How does this skill prevent agent hallucinations?

It includes specialized 'Anti-Hallucination Warning' prompt templates that instruct agents to verify tool results before claiming a task is successfully completed.

Why am I getting sandbox restriction errors for /tmp paths?

Both Maestro and Agentica restrict access to certain directories. This skill recommends using project-relative paths like 'workspace/' to ensure consistent file access.

What is the required response format for Agentica agents?

Agents must return results as Python code blocks using a return statement, which Agentica's REPL parser then extracts and processes.

Why do my agents report 'permission denied' for file operations?

This is typically caused by missing the --allowedTools flag in the Claude CLI -p mode. This skill provides the specific subprocess configuration to enable Read, Write, Edit, and Bash tools.

Agentica-Claude Proxy Integration

Name: Agentica-Claude Proxy Integration
Author: scooter-lacroix

byscooter-lacroix

0•

API Development

Bridges Agentica SDK agents with Maestro CLI tools to enable robust tool execution and anti-hallucination guardrails.

This skill provides a standardized architectural framework for connecting the Agentica SDK to the Maestro CLI proxy. It addresses critical integration requirements such as SSE streaming compliance, REPL response formatting, and permission management via the --allowedTools flag. By implementing the provided anti-hallucination prompt engineering and path sandboxing strategies, developers can build reliable multi-agent swarms that interact safely with local file systems and bash environments while maintaining strict operational consistency.

Key Features

01Standardizes REPL response parsing for Python code execution

020 GitHub stars

03Optimizes path sandboxing to prevent file access errors in local environments

04Configures SSE streaming formats for Agentica-to-Maestro communication

05Manages CLI permissions using the --allowedTools flag for Write/Edit operations

06Implements anti-hallucination prompt engineering to ensure tool invocation

Use Cases

01Standardizing response formats for custom LLM proxies using the Agentica SDK

02Setting up multi-agent swarms that require local file system access

03Debugging permission errors when Claude CLI tools fail to write or edit files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add scooter-lacroix/maestro agentica-claude-proxy

For use in Claude.ai and ChatGPT

Download Skill