counter_1 Foundations — Ship with Agents

Your engineers are ready to start working with agents. This engagement helps them get setup.

Author

Select and integrate agents into the development workflow. Establish patterns for prompting, context management, and permission handling that work with your existing practices.

• Evaluate team- and project-specific confidentiality, privacy, and security requirements
• Select agentic service providers, package levels and cost monitoring processes
• Recommend agent tooling and integrations (IDEs vs terminal based, backend vs mobile vs frontend)
• Capture guidelines (coding, security, infrastructure, testing, …), identify owners and setup feedback loops
• Define model for engineers to discover and fetch relevant artifacts
• Ensure well-defined conventions for requirements (including user story, acceptance criteria, success and failure paths, error handling, logging and metrics)
• Identify risk areas and manage default permission sets
• Learn effective planning mode, context management practices, persistent memory tooling, skills, commands
• Setup communication channels and community of practice (CoP) to disseminate learnings

Test

Build confidence in agent-generated code. Develop testing strategies that validate agent output and catch issues before they reach production.

• Determine, setup or build MCP servers to access testing infrastructure
• Automate deployment and management of testing infrastructure
• Setup testing feedback loops to teach agents how to validate code themselves
• Learn how to use TDD effectively
• Establish metrics for agent-assisted code quality

Ship

Deploy with confidence. Create the guardrails and review processes that let your team ship agent-assisted work at production quality.

• Establish partial commit and push workflows to enable early feedback
• Set up automated agent review for published branches and PRs
• Create code review processes specific to agent generated code

Engineers who want to integrate AI agents into their daily workflow without sacrificing code quality or team velocity.

Your team ships with agents. Engineers have the skills, patterns, and confidence to leverage AI agents as part of their standard development process.