nexus-agent¶

Create and manage custom AI agents.

Synopsis¶

nexus-agent COMMAND [OPTIONS]

Description¶

Create specialized AI personas that can be invoked via MCP tools. Agents are defined in YAML files in the agents/ directory.

Commands¶

Command	Description
`init`	Create a new agent
`list`	List configured agents
`templates`	Show available templates

nexus-agent init¶

Create a new custom agent.

From Template (Recommended)¶

nexus-agent init code_reviewer --from-template code_reviewer

Interactive¶

nexus-agent init my_agent
# Follow prompts for role, goal, backstory

With Custom Model¶

nexus-agent init security_expert \
  --from-template security_auditor \
  --model claude-opus-4.5

nexus-agent templates¶

List available agent templates.

nexus-agent templates

Output:

Available Agent Templates:

  code_reviewer     - Reviews code for bugs, security issues, and best practices
  doc_writer        - Creates and updates technical documentation
  debug_detective   - Analyzes errors and proposes fixes
  refactor_architect - Suggests code restructuring and design patterns
  test_engineer     - Generates test cases and improves coverage
  security_auditor  - Identifies vulnerabilities and recommends fixes
  api_designer      - Reviews and designs REST/GraphQL APIs
  performance_optimizer - Finds performance bottlenecks

nexus-agent list¶

Show configured agents.

nexus-agent list

Output:

Configured Agents:

  ✅ code_reviewer
     Role: Senior Code Reviewer
     Model: claude-sonnet-4.5

  ✅ my_custom_agent
     Role: Project Expert
     Model: auto

Agent Configuration¶

Agents are defined in agents/<name>.yaml:

name: "code_reviewer"
display_name: "Code Reviewer"
description: "Delegate code review tasks to the Code Reviewer agent."

profile:
  role: "Senior Code Reviewer"
  goal: "Identify bugs, security issues, and suggest improvements"
  backstory: "Expert developer with 10+ years of experience in code quality."
  tone: "Professional and constructive"

memory:
  enabled: true
  rag_limit: 5
  search_types: ["code", "documentation", "lesson"]

tools: []  # Empty = all tools available

llm_config:
  model_hint: "claude-sonnet-4.5"
  fallback_hints: ["auto"]
  temperature: 0.5
  max_tokens: 4000

Using Agents¶

After creating an agent, use it via MCP:

Use the ask_code_reviewer tool to review this function for security issues.

Or invoke directly:

# Via MCP tool
ask_code_reviewer("Review the authentication module for security issues")