Model Online / On-Site
Duration
Modules 5 modules

Course Objective & Scope

Having every developer re-explain build commands, architectural boundaries and naming conventions in every chat session wastes time—yet dumping thousands of lines into a monolithic instruction file dilutes model attention until rules are ignored. This course teaches how to engineer concise, modular repository instruction systems that coding agents actually follow.

Participants master the open AGENTS.md standard alongside CLAUDE.md, .cursor/rules/*.mdc and .github/copilot-instructions.md, directory hierarchies, glob-scoped activation, canonical positive/negative examples, and regression-testing instruction effectiveness on their own repositories.

Who Should Attend?

  • Tech leads and software architects
  • Platform, DevEx and developer-tooling engineers
  • Senior software engineers
  • Prerequisite: experience with Git workflows and at least one AI coding assistant or agent

Key Takeaways

  • Set up a single-source AGENTS.md architecture compatible across IDEs and CLI coding agents
  • Write high-signal instructions covering exact build/test commands, module boundaries and golden file references
  • Scope rules by path and file glob so frontend, backend and database rules load only when relevant
  • Offload formatting and syntax rules to deterministic linters/hooks instead of wasting prompt tokens
  • Convert recurring PR review feedback systematically into version-controlled repo rules

Curriculum

The Instruction Ecosystem: AGENTS.md, CLAUDE.md and IDE Rules
  • How the AGENTS.md standard, CLAUDE.md, .cursor/rules and Copilot instructions are loaded
  • Maintaining a single source of truth across multi-tool engineering teams
Writing High-Signal Instructions and Avoiding Anti-Patterns
  • Essential sections: architecture map, exact build/test commands, domain constraints and gotchas
  • Pruning bloated, contradictory or vague prose that causes instruction following to degrade
Hierarchical and Glob-Scoped Modular Rule Architecture
  • Structuring root vs. sub-package AGENTS.md files in monorepos and layered apps
  • Configuring frontmatter and file globs so rules activate only on relevant paths
Separating Deterministic Linters/Hooks from Natural-Language Rules
  • Deciding what belongs in ESLint/Ruff/ArchUnit/pre-commit hooks versus AGENTS.md
  • Pointing agents to canonical golden files in the repo instead of pasting long code blocks
Testing, Versioning and Evolving Rules from PR Feedback
  • Validating rule adherence with representative coding tasks (instruction evals)
  • Turning repeated PR review comments into living repository instruction updates

Other Courses in This Category

  • Spec-Driven Development (SDD) with Coding Agents

    Training on making structured specifications (SPEC.md, OpenAPI, JSON Schema, TypeSpec, state machines and executable acceptance tests) the primary engineering artifact that drives coding agents deterministically.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View
  • Feature Delivery, Refactoring and Migrations with Coding Agents

    Practical playbook for executing the three core engineering workloads with coding agents: end-to-end multi-file feature delivery, large-scale refactoring and framework/SDK/database migrations combining codemods with agent loops.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View
  • Agentic Testing and Autonomous QA Workflows

    Hands-on training in autonomous QA workflows using Playwright MCP, browser-driving testing agents, self-healing E2E tests and autonomous exploratory testing across UI and API user journeys.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View
  • AI-Assisted Code Review and Human Approval Workflows

    Solving the PR review bottleneck in AI-accelerated teams through stacked diffs, low-noise AI review bots, evidence-backed PR templates and risk-tiered human approval workflows.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View
  • Sandboxing, Tool Permissions and Safe Execution for Coding Agents

    Advanced training on running autonomous coding agents safely on developer machines and servers using container/microVM sandboxes, filesystem/network egress controls, least-privilege MCP permissions and secret isolation.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View
  • Team Workflows, Reusable Agent Skills and Context Handoffs

    Scaling AI coding from solo tricks to team workflows: building version-controlled agent skills (SKILL.md), custom slash commands, and structured context handoffs across sessions, agents and engineering roles.

    AI-Assisted Software Development Corporate
    Live Course
    5 modules
    View

Get in touch