Module 2 — Multi-agent team patterns

18 patterns diagnosing how multi-agent crews work (or don't). Includes the foundational After-Action Review (#30) that every other pattern composes off.

The 18 patterns:

1. #13 GRPI — vstack_grpi — generate a working agreement from a team setup request.
2. #14 Process gain/loss — vstack_process_gain_loss — Steiner; did the crew beat best-individual or pay coordination cost?
3. #15 Social loafing — vstack_social_loafing — Latane-Williams-Harkins.
4. #16 Superflocks — vstack_superflocks — Heffernan; routing layer over-concentrates traffic.
5. #17 Lencioni — vstack_lencioni — five dysfunctions pyramid (the layered diagnostic).
6. #18 Trust Triangle — vstack_trust_triangle — Frei & Morriss: logic / authenticity / empathy.
7. #19 McAllister trust — vstack_mcallister_trust — cognitive- vs. affective-based trust.
8. #20 Edmondson — vstack_psych_safety — psychological-safety scale.
9. #21 Glaser — vstack_glaser_conversation — Level I/II/III conversational intelligence.
10. #22 Stone-Heen — vstack_feedback_triggers — truth/relationship/identity triggers.
11. #23 Plus/Delta — vstack_plus_delta — structured behavioral inter-agent feedback.
12. #24 SMART goals — vstack_smart_goal — Doran SMART criteria + kill criteria.
13. #25 Group decision — vstack_group_decision — facilitator-canon decision models + protocol generator.
14. #26 Debate pathology — vstack_debate_pathology — groupthink / polarization / contagion.
15. #27 Bias stack — vstack_bias_stack — Kahneman/Tversky cognitive biases.
16. #28 Devil's advocate — vstack_devils_advocate — role-separation diagnostic.
17. #29 Thomas-Kilmann — vstack_thomas_kilmann — five conflict styles.
18. #30 AAR (foundational) — vstack_aar — Wharton 4-step After-Action Review.

See PATTERNS.md at the repo root for full per-pattern detail including citations.