> Practical work on how AI is built, understood, and used. Explore tools developed at Aelion Path alongside open-source OWASP projects Helen Oakley co-leads.
A structured workflow for AI-assisted development, keeping task scope, review criteria, and evidence alongside the code.
OWASP AIBOM Generator and OWASP FinBot CTF make AI supply chain transparency and agentic security learning practical.
AI Software Development Life Cycle (SDLC) Harness gives developers a structured way to direct and review work performed with AI coding agents. Here, “harness” means the workflow structure around the agent’s work: defining the task, guiding its progress, and recording what was checked. It keeps scope, acceptance criteria, and evidence alongside the code in the project repository, with developers retaining authority over key decisions and final acceptance.

AI-assisted development needs a clear record of what was requested, what changed, and what was checked. Harness brings that record into the repository so it can be reviewed alongside the implementation.
Record the task's scope, acceptance criteria, and planned verification before implementation.
Use the current task context and workflow status to guide the developer or coding agent's next action.
Record actual checks and results, then review the change with the developer retaining authority over acceptance.
Harness is designed for teams bringing AI coding agents into their development workflow. It structures the work and its review record; it does not execute the coding agent or replace testing, security controls, or engineering judgment.

Helen Oakley created OWASP AIBOM Generator and OWASP FinBot CTF, turning practical challenges in AI transparency and agentic security into open-source tools. Both continue to evolve under her leadership with contributions from the OWASP community, under the OWASP GenAI Security Project. She also contributes to the OWASP Agent Control Standard (ACS), focusing on AgentBOM.
Make AI model information easier to inspect.
OWASP AIBOM Generator creates AI Bills of Materials from model information on Hugging Face. It organizes available metadata into a CycloneDX document and highlights gaps through completeness scoring.
For security, engineering, and AI teams, this provides a practical starting point for examining model provenance, documentation, and supply chain transparency.
Explore agentic AI security through hands-on learning.
OWASP FinBot CTF is an intentionally vulnerable application for learning how agentic AI systems can fail. Its vendor-management scenario gives participants a realistic setting for exploring interactions between AI agents, tools, business rules, and sensitive information.
Capture-the-flag challenges turn those interactions into practical learning for security practitioners, developers, and educators.
Making agent composition and dependencies visible.
The OWASP Agent Control Standard (ACS) is an evolving open standard for inspecting, tracing, and controlling AI agents. Helen leads its AgentBOM (AgBOM) work, focused on describing the components and dependencies that shape an agent’s operation.
An Agent Bill of Materials provides a structured view of the tools, models, and dependencies an agent uses—including changes as it operates.
Access the tools, documentation, and GitHub repositories.
Generate an AI Bill of Materials for a model hosted on Hugging Face.
Explore agentic AI vulnerabilities through hands-on security challenges.
Explore an open standard for inspecting, tracing, and controlling AI agents at runtime.
Contribute into OWASP AIBOM Generator
Contribute into OWASP FinBot CTF
Contribute into OWASP ACS