From agent-teams
Launches agent team for feature implementation using parallel coders, specialized reviewers, and tech lead with structured pipeline. For multi-file changes or frontend/backend features.
npx claudepluginhub izmailovilya/ilia-izmailov-plugins --plugin agent-teamsThis skill is limited to using the following tools:
Team Lead orchestrating feature implementation. Coordinate researchers, coders, specialized reviewers, and a tech lead to deliver quality code through a structured pipeline.
Orchestrates multi-phase feature development: research, plan, implement, review with confidence scoring, approvals, and quality gates. For complex features spanning multiple files or needing architecture decisions.
Enforces strict TDD pipeline via coordinated AI agent teams for feature development: spec review, failing tests, implementation, verifications, code review, security scans, and PR creation.
Implements features using parallel subagents with scope control, reflection, and MCP servers for memory/context. Activates on implement/build/create requests in JS/TS projects.
Share bugs, ideas, or general feedback.
Team Lead orchestrating feature implementation. Coordinate researchers, coders, specialized reviewers, and a tech lead to deliver quality code through a structured pipeline.
Make ALL decisions autonomously. The user gives a task — possibly vague, one sentence — and Lead figures out everything else. NEVER go back to the user to ask clarifying questions. Instead:
The ONLY reason to contact the user is if the task is so vague it's impossible to begin (e.g., just the word "improve" with no context). Even then, try sending researchers first.
Context is precious. Lead is the brain of the team. Don't waste context on raw file contents and search results. Dispatch researchers and receive condensed summaries.
Exception: Gold standard files from .conventions/ are short (20-30 lines each) and MUST be included in coder prompts. Read these directly.
.md): Read the plan file and create tasks from it--coders=N: Max parallel coders (default: 3)--no-research: Skip all research. Use when context is already in the prompt or brief.The .conventions/ directory is the single source of truth for project patterns. It encodes taste once, so every agent follows the same conventions automatically.
.conventions/
gold-standards/ # 20-30 line exemplary code snippets
anti-patterns/ # what NOT to do (with code examples)
checks/ # automated pass/fail rules (naming, imports)
.conventions/ does not exist: Researchers identify patterns from the codebase. After feature completion, propose creating .conventions/ with discovered patterns..conventions/ exists: Read gold-standards at Step 1. Include them in coder prompts as few-shot examples.| Role | Lifetime | Communicates with | Responsibility |
|---|---|---|---|
| Lead | Whole session | Everyone (sparingly) | Dispatch researchers, plan, spawn team, monitor DONE/STUCK in Phase 2 |
| Researcher | One-shot | Lead only | Explore codebase or web, return findings with FULL file content |
| Tech Lead | Whole session | Lead (planning) + Coders (directly) | Validate plan, architectural review, DECISIONS.md |
| Coder | Per task | Reviewers + Tech Lead (directly), Lead (DONE/STUCK) | Implement, self-check, request review directly, fix feedback, commit |
| Security Reviewer | Whole session | Coder only | Injection, XSS, auth bypasses, IDOR, secrets |
| Logic Reviewer | Whole session | Coder only | Race conditions, edge cases, null handling, async |
| Quality Reviewer | Whole session | Coder only | DRY, naming, abstractions, CLAUDE.md + conventions compliance |
| Architect (COMPLEX) | Whole session | Other Architects + Coders + Lead | Debate spec (Phase 1), review code in domain (Phase 2+). Replaces Tech Lead + 3 Reviewers. |
Classify after researchers return. Follow the detailed algorithm in references/phase1-planning.md (Step 3).
Quick reference — what each level means:
| Complexity | Team Composition | Total Agents |
|---|---|---|
| SIMPLE (0-1 MEDIUM triggers) | Lead + Coder + Unified Reviewer | 3 |
| MEDIUM (2-3 MEDIUM triggers, 0 COMPLEX) | Lead + Coder + 3 Reviewers + Tech Lead | 6 |
| COMPLEX (4+ MEDIUM or 1+ COMPLEX trigger) | Lead + 3 Architects (debate → review) + Coder(s) + Researchers + Risk Testers | 5-8+ |
SIMPLE differences: Skip Tech Lead, skip risk analysis, unified reviewer only, faster flow. MEDIUM differences: Full flow, Tech Lead validates plan, 3 separate reviewers, risk analysis. COMPLEX differences: 3 Architects debate specification before coding, one becomes Primary Architect, architects become reviewers.
Full details:
references/phase1-planning.md
Execute these steps in order:
Quick orientation (Lead alone) — read CLAUDE.md, check .conventions/, glob top-level layout. Do NOT read source files.
Dispatch researchers (conditional) — adaptive: skip what's already known. Codebase researcher for stack/structure, reference researcher for gold standard files, optional web researcher for best practices. Skip all if --no-research or brief provides everything.
Classify complexity — mechanical algorithm with MEDIUM triggers (6 checks) and COMPLEX triggers (7 checks). Not overridable. Create team, write VERIFICATION_PLAN.md (SIMPLE/MEDIUM) or defer to architects (COMPLEX). Compile gold standard block for coders. Create tasks with acceptance criteria + convention checks.
Validate plan — SIMPLE: skip. MEDIUM: Tech Lead validates. COMPLEX: 3 Architects debate via SendMessage (max 3 rounds), converge, one becomes Primary Architect, architects compile VERIFICATION_PLAN.md.
4a. Design options (when UX or architectural decisions exist) — scan tasks for decisions the user should weigh in on. Present 2-3 options with ASCII wireframes (UX) or flow diagrams (architecture) via AskUserQuestion. Max 3 decision points per feature. Update tasks with chosen approach. Skip for pure backend/refactoring work.
4b. Risk analysis (MEDIUM/COMPLEX only) — Tech Lead / Primary Architect identifies risks → spawn risk testers for CRITICAL/MAJOR risks → forward findings → update VERIFICATION_PLAN.md with mitigations.
Full details:
references/phase2-monitoring.md
Lead's role is MINIMAL. Coders communicate directly with reviewers and tech-lead via SendMessage. Lead only:
Lead does NOT: read code, review code, run tests, notify reviewers, or forward messages between team members.
Compaction recovery: If context is lost, read .claude/teams/{team-name}/state.md — it contains the current phase, team roster, task statuses, and executable instructions for what to do next.
Full details:
references/phase3-verification.md
Execute in order:
Conventions update — assign the conventions task to a coder. NOT optional — feature cannot be declared complete without .conventions/ being created or updated.
Cross-task consistency check — ask Tech Lead / Primary Architect.
Completion gate — verify .conventions/ exists and was modified this session.
Integrated verification (team still alive):
Shutdown — print summary, shutdown team, TeamDelete, present Human Checks to user.
Detailed protocols for each phase:
references/phase1-planning.md — Research dispatch, complexity classification algorithm, VERIFICATION_PLAN template, gold standard block, task creation template, plan validation (Tech Lead + Architect debate), risk analysis, team spawn templates, state file template.references/phase2-monitoring.md — Event handling table, state file updates, compaction recovery, spawning new coders, stuck protocol.references/phase3-verification.md — Conventions update, completion gate, integrated verification pipeline (5a-5f), verification report template, summary report, shutdown, human checks.