mashinsonvv submission — Mentor & Student (Negotiation)

[M-VS-V]

mashinsonvv — Build a Mentor That Knows You Learned

Subject: Negotiation · 10 lessons · two agents (mentor Nadia, student Sam) · model DeepSeek deepseek-chat.

Everything lives in the mashinsonvv/ folder: the single 5-section deliverable is mashinsonvv_submission.pdf, alongside fully runnable code in mentor_student/.

What it does

Two AI agents run a ten-lesson negotiation course entirely through conversation. The student is built to fake having practised on lessons 3, 6 and 9 (and cut a corner on 2 and 8); the mentor must catch the bluff and refuse to advance without real evidence of application, not recall.

• Verification by conversation: ask for the exact words → ask what went wrong (a faker's story is suspiciously smooth) → run a live transfer test on a brand-new case in-chat.
• Memory tool: the mentor keeps a persistent <memory> block (per-lesson status + notes + caught bluffs), hidden from the student, so it circles back to weak spots.
• Advancement rule enforced in code (orchestrator.py): a lesson can't advance until it is verified, so a bluff can never silently move the course forward.

The five sections

1. Mentor prompt — mentor_student/prompts/mentor_system.md
2. Student prompt (with bluff mechanics) — mentor_student/prompts/student_system.md
3. Tools used — mentor_student/submission/tools.md
4. Full untrimmed transcript — mentor_student/output/transcript.md
5. Rationale & self-eval — mentor_student/submission/rationale.md

Honesty note

Our first student never bluffed at all — too agreeable, exactly as the brief warns. We moved the unreliability schedule out of the prompt and into the orchestrator (a private per-turn directive); after that the student bluffed and was caught on lessons 3/6/9. The rationale is candid that, stripped of the scripted confession, the mentor is really a specificity detector rather than a lie detector — a consistent confabulator would likely pass, which is why the live transfer test is the real backstop.

Run: python -m mentor_student.run_course then python -m mentor_student.build_pdf (needs DEEPSEEK_API_KEY; reportlab only for the PDF).