content-builder-agent

Content Builder Agent

A content writing agent for writing blog posts, LinkedIn posts, and tweets with cover images included.

This example demonstrates how to use Agent-RT with OpenBox governance through three filesystem primitives:

• Memory (AGENTS.md) – persistent context like brand voice and style guidelines
• Skills (skills/*/SKILL.md) – workflows for specific tasks, loaded on demand
• Subagents (subagents.yaml) – specialized agents for delegated tasks like research

The content_writer.py script shows how to combine these into a working agent using Agent-RT with OpenBox governance.

Quick Start

# Set API keys
export OPENAI_API_KEY="..."
export OPENAI_MODEL="gpt-4o-mini"
export OPENBOX_API_URL="https://core.openbox.ai"
export OPENBOX_API_KEY="obx_live_..."
export OPENBOX_AGENT_DID="did:aip:..."          # Required by default for newly registered agents
export OPENBOX_AGENT_PRIVATE_KEY="..."          # Required by default for newly registered agents
export GOOGLE_API_KEY="..."      # For image generation
export TAVILY_API_KEY="..."      # For web search (optional)

# Run (uv automatically installs dependencies on first run)
cd examples/content-builder-agent
uv run python content_writer.py "Write a blog post about prompt engineering"

More examples:

uv run python content_writer.py "Create a LinkedIn post about AI agents"
uv run python content_writer.py "Write a Twitter thread about the future of coding"

OpenBox enables DID signing by default for newly registered agents. If signing has been explicitly disabled for this agent in OpenBox, you can omit OPENBOX_AGENT_DID and OPENBOX_AGENT_PRIVATE_KEY.

To use Anthropic instead of OpenAI, set ANTHROPIC_API_KEY and ANTHROPIC_MODEL and install the matching Agent-RT provider extra. The example installs the local SDK in editable mode through [tool.uv.sources].

How It Works

The agent is configured by files on disk, not code:

content-builder-agent/
├── AGENTS.md                    # Brand voice & style guide
├── subagents.yaml               # Subagent definitions
├── skills/
│   ├── blog-post/
│   │   └── SKILL.md             # Blog writing workflow
│   └── social-media/
│       └── SKILL.md             # Social media workflow
└── content_writer.py            # Wires it together (includes tools)

File	Purpose	When Loaded
AGENTS.md	Brand voice, tone, writing standards	Always (system prompt)
subagents.yaml	Research subagent config (prompt, model, tools)	When research tool runs
skills/*/SKILL.md	Content-specific workflows	Always (appended to system prompt)

Architecture

# Load memory + skills into system prompt
agents_md = (EXAMPLE_DIR / "AGENTS.md").read_text()
skills_text = load_skills(EXAMPLE_DIR / "skills")

# Create OpenBox governance
governance = await create_openbox_agentrt(
    api_url=os.environ["OPENBOX_API_URL"],
    api_key=os.environ["OPENBOX_API_KEY"],
    agent_did=os.environ.get("OPENBOX_AGENT_DID"),
    agent_private_key=os.environ.get("OPENBOX_AGENT_PRIVATE_KEY"),
    agent_name="ContentWriter",
)

# Create a governed agent loop
loop = governance.govern_loop(AgentLoop(load_model(), tool_registry=registry))

# Run with governance applied automatically
result = await loop.run_streaming(agent, [user_message(task)], print_progress)

Flow:

1. Agent receives task → loads relevant skill (blog-post or social-media)
2. Calls research tool → runs the researcher subagent as its own governed run → saves to research/
3. Reads research findings → writes content → saves to blogs/, linkedin/, or tweets/
4. Generates cover image with Gemini → saves alongside content

File tools refuse paths that would escape the example directory.

Output

blogs/
└── prompt-engineering/
    ├── post.md       # Blog content
    └── hero.png      # Generated cover image

linkedin/
└── ai-agents/
    ├── post.md       # Post content
    └── image.png     # Generated image

research/
└── prompt-engineering.md   # Research notes

Customizing

Change the voice: Edit AGENTS.md to modify brand tone and style.

Add a content type: Create skills/<name>/SKILL.md with YAML frontmatter:

---
name: newsletter
description: Use this skill when writing email newsletters
---
# Newsletter Skill
...

Add a subagent: Add to subagents.yaml:

editor:
  description: Review and improve drafted content
  model: inherit
  system_prompt: |
    You are an editor. Review the content and suggest improvements...
  tools: []

Add a tool: Define an async handler and ToolDefinition in content_writer.py, add it to TOOLS, and include its name in the build_registry([...]) list. Classify it for policies through tool_type_map.

Key Differences from LangChain Version

Aspect	LangChain	Agent-RT
Agent creation	create_agent() from langchain	AgentConfig + AgentLoop
Memory/Skills	Loaded manually into system prompt	Loaded manually into system prompt
Subagents	Custom research tool spawning sub-agent	Custom research tool running a nested governed AgentLoop
Governance	create_openbox_langchain_middleware() via AgentMiddleware	create_openbox_agentrt() + governance.govern_loop(...)
Execution	Sync (stream)	Async (run_streaming)
Tools	@tool decorator	ToolDefinition + async ToolRegistry handler

Requirements

• Python 3.11–3.14
• OPENAI_API_KEY + OPENAI_MODEL - For the main agent and researcher
• OPENBOX_API_URL + OPENBOX_API_KEY - For OpenBox governance
• GOOGLE_API_KEY - For image generation (Gemini)
• TAVILY_API_KEY - For web search (optional, research still works without it)