agenticflow-agent

Create, run, and update single AgenticFlow agents via the af CLI.

12|8|Updated Oct 19, 2025

Search Tags:#agent #mcp #iteration #ai-agent #system-prompt #af-cli

Ready to run

npx skills add https://github.com/PixelML/agenticflow-skill --skill agenticflow-agent

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agenticflow-agent
Source: https://github.com/PixelML/agenticflow-skill/tree/main/skills/agenticflow-agent
Command: npx skills add https://github.com/PixelML/agenticflow-skill --skill agenticflow-agent

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill enables you to create, run, and iterate on a single AgenticFlow AI agent — one chat endpoint, one persona, and one set of rules — for customer-facing bots or internal assistants.

Core Features & Use Cases

Single-agent lifecycle: create, update (including --patch partial updates), run, and delete.
Supports model configuration, system prompts, optional MCP tool attachments, and a code execution sandbox.
Iterative development with schema guidance, robust commands, and safe patching to preserve attachments and tools.

Quick Start

Create your first AgenticFlow agent with a minimal system prompt and run a test conversation to verify behavior.

Frequently Asked Questions about agenticflow-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I build and iterate on a single AI agent with a custom system prompt?▼

To build and iterate on a single AI agent, you define a custom system prompt and persona, then use patch-based updates to safely refine its behavior without losing attachments. This creates one chat surface without orchestrating multiple agents.

What is the best way to attach MCP tools to an AI agent?▼

Attaching MCP tools involves configuring them during the agent creation or update lifecycle, ensuring they remain preserved across patch-based updates. This enables the AI agent to leverage external tool capabilities within its single chat surface.

Can I update an AI agent's configuration without losing its existing MCP tool attachments?▼

Yes, you can update an AI agent without losing MCP tool attachments by applying safe patch-based partial updates. This method modifies the configuration while deliberately preserving existing tools and model settings.

Does AgenticFlow support a code execution sandbox for AI agents?▼

Yes, AgenticFlow supports an optional code execution sandbox for AI agents. You can configure this setting alongside model configuration and system prompts to enable dynamic code execution within the agent's chat surface.

How do I inspect the schema of an AI agent before running a test conversation?▼

You inspect an AI agent schema by using schema guidance commands via the af CLI. This allows you to verify the model configuration, system prompt, and attached tools before running a test conversation.