AI Agent Team Guide

The @saas template ships with a 12-agent Claude Code team for automated development.

Agent Roster

Agent	Command	Role
@spec-writer	/athena:spec	Feature spec, OpenAPI-first design
@reviewer	/athena:qa --review-only	Code review + security audit
@qa	/athena:qa --test-only	Test execution + 80% coverage gate
@evaluator	/athena:qa --eval-only	Independent acceptance evaluation
@best-practice	—	Architecture + trade-off decisions
@debugger	—	Errors, failing tests (auto-delegated)
@deployer	/athena:deploy	7-gate deploy protocol
@memory-curator	/athena:promote	Extract wisdom → Tier 0
@strategist	/athena:plan	Audit, research, propose epics
@orchestrator	/athena:batch	Parallel epic coordination
@designer	/athena:design	Design tokens → React page
@dba	/athena:dba	Migration review, schema design

Invocation

# Full pipeline (one step at a time)
/athena:loop

# Full pipeline (auto-advance)
/athena:loop auto

# Parallel epics (max 4 concurrent)
/athena:batch auto

# Confidence-gated autopilot
/athena:autopilot

When to Use Each Agent

Building a new feature:

/athena:spec "feature description"
/athena:implement
/athena:qa
/athena:pr

Diagnosing a bug:

@debugger — describe the error, share logs

Reviewing for production:

/athena:qa --review-only   # security + code quality
/athena:deploy staging     # 7-gate pre-deploy check

Optimizing DB:

/athena:dba inspect-migrations
/athena:dba lint

Effort Tiers

Pass --effort <tier> to scale depth vs. cost:

Tier	Use Case
quick	Rapid iteration, draft specs
standard	Normal development (default)
thorough	Pre-merge quality gate
ultra	Security audit, critical-path review

/athena:qa --effort ultra   # maximum review depth
/athena:spec --effort quick "small tweak"