Overview
Project Anvil is a multi-agent harness for generating and remediating Ansible roles. It turns natural-language requirements into infrastructure code and repairs existing roles using policy-driven quality checks, deployment tests and compliance controls.
The project is an implementation artifact of PhD research at the University of Waikato on code quality and cyber security for infrastructure automation. It is a research proof of concept under active development.
Core Capabilities
- Generate Ansible roles from requirements, with iterative testing and repair.
- Remediate existing roles by fixing lint and syntax issues, applying best practices and adding compliance controls.
- Measure code quality across nine dimensions and 96 attributes, using configurable YAML policies and an Ansible Galaxy benchmark dataset.
- Assess deployed roles using CIS-CAT Pro and manual control checks.
How It Works
Generated and remediated roles move through lint, syntax, quality diagnostics, Docker deployment and live service checks. Failed checks trigger agent-driven repair loops informed by failure-specific skills and previously learned patterns.
A supervisor agent checks request feasibility, monitors the pipeline and produces a final run report. Quality weights, thresholds and compliance mappings are maintained as policies in the knowledge base.
Current Scope
The current implementation supports Ansible, MongoDB 7 Enterprise, HIPAA mappings and CIS Controls v8 across IG1, IG2 and IG3, with live CIS-CAT Pro assessment. Additional languages, technologies and regulatory frameworks are planned.