reldo 0.2.1

Claude-powered code review orchestrator that coordinates specialized review agents

Reldo

"The librarian has reviewed your code."

A Claude-powered code review orchestrator that coordinates specialized review agents.

Named after the Varrock Palace librarian in RuneScape who researches and checks things against ancient tomes.

Installation

# From PyPI
pip install reldo

# Or with uv
uv tool install reldo

Usage

As a Library

from reldo import Reldo, ReviewConfig
from pathlib import Path

# Load config from file
config = ReviewConfig.from_file(Path(".claude/reldo.json"))
reldo = Reldo(config=config)

# Run a review - single prompt argument
result = await reldo.review(
    prompt="Review app/Models/User.php for backend conventions. Context: Added user registration."
)

print(result.text)
print(f"Tokens used: {result.total_tokens}")

As a CLI

# Basic review
reldo review --prompt "Review app/Models/User.php"

# With JSON output (for CI)
reldo review --prompt "Review $(git diff --name-only HEAD)" --json --exit-code

# With custom config
reldo review --prompt "..." --config .claude/reldo.json

Configuration

Create .claude/reldo.json:

{
  "prompt": ".claude/reldo/orchestrator.md",
  "allowed_tools": ["Read", "Glob", "Grep", "Bash", "Task"],
  "mcp_servers": {
    "my-server": {
      "command": "node",
      "args": ["./mcp-server.js"]
    }
  },
  "agents": {
    "backend-reviewer": {
      "description": "Reviews PHP/Laravel code",
      "prompt": ".claude/reldo/agents/backend-reviewer.md",
      "tools": ["Read", "Glob", "Grep"]
    },
    "frontend-reviewer": {
      "description": "Reviews Vue/TypeScript code",
      "prompt": ".claude/reldo/agents/frontend-reviewer.md"
    }
  },
  "model": "claude-sonnet-4-20250514",
  "timeout_seconds": 300
}

Configuration Options

Option	Type	Default	Description
prompt	string	required	Path to orchestrator prompt file
allowed_tools	string[]	["Read", "Glob", "Grep", "Bash", "Task"]	Tools available to the orchestrator
mcp_servers	object	{}	MCP server configurations
agents	object	{}	Sub-agent definitions
model	string	"claude-sonnet-4-20250514"	Claude model to use
timeout_seconds	int	180	Maximum review duration
cwd	string	current directory	Working directory
logging	object	{"enabled": true, ...}	Logging configuration

Agent Definition

Each agent in the agents object has the following properties:

Property	Type	Required	Description
description	string	yes	Description of when to use this agent
prompt	string	yes	Path to agent prompt file
tools	string[]	no	Tools available to this agent. If omitted, inherits all tools from the orchestrator
model	string	no	Model override ("sonnet", "opus", "haiku", or "inherit")

Tool Inheritance

If you omit the tools property from an agent definition, it inherits all tools from the parent orchestrator, including any MCP server tools:

{
  "allowed_tools": ["Read", "Glob", "Grep", "Bash", "Task"],
  "mcp_servers": {
    "laravel-boost": {
      "command": "php",
      "args": ["artisan", "boost:mcp"]
    }
  },
  "agents": {
    "full-access-reviewer": {
      "description": "Has access to all orchestrator tools + MCP",
      "prompt": ".claude/agents/full-reviewer.md"
    },
    "limited-reviewer": {
      "description": "Only has read access",
      "prompt": ".claude/agents/limited-reviewer.md",
      "tools": ["Read", "Glob", "Grep"]
    }
  }
}

MCP Server Configuration

Reldo supports MCP (Model Context Protocol) servers for extended functionality:

{
  "mcp_servers": {
    "server-name": {
      "command": "executable",
      "args": ["arg1", "arg2"],
      "env": {
        "ENV_VAR": "value"
      }
    }
  }
}

Variable Substitution

MCP server configurations support variable substitution:

• ${cwd} - Replaced with the working directory

{
  "mcp_servers": {
    "serena": {
      "command": "uvx",
      "args": ["serena", "start-mcp-server", "--project", "${cwd}"]
    }
  }
}

CLI Reference

reldo review --prompt "..."     # Review prompt
             --config PATH      # Config file (default: .claude/reldo.json)
             --cwd PATH         # Working directory
             --json             # Output as JSON
             --verbose          # Verbose logging
             --no-log           # Disable session logging
             --exit-code        # Exit 1 if review fails (for CI)

Documentation

• Product Requirements Document

License

MIT