Include all .md files in vendor.tau2 data to fix simulation_guidelines errors

Author: xzrderek

eval_protocol/benchmarks/test_tau_bench_airline.py

@@ -119,7 +119,7 @@ def tau_bench_airline_to_evaluation_row(data: List[Dict[str, Any]]) -> List[Eval
     rollout_processor=MCPGymRolloutProcessor(),
     rollout_processor_kwargs={"domain": "airline"},
     passed_threshold={"success": 0.4, "standard_error": 0.02},
-    num_runs=8,
+    num_runs=2,
     mode="pointwise",
     max_concurrent_rollouts=50,
     server_script_path=_get_server_script_path(),

eval_protocol/pytest/utils.py

@@ -300,12 +300,12 @@ async def execute_row_with_backoff(task: asyncio.Task, row: EvaluationRow) -> Ev
                     except Exception as retry_error:
                         # Backoff gave up
                         row.rollout_status.status = RolloutStatus.Status.ERROR
-                        row.rollout_status.termination_reason = str(retry_error)
+                        # row.rollout_status.termination_reason = str(retry_error)
                         return row
                 else:
                     # Non-retryable exception - fail immediately
                     row.rollout_status.status = RolloutStatus.Status.ERROR
-                    row.rollout_status.termination_reason = str(e)
+                    # row.rollout_status.termination_reason = str(e)
                     return row
 
         # Process all tasks concurrently with backoff retry

pyproject.toml

@@ -147,7 +147,7 @@ include = ["eval_protocol*", "development*", "vendor*"]
 "eval_protocol" = ["../vite-app/dist/**/*"]
 "eval_protocol.mcp_servers.tau2" = ["*.md", "tests/system_prompts/*.md"]
 "eval_protocol.benchmarks" = ["data/*.jsonl"]
-"vendor.tau2" = ["data/domains/*/policy.md"]
+"vendor.tau2" = ["data/**/*.md"]
 
 
 [tool.versioneer]