Why AI-written test cases miss the business—and how to give a Skill the right context
A context-led guide to using Requirements Analysis, Test Case Writing, and Test Case Reviewer with real project files and business rules.
#Agent Skills
Skill Regression Testing: Proving Changes Did Not Break Existing Behavior
Assess description, instruction, reference, model, and runtime changes with targeted regression evidence that catches silent Skill failures.
Agent Skill Compatibility: Agents, Models, Tools, and Environments
Bring Agent Host, Model, Tools, Environment, and Language into compatibility evidence, and separate specification compatibility from runtime compatibility.
Skill Benchmarking: Is the Skill Actually Better Than the Baseline?
Use a controlled with-Skill versus without-Skill comparison to determine whether an Agent Skill creates real uplift instead of treating one PASS as effectiveness evidence.
Designing Skill Evals: From Success Criteria to Evidence
Turn Skill success criteria into executable Eval Cases, choose a suitable judge, and preserve run metadata, traces, artifacts, and limitations.
Skill Behavior Testing: Evaluate Contracts, Not Exact Answers
Define observable Agent Skill behavior with MUST, SHOULD, and MUST NOT, then evaluate outcomes, trajectories, artifacts, and side effects instead of fixed wording.
Skill Trigger Testing: Can the Agent Select the Right Skill?
Test Skill routing from discovery to trigger and neighboring capability selection with positive, negative, boundary, and paraphrase cases.
The Skill Testing Pyramid: How Much Testing Is Enough?
Use L1 through L4 test depth and E1 through E4 evidence levels to choose how far an Agent Skill should be tested.
Defining Agent Skill Quality: The SQEM Model
Use SQEM’s seven quality domains and eight principles to connect Agent Skill quality claims to evidence without hiding everything behind one score.
What Is an Agent Skill? From Specification to Runtime
Break an Agent Skill into SKILL.md, metadata, references, progressive disclosure, and the runtime chain that turns a package into behavior.
Testing Agent Skills: Why “It Works” Does Not Mean “It’s Good”
Why a valid SKILL.md, a successful invocation, and plausible output are still insufficient evidence of Agent Skill quality.
LLM Evaluation Design: From task sets and scoring criteria to model regression gates
Design LLM evaluations with representative task sets, calibrated rubrics, judge audits, and regression thresholds tied to real tasks.