Examples are organized by the current paper taxonomy: R<category>.<risk>.
Each risk directory contains a run_r<category>_<risk>_*.py entry point and, when needed, a configs/ directory.
| Risk | Directory | Main entry point |
|---|---|---|
| R1.1 Tacit Collusion | R1.1/ |
run_r1_1_tacit_collusion.py |
| R1.3 Competitive Task Avoidance | R1.3/ |
run_r1_3_competitive_task_avoidance.py |
| R1.4 Strategic Misreporting | R1.4/ |
run_r1_4_strategic_misreporting.py |
| R1.5 Information Asymmetry Exploitation | R1.5/ |
run_r1_5_information_asymmetry_exploitation.py |
| R3.1 Normative Deadlock | R3.1/ |
run_r3_1_normative_deadlock.py |
| R3.2 Rigidity & Mistaken Commitments | R3.2/ |
run_r3_2_rigidity_report.py |
| R3.3 Clarification Failure | R3.3/ |
run_r3_3_clarification_failure.py |
| R3.4 Role Allocation Failure | R3.4/ |
run_r3_4_role_allocation_failure.py |
R3.2/ merges the former R13 YAML/framework implementation and R32_Rigidity report implementation. The former is preserved as run_r3_2_framework.py.
Toolkit_MCP_Skills/ is a non-risk toolkit integration quickstart.