Agentica Claude Proxy

This is integration glue for running Agentica agents through Claude Code's CLI as their tool provider. The main value is the documented gotchas: you need the allowedTools flag or file operations silently fail, agents will hallucinate success without the anti-hallucination prompt engineering, and the response format must be SSE streaming with Python code blocks containing return statements. It's essentially a reference implementation with a proxy server that bridges Agentica's expectations to Claude's REPL mode. Worth reading if you're chaining agents with Claude Code tools and hitting permission errors or phantom successes. The architecture diagram and error table are the most useful parts.

skills.sh installs: 485
GitHub stars: 3.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add parcadei/continuous-claude-v3 --skill agentica-claude-proxy --agent claude-code

Installs into .claude/skills of the current project.

Files

Agentica-Claude Code Proxy Integration

Use this skill when developing or debugging the Agentica-Claude proxy integration.

When to Use

Setting up Agentica agents to use Claude Code tools
Debugging agent hallucination issues
Fixing permission errors in file operations
Understanding the REPL response format

Architecture Overview

Agentica Agent → S_M_BASE_URL → Claude Proxy → claude -p → Claude CLI (with tools)
                 (localhost:2345)   (localhost:8080)

Critical Requirements

1. --allowedTools Flag (REQUIRED)

Claude CLI in -p mode restricts file operations. You MUST add:

subprocess.run([
    "claude", "-p", prompt,
    "--append-system-prompt", system_prompt,
    "--allowedTools", "Read", "Write", "Edit", "Bash",  # REQUIRED
])

Without this, agents will report "permission denied" for Write/Edit operations.

2. SSE Streaming Format (REQUIRED)

Agentica expects SSE streaming, not plain JSON:

# Response format
yield f"data: {json.dumps(chunk)}\n\n"
yield "data: [DONE]\n\n"

3. REPL Response Format (REQUIRED)

Agents MUST return results as Python code blocks with a return statement:

SKILL.md on GitHub

Details

Repository: github.com/parcadei/continuous-claude-v3
Categories: Backend & APIs, Documentation, CLI & Terminal
First seen: Jun 3, 2026