Key Points

Introduction to Advanced Research Software AI Coding

Treat and review AI-generated code before use with at least the same level of scrutiny as human-written code.
Intentional software engineering discipline and practices becomes more important, not less, when using AI-assisted development.
Following established software engineering stage-gated processes for requirements, design, implementation, and testing can help mitigate the risks.
Apply automated quality checks such as linting, testing, static analysis, and security scanning.
Restrict AI agents using least-privilege access and avoid using them on production systems.
Ensure human approval is required for high-risk actions such as deployments, infrastructure changes, and data access.

Ensure GitHub Copilot extension is installed and you’re logged into your GitHub account.
Create a Python virtual environment and install required packages (NumPy and Matplotlib).
Instructions files help standardize code style, enforce data integrity, and improve Copilot’s suggestions across a whole project.

Use the planning agent to create structured plans before implementation.
Create custom skills to encapsulate common development tasks.
Always review AI-generated code, tests, and documentation carefully.
Consider whether dedicated tools might be more appropriate than AI for certain tasks.
AI is most valuable for scaffolding, boilerplate generation, and preliminary content. The real value is what you add yourself and how you refine it.

Create separate agents for distinct development phases to maintain clarity and enable careful review.
Custom agents provide more control and flexibility than built-in planning tools.
Use YAML front matter to specify agent metadata, tools, and permissions following the principle of least privilege.
Each agent output should be reviewed and refined before proceeding to the next phase.
Specification-driven development reduces ambiguity and keeps developers in control of the process.
Extend this approach by tailoring agents to your domain, adding new phases, or creating agents for maintenance and agile workflows.