Agency-Agents¶

What it is¶

Agency-Agents is a comprehensive suite of 110+ specialized AI agent personas designed to transform generic coding assistants into a "complete AI agency." In June 2026, it serves as a critical configuration layer for IDE-based agents like Claude 4.8 Opus and GPT-5.5, providing them with domain-specific identities, missions, and success metrics through Native MCP 3.0 support.

What problem it solves¶

It reduces AI hallucinations and improves technical output by providing "off-the-shelf" expert personas. It moves beyond generic "write code" prompts to specialized, opinionated domain expertise—from Frontend Architects and Security Engineers to Reality Checkers and specialized Business Analysts.

Where it fits in the stack¶

Agents / Personas / Framework. It acts as a system prompt and configuration layer for terminal and IDE-based agents. It sits between the raw LLM and the application-specific workflow.

Typical use cases¶

Multi-agent IDE Workflows: Invoking a "Backend Architect" for initial design and a "Security Engineer" for a final PR review.
Reality Checking: Using the "Reality Checker" persona to find logical flaws in proposed solutions before implementation.
Specialized Engineering: Deploying "Performance Tuning" or "Documentation Specialist" personas for specific project phases.
Business Logic Review: Using "Financial Risk Analyst" or "Product Manager" personas to evaluate feature impact.

Strengths¶

High Specialization: 110+ personas covering development, security, business, and creative roles.
Native MCP 3.0: Personas are now exposed as Model Context Protocol resources, allowing seamless discovery by compliant agents.
Claude 4.8 Optimized: Personas updated with specialized PreToolUse and PostToolUse logic for high-precision tool calling.
Model Agnostic: Works with any frontier model through any interface.
Improved Grounding: Drastically reduces hallucinations by narrowing the agent's focus and providing specific constraints.

Limitations¶

Manual Integration: Requires cloning the repo and manually referencing files in most tools.
Context Overhead: Long system prompts from complex personas can consume a significant portion of the context window in older models.
Maintenance: Personas may need periodic updates to align with the capabilities of new models (e.g., Claude 4.8).

When to use it¶

When you need more than just a general-purpose assistant and want a virtual "team" of experts.
For complex engineering tasks that require multiple perspectives (architecture, security, testing).
When using tools like Claude Code, Cursor, or Aider that allow custom system instructions.

When not to use it¶

For simple, one-off tasks where a general-purpose assistant is sufficient.
If you have already developed highly customized, proprietary system prompts for your specific domain.

Getting started¶

Installation¶

# Clone the repository to your local machine
git clone https://github.com/msitarzewski/agency-agents.git ~/.agency-agents

MCP 3.0 Setup¶

Expose the persona library to your agents by adding the Agency MCP server to your configuration:

mcp install agency-agents --path ~/.agency-agents

Integration with Claude Code¶

To use a persona with Claude Code, you can reference the markdown file or use the MCP resource:

claude "Use the @agency/backend-architect persona to design a FastAPI service."

CLI examples¶

# List all available agent personas
ls ~/.agency-agents/agents/

# Use the 'Security Engineer' persona with Aider
aider --model claude-4-8-opus-20260528 --message-file ~/.agency-agents/agents/security-engineer.md

# Search for a specific specialist (e.g., Frontend)
ls ~/.agency-agents/agents/ | grep "frontend"

API examples¶

Python Integration¶

You can programmatically load these personas into your own agentic frameworks.

import os

def load_persona(agent_name):
    # Ensure name ends with .md
    filename = agent_name if agent_name.endswith(".md") else f"{agent_name}.md"
    path = os.path.expanduser(f"~/.agency-agents/agents/{filename}")

    if not os.path.exists(path):
        raise FileNotFoundError(f"Persona {agent_name} not found at {path}")

    with open(path, "r") as f:
        return f.read()

# Load the 'Reality Checker' persona for a critique loop
reality_checker_prompt = load_persona("reality-checker")

# Example: Sending to an LLM via a standard client
# response = client.chat.completions.create(
#     model="claude-4-8-opus-20260528",
#     messages=[
#         {"role": "system", "content": reality_checker_prompt},
#         {"role": "user", "content": "Critique my proposed architecture for a distributed cache."}
#     ]
# )

Sources / References¶

Contribution Metadata¶

Last reviewed: 2026-06-28
Confidence: high