APPENDIX H: RESUME PORTFOLIO BUILDING CAPSTONE PROJECTS

ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS These two prompts reflect tasks a working harness engineer faces after finishing a first harness. Replace each bracketed placeholder with your own material before use, and never paste secret keys or real personal data into an assistant. [START OF PROMPT] I have built a small Python agent harness with a tool registry, a folder-based permission gate, a step limit, and a JSON Lines log. Here is a summary of how the gate works: [DESCRIBE YOUR GATE RULES IN PLAIN WORDS]. Act as a careful security reviewer. List ten different path strings or tool arguments that someone might use to try to escape the allowed folders, such as paths with parent-folder segments, absolute paths, shortcuts, or unusual characters. For each one, say what a correct gate should do and how I could write a small automated test for it. Do not give me exploit code; keep the examples to harmless test strings. Output format: a numbered list with the headings Attempt, Expected Gate Behavior, and Test Idea. [END OF PROMPT] [START OF PROMPT] I am preparing a portfolio entry for a small agent harness project. Facts you may use, and nothing else: [PASTE YOUR TRUE FACTS: WHAT IT DOES, NUMBER OF TESTS, WHICH MODEL YOU USED, WHAT IS NOT DONE]. Write three resume bullet options that are truthful, specific, and under 30 words each, using strong verbs but no exaggeration. Then write a five-sentence project summary for a readme file in plain language for a non-technical recruiter. Finally, point out any phrase in my facts that could be misread as a claim stronger than the evidence supports. Output format: Bullet Options, Readme Summary, Possible Overstatements. [END OF PROMPT] PROJECT 2: THE GARDEN LIBRARIAN (A CONTEXT-MANAGED RESEARCH AGENT)