Qualifire AI Releases Rogue: An Finish-to-Finish Agentic AI Testing Framework, Evaluating the Efficiency of AI Brokers

Agentic methods are stochastic, context-dependent, and policy-bounded. Typical QA—unit checks, static prompts, or scalar “LLM-as-a-judge” scores—fails to show multi-turn vulnerabilities and offers weak audit trails. Developer groups want protocol-accurate conversations, specific coverage checks, and machine-readable proof that may gate releases with confidence.

Qualifire AI has open-sourced Rogue, a Python framework that evaluates AI brokers over the Agent-to-Agent (A2A) protocol. Rogue converts enterprise insurance policies into executable eventualities, drives multi-turn interactions towards a goal agent, and outputs deterministic reviews appropriate for CI/CD and compliance critiques.

Fast Begin

Stipulations

• uvx – If not put in, observe uv set up information
• Python 3.10+
• An API key for an LLM supplier (e.g., OpenAI, Google, Anthropic).

Set up

Choice 1: Fast Set up (Advisable)

Use our automated set up script to stand up and operating shortly:

# TUI
uvx rogue-ai
# Internet UI
uvx rogue-ai ui
# CLI / CI/CD
uvx rogue-ai cli

Choice 2: Handbook Set up

(a) Clone the repository:

git clone https://github.com/qualifire-dev/rogue.git
cd rogue

(b) Set up dependencies:

If you’re utilizing uv:

Or, if you’re utilizing pip:

(c) OPTIONALLY: Arrange your atmosphere variables: Create a .env file within the root listing and add your API keys. Rogue makes use of LiteLLM, so you’ll be able to set keys for varied suppliers.

OPENAI_API_KEY="sk-..."
ANTHROPIC_API_KEY="sk-..."
GOOGLE_API_KEY="..."

Working Rogue

Rogue operates on a client-server structure the place the core analysis logic runs in a backend server, and varied shoppers hook up with it for various interfaces.

Default Habits

If you run uvx rogue-ai with none mode specified, it:

1. Begins the Rogue server within the background
2. Launches the TUI (Terminal Consumer Interface) shopper

Out there Modes

• Default (Server + TUI): uvx rogue-ai – Begins server in background + TUI shopper
• Server: uvx rogue-ai server – Runs solely the backend server
• TUI: uvx rogue-ai tui – Runs solely the TUI shopper (requires server operating)
• Internet UI: uvx rogue-ai ui – Runs solely the Gradio net interface shopper (requires server operating)
• CLI: uvx rogue-ai cli – Runs non-interactive command-line analysis (requires server operating, excellent for CI/CD)

Mode Arguments

Server Mode

uvx rogue-ai server [OPTIONS]

Choices:

• –host HOST – Host to run the server on (default: 127.0.0.1 or HOST env var)
• –port PORT – Port to run the server on (default: 8000 or PORT env var)
• –debug – Allow debug logging

TUI Mode

uvx rogue-ai tui [OPTIONS]
Internet UI Mode
uvx rogue-ai ui [OPTIONS]

Choices:

• –rogue-server-url URL – Rogue server URL (default: http://localhost:8000)
• –port PORT – Port to run the UI on
• –workdir WORKDIR – Working listing (default: ./.rogue)
• –debug – Allow debug logging

Instance: Testing the T-Shirt Retailer Agent

This repository features a easy instance agent that sells T-shirts. You need to use it to see Rogue in motion.

Set up instance dependencies:

If you’re utilizing uv:

or, if you’re utilizing pip:

pip set up -e .[examples]

(a) Begin the instance agent server in a separate terminal:

If you’re utilizing uv:

uv run examples/tshirt_store_agent

If not:

python examples/tshirt_store_agent

This may begin the agent on http://localhost:10001.

(b) Configure Rogue within the UI to level to the instance agent:

• Agent URL: http://localhost:10001
• Authentication: no-auth

(c) Run the analysis and watch Rogue take a look at the T-Shirt agent’s insurance policies!

You need to use both the TUI (uvx rogue-ai) or Internet UI (uvx rogue-ai ui) mode.

The place Rogue Suits: Sensible Use Circumstances

• Security & Compliance Hardening: Validate PII/PHI dealing with, refusal habits, secret-leak prevention, and regulated-domain insurance policies with transcript-anchored proof.
• E-Commerce & Assist Brokers: Implement OTP-gated reductions, refund guidelines, SLA-aware escalation, and tool-use correctness (order lookup, ticketing) underneath adversarial and failure situations.
• Developer/DevOps Brokers: Assess code-mod and CLI copilots for workspace confinement, rollback semantics, rate-limit/backoff habits, and unsafe command prevention.
• Multi-Agent Programs: Confirm planner↔executor contracts, functionality negotiation, and schema conformance over A2A; consider interoperability throughout heterogeneous frameworks.
• Regression & Drift Monitoring: Nightly suites towards new mannequin variations or immediate modifications; detect behavioral drift and implement policy-critical move standards earlier than launch.

What Precisely Is Rogue—and Why Ought to Agent Dev Groups Care?

Rogue is an end-to-end testing framework designed to judge the efficiency, compliance, and reliability of AI brokers. Rogue synthesizes enterprise context and threat into structured checks with clear goals, ways and success standards. The EvaluatorAgent runs protocol right conversations in quick single flip or deep multi flip adversarial modes. Carry your individual mannequin, or let Rogue use Qualifire’s bespoke SLM judges to drive the checks. Streaming observability and deterministic artifacts: stay transcripts,move/fail verdicts, rationales tied to transcript spans, timing and mannequin/model lineage.

Beneath the Hood: How Rogue Is Constructed

Rogue operates on a client-server structure:

• Rogue Server: Incorporates the core analysis logic
• Consumer Interfaces: A number of interfaces that hook up with the server:
  • TUI (Terminal UI): Trendy terminal interface constructed with Go and Bubble Tea
  • Internet UI: Gradio-based net interface
  • CLI: Command-line interface for automated analysis and CI/CD

This structure permits for versatile deployment and utilization patterns, the place the server can run independently and a number of shoppers can hook up with it concurrently.

Abstract

Rogue helps developer groups take a look at agent habits the best way it truly runs in manufacturing. It turns written insurance policies into concrete eventualities, workouts these eventualities over A2A, and information what occurred with transcripts you’ll be able to audit. The result’s a transparent, repeatable sign you should use in CI/CD to catch coverage breaks and regressions earlier than they ship.

Due to the Qualifire crew for the thought management/ Assets for this text. Qualifire crew has supported this content material/article.