fix: avoid duplicate agent_run telemetry events

Author: abrichr

openadapt_evals/cli/main.py

@@ -18,9 +18,6 @@
 
 import argparse
 import sys
-import time
-
-from openadapt_evals.telemetry import track_agent_run, track_agent_run_completed
 
 
 def main(argv: list[str] | None = None) -> int:
@@ -252,43 +249,13 @@ def _cmd_mock(args: argparse.Namespace) -> int:
     """Run mock evaluation."""
     # Delegate to existing CLI implementation
     from openadapt_evals.benchmarks.cli import cmd_mock
-    start = time.perf_counter()
-    track_agent_run(
-        phase="start",
-        entrypoint="oa evals mock",
-        mode="mock",
-        agent_class=getattr(args, "agent", "mock"),
-    )
-    rc = cmd_mock(args)
-    track_agent_run_completed(
-        entrypoint="oa evals mock",
-        mode="mock",
-        agent_class=getattr(args, "agent", "mock"),
-        return_code=rc,
-        duration_seconds=round(time.perf_counter() - start, 3),
-    )
-    return rc
+    return cmd_mock(args)
 
 
 def _cmd_run(args: argparse.Namespace) -> int:
     """Run live evaluation."""
     from openadapt_evals.benchmarks.cli import cmd_live
-    start = time.perf_counter()
-    track_agent_run(
-        phase="start",
-        entrypoint="oa evals run",
-        mode="live",
-        agent_class=getattr(args, "agent", "unknown"),
-    )
-    rc = cmd_live(args)
-    track_agent_run_completed(
-        entrypoint="oa evals run",
-        mode="live",
-        agent_class=getattr(args, "agent", "unknown"),
-        return_code=rc,
-        duration_seconds=round(time.perf_counter() - start, 3),
-    )
-    return rc
+    return cmd_live(args)
 
 
 def _cmd_probe(args: argparse.Namespace) -> int:

scripts/record_waa_demos.py

@@ -95,8 +95,6 @@
 
 from openadapt_evals.constants import HARDER_TASK_IDS
 from openadapt_evals.telemetry import (
-    track_agent_run,
-    track_agent_run_completed,
     track_demo_recorded,
 )
 
@@ -2450,22 +2448,8 @@ def cmd_eval_dc(
 
     print(f"Running eval-suite with demo-conditioned demos from {demo_dir}")
     print(f"Command: {' '.join(cmd)}\n")
-    track_agent_run(
-        phase="start",
-        entrypoint="record_waa_demos.py eval",
-        mode="demo-conditioned",
-        num_tasks=len([t.strip() for t in tasks.split(",") if t.strip()]),
-        max_steps=max_steps,
-        run_name=suite_name,
-    )
 
     result = subprocess.run(cmd)
-    track_agent_run_completed(
-        entrypoint="record_waa_demos.py eval",
-        mode="demo-conditioned",
-        return_code=result.returncode,
-        run_name=suite_name,
-    )
     if result.returncode != 0:
         sys.exit(result.returncode)