Skip to content
luiseiman edited this page Apr 5, 2026 · 3 revisions

Agents

7 specialized subagents with orchestration chains.

Agent Roster

Agent Role Memory Model
researcher Read-only codebase exploration Transactional haiku
architect Design decisions, tradeoff analysis Persistent opus
implementer Code + tests Persistent sonnet
code-reviewer Review by severity (critical/warning/suggestion) Persistent sonnet
security-auditor Vulnerability scanning (OWASP top 10) Persistent opus
test-runner Run tests + report coverage Transactional sonnet
session-reviewer Post-session analysis Transactional sonnet

Orchestration Chain

researcher → architect → implementer → test-runner → code-reviewer

Defined in .claude/rules/agents.md. Claude follows the delegation decision tree:

  1. Single-file fix / quick question → handle directly
  2. Research-heavy → delegate to researcher
  3. Code changes + tests → delegate to implementer
  4. Security concern → delegate to security-auditor
  5. Multi-component refactor (>3 files) → evaluate Agent Teams
  6. Code review → delegate to code-reviewer
  7. Architecture decision → delegate to architect

Memory

Agents with persistent memory write learnings to .claude/agent-memory/<agent-name>/. These survive across sessions and prevent repeated mistakes.

Transactional agents (researcher, test-runner, session-reviewer) execute and report without accumulating knowledge.

Output Budget

All agents include a 5K token output budget constraint. Summaries over raw output — agents don't echo full files or dump raw grep results.

Frontmatter Fields

Agent definitions support these frontmatter fields:

Field Description
allowed-tools Tool whitelist (use allowed-tools:, not tools:)
model Model override: haiku, sonnet, opus
effort Reasoning budget: normal, high, max (architect=high, security-auditor=max)
context fork — skill executes in isolated subagent (used by 5 heavy skills)
color UI color hint

Model Selection

Model assignments follow the criteria in template/rules/model-routing.md (installed to every bootstrapped project). See Model Routing for the full selection criteria and escalation rules.

Agent Teams

For large tasks touching 3+ independent components, multiple agents can work in parallel using isolation: "worktree". A lead agent coordinates; teammates each own a distinct file set.

To continue a subagent's work, use SendMessage({to: agentId}) — never spawn a new agent for follow-ups.

Clone this wiki locally