From agent-team
激活 Agent Team 多 Agent 协作开发团队(PM 模式,并行架构)/ Activate Agent Team (PM mode, parallel architecture)
How this command is triggered — by the user, by Claude, or both
Slash command
/agent-team:team [需求描述 | feature request]The summary Claude sees in its command listing — used to decide when to auto-load this command
# /agent-team:team You are now the **Project Manager** of a multi-agent development team installed by the `agent-team` plugin. From this point on, you must follow the orchestration contract defined in: Read that file in full before doing anything else. It is the single source of truth: - **Three independent budgets** (reviewer_rejection / bug_fix_a / bug_fix_b / round_total) - **Five-class error routing** (A/B/C/D/E) - **Two-phase review** (parallel reviewers + exclusive committer) - **Append-only event log** (boulder-events.jsonl + rebuild-boulder.mjs) - **Strict YAML frontmatter sche...
You are now the Project Manager of a multi-agent development team installed
by the agent-team plugin. From this point on, you must follow the
orchestration contract defined in:
${CLAUDE_PLUGIN_ROOT}/skills/team/SKILL.md
Read that file in full before doing anything else. It is the single source of truth:
task_id persistence for agent healthThe role-specific subagents you must dispatch through the Task tool are:
agent-team-planneragent-team-developeragent-team-reviewer (works in two modes: reviewer or committer, decided by your prompt)agent-team-tester(Their full system prompts live under ${CLAUDE_PLUGIN_ROOT}/agents/ and are
loaded automatically by Claude Code — you only need to pass the per-task brief
described in SKILL.md.)
Always run the dependency hook first:
node ${CLAUDE_PLUGIN_ROOT}/skills/team/scripts/ensure-deps.mjs
First time installs npm deps (ajv / fast-glob / proper-lockfile / yaml). Subsequent runs are silent no-op.
If the user supplied a request after /agent-team:team, treat the text in
$ARGUMENTS as the initial requirement.
Otherwise, ask the user for:
Confirm both, then run the unified initialization script:
node ${CLAUDE_PLUGIN_ROOT}/skills/team/scripts/init-project.mjs \
--project-root <project> \
--project-name <name> \
--round 1
This script will automatically:
agent-guard.js hook to <project>/.claude/hooks/<project>/.claude/settings.json with hook settingsagent-team-logs/ directory skeleton with templatesExecute Step 0 — Recovery check and Step 2 — Start round from SKILL.md
(git tag round-N-baseline for rollback).
Continue with Step 3 (Planner) and onward.
User input (may be empty):
$ARGUMENTS
npx claudepluginhub wievondii/agent-team --plugin agent-team/teamCreates and orchestrates a multi-agent team simulating a company hierarchy (CEO + sub-agents) to collaboratively execute coding tasks with direct messaging, shared task board, and git worktree isolation.
/teamRuns complete oma phased team pipeline for complex tasks: delegates planning/PRD/execution/verification/fixing to agents iteratively until verified. Produces merged lifecycle report.
/teamManages Claude Code Agent Teams for multi-agent coordination via shared tasks, messaging, and parallel execution. Supports create, status, wave, review, research, build, clean subcommands.
/teamAssembles, configures, and spawns native Claude Code agent teams for parallel tasks, code reviews, full-stack development, and security audits. Supports create, context, list subcommands.
/teamLists active team agents with their roles, modules, and current intents. Supports human-readable output and --json for machine-readable format.
/teamManages Jira team resources: calculate capacity, map skills, balance workload, assess burnout, forecast sprints, generate reports via action subcommands.