Eric (@tanker327)

Public skills

• tanker327/team — Decide whether a task warrants a multi-agent team and, if so, design it (roles, per-agent models, file ownership, task waves) and get user approval before launching. Use when the user wants work split across several parallel Claude agents/teammates — e.g. "agent team", "swarm", "throw a team of claudes at this", "split this across agents", "multi-agent workflow" — or describes work that benefits from parallel agents even without naming them: a multi-layer feature (backend/frontend/data/tests) built concurrently; a large migration, audit, or refactor spanning many independent files/modules/services; a multi-perspective code or architecture review (e.g. security + performance + tests at once); or competing-hypotheses debugging where several theories are investigated in parallel and challenged. Treat these as a cue to plan a team, not to work solo. Do NOT use for single-file, small, or tightly sequential work, debugging one known error, building an AI "agent" product (Agent SDK), Docker/infra "swarm" clusters, bulk email to "agents", or human/game "team composition".