AI-Assisted Development.

Leveraging AI coding assistants effectively: context engineering, code review, pair programming, and prompt crafting. Covers 8 sub-skills.

• AI-Assisted Development · 1 of 3

  AI Limitations & Hallucinations

  An agent is asked about a library released six months ago. What is the safest default approach to avoid stale guidance?

  AAsk the user to paste the library README into the chat insteadBAnswer from the model's pretraining knowledge of the libraryCRefuse to discuss a library released after the model cutoffDPull the current docs through a fetch tool before answering

• AI-Assisted Development · 2 of 3

  AI Testing & Validation

  When evaluating a mutation-test report against a coverage report on the same suite, which question does the mutation report answer that coverage cannot?

  AWhether the source files have at least one matching test file by nameBWhether the suite executes each line and branch in the source treeCWhether the assertions actually fail when the code is alteredDWhether each test file imports the module that the test is exercising

• AI-Assisted Development · 3 of 3

  AI-Assisted Refactoring

  Why are AI-assisted refactors of legacy languages like COBOL particularly risky despite the AI knowing the syntax?

  ACOBOL programs are too short to provide meaningful context for AI refactoringBCOBOL syntax was excluded from major LLM training datasets for licensing reasonsCNo grasp of the runtime behavior, memory models or the platform quirksDLegacy languages cannot be parsed by the tokenizers used in modern LLMs