Agent: Claude

Purpose

• reasoning
• long-context analysis
• document writing
• architecture discussion

---

When to Use

• structure design
• long document review
• workflow organization
• change plan review before implementation

---

Model

Item	Value
Provider	Anthropic
Default Model	claude-sonnet-4-6
Higher Tier	claude-opus-4-6
API	Anthropic Messages API

---

Configuration

{
  "agent": "claude",
  "model": "claude-sonnet-4-6",
  "max_tokens": 4096,
  "system_prompt": "You are a helpful assistant with strong reasoning skills."
}

---

Setup

---

Rule Files

Path	Role
claude.md	project rule file
C:\\Users\\<user>\\.claude\\projects\\d--works-projects-local-ai-agent-mcp\\memory	project memory

---

Authentication

Method 1:

claude auth login

Method 2:

export ANTHROPIC_API_KEY="sk-ant-..."

Method 3:

$env:ANTHROPIC_API_KEY = "sk-ant-..."

---

Verify

claude auth status
claude -p "hello"

---

MCP Notes

• MCP tool usage available
• strong on long-context reasoning work
• better fit for analysis and design than local execution itself

---