Parallel feature work is governed by file ownership — no two implementers may touch the same file, and boundaries are declared as interface contracts before anyone writes code. The guidance caps a useful team at two to four members, on the grounds that coordination overhead grows faster than throughput.
Similar skills
Dispatching Parallel Agents
SkillSplits genuinely independent problems across agents that run at the same time.
Claude-Mem
SkillCarries what an agent learned in one session into the next.
Claude Code Game Studios
SkillRuns a solo game project the way a studio runs one, with roles and gates.

Comments(0)
Sign in to comment