// Coordinate Claude Code agent teams for td-watch development. Use when creating teams, assigning tasks, running team sessions, or when any agent needs to understand team coordination rules. Covers roles, file ownership, friction protocol, quality gates, and common team configurations.
| name | agent-teams |
| description | Coordinate Claude Code agent teams for td-watch development. Use when creating teams, assigning tasks, running team sessions, or when any agent needs to understand team coordination rules. Covers roles, file ownership, friction protocol, quality gates, and common team configurations. |
| Role | Agent file | Writes code | Key tools |
|---|---|---|---|
| Lead | (main session, delegate mode) | Never | Task management, SendMessage |
| Implementer | .claude/agents/implementer.md | Yes | Read, Write, Edit, Glob, Grep, Bash |
| Reviewer | .claude/agents/reviewer.md | No | Read, Glob, Grep, Bash |
| Researcher | .claude/agents/researcher.md | No | Read, Glob, Grep, Bash, WebFetch, WebSearch |
| Tester | .claude/agents/tester.md | Test files only | Read, Write, Edit, Glob, Grep, Bash |
The lead never writes code. Use delegate mode (Shift+Tab). Lead's job: break down work, assign tasks, triage friction, run quality gates, synthesize results.
Split by file ownership. Example: one owns server routes, one owns page components.
Researcher investigates first, implementer builds after findings arrive.
For building multiple independent components/pages in parallel.
For security-sensitive or contract-critical code. Review loop after implementation.
blockedBy when a task genuinely depends on another (page depends on server route). Don't use it for sequencing preference.Run after implementation completes, before marking work done:
npx svelte-check # type errors
npm run test:unit # component tests
npm run test:server # server route tests
npm run test:e2e # integration tests
Lead can run these directly or assign to a tester teammate.
Friction = any point where an agent can only proceed by guessing, bypassing, or working around.
~/code/td/ — td-sync server (Go)~/code/td-watch/ — td-watch (SvelteKit)~/code/sidecar/ — sidecar TUI (Go), reference onlyLabel which repo each task targets. Prefer separate tasks per repo. Researcher can explore ~/code/td/ while implementer works in ~/code/td-watch/.
When the lead senses context pressure (many exchanges, approaching limits), write a summary task containing:
For programmatic agent dispatch with full coding capability and streaming output, see:
references/agentic-coding-patterns.md — Detailed reference covering:
query() async generator, subagent orchestration with explicit role definitions, session persistence/resumption. TypeScript + Python.Both support full agentic coding with subagent spawning. The integration shape from Go is identical: launch subprocess, read structured events from stdout.
references/agentic-coding-patterns.md — Codex app-server + Claude Agent SDK patternsdocs/agent-teams.md — full team workflow documentationdocs/sync-admin-web-spec.md — system spec, API contractsdocs/design-system.md — design tokens, component specsdocs/agent-dev-guide.md — dev environment, testing, loggingTell Claude what to build and how to structure the team:
I'm working on [feature]. Create an agent team with delegate mode.
Spawn:
- [role] for [scope with file ownership]
- [role] for [scope with file ownership]
Break the work into tasks based on the spec. Assign clear file ownership.
Create and maintain DESIGN.md files — a design system specification optimized for AI consumption. Use when creating a project's design system file, extracting tokens from existing CSS/code, auditing design consistency, or onboarding an agent to a project's visual language.
Write naturally and avoid AI-detectable patterns. Use when (1) generating any written content, (2) reviewing/editing text for AI-like patterns, (3) user asks to make writing sound more human/natural, or (4) improving text that sounds robotic or generic. Covers vocabulary, structure, tone, and formatting tells that signal AI authorship.
Use the Perch plan CLI to create, read, update, and manage architectural plans. Covers project lifecycle, tree building, reviews, snapshots, facets, inputs, and export. Use when asked to retrieve a plan by name, build a plan tree, run a review, or interact with the planner in any way.
Set up and run autonomous overnight coding loops that process td epic tasks one at a time. Includes the loop runner, Rich TUI cockpit dashboard, and prompt template. Use when automating multi-day feature development across many tasks with td for state management.
Adapt AI agent runtimes (Claude Code SDK, Codex CLI app-server) into multi-turn conversational systems. Covers session lifecycle, resume/fork, event normalization, JSON-RPC 2.0 over stdio, tool policy enforcement, and rich component rendering (agent→UI generative components, UI→agent bidirectional state, component registries, surface targets). Use when building orchestrators, chat UIs, or sidecar processes that need persistent multi-turn agent sessions with rich interactive interfaces.
Orchestrate development work through sub-agents using td for state. Use when given a td task ID, text idea, markdown plan, or td epic to execute through plan-implement-review loops.