Agents

Overview

Sub-agents are markdown files in .claude/agents/ that define specialized AI personas with specific tools, knowledge, and behaviors. We run a 4-tier hierarchy: Tier 0 (Captain — CEO brain), Tier 1 (Business — revenue, growth, support), Tier 2 (Product — product, analyst), Tier 3 (Tech — 31 specialists across stack, design, UI, DevOps, VCS). Agents run in isolated contexts and can be launched in parallel via worktrees.

Current Progress

• 40 agents defined across 4 tiers with clear hierarchy
• 9 leadership agents (captain, revenue, growth, support, product, analyst, tech-lead, ops, guardian)
• 31 specialist agents covering stack, design, UI, DevOps, VCS, domain
• Worktree isolation tested for parallel execution

Areas of Improvement

• Agent quality varies — some need better tool restrictions
• No automated testing for agent behavior (prompt regression tests)
• Agent SDK integration not yet started for CI/CD pipelines
• Some agents have overlapping scope — need clearer boundaries

References

Setup

1. Create .claude/agents/ directory
2. Add markdown files with agent definitions (name, tools, instructions)
3. Specify subagent_type in agent definitions for the Agent tool
4. Set model override per agent (opus, sonnet, haiku)
5. Use isolation: worktree for parallel agent execution

Usage

• Agent tool with subagent_type selects the right specialist
• Orchestration agent delegates to multiple specialists in parallel
• Captain agent manages weekly allocation and strategic decisions
• Tech-lead agent handles architecture across all 14 repos
• Product agents (analyst, guardian) run quality and market checks

Config Files