fix: address three review comments on env/cli handling

- write_env_file: remove .upper() to preserve original key case
- load_env_vars: strip surrounding single/double quotes from values
- constants.py: fix Ollama key_name from OPENAI_API_BASE to OLLAMA_HOST
- _test_new_agents: replace asyncio.run() loop with new_event_loop + run_until_complete

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
alex-clawd
2026-05-13 12:12:57 -07:00
parent b5396ea290
commit 023bb7e6b8
3 changed files with 15 additions and 4 deletions

View File

@@ -864,6 +864,9 @@ def _test_new_agents(
all_passed = True
agents_tested: set[str] = set()
_loop = asyncio.new_event_loop()
asyncio.set_event_loop(_loop)
for iteration in range(n_iterations):
if n_iterations > 1:
click.secho(f"\n Iteration {iteration + 1}/{n_iterations}", fg="cyan")
@@ -876,10 +879,10 @@ def _test_new_agents(
with ArtifactsSandbox():
if verbose:
with VerboseBenchmarkOutput():
all_results = asyncio.run(_run_all())
all_results = _loop.run_until_complete(_run_all())
else:
with SuppressBenchmarkOutput():
all_results = asyncio.run(_run_all())
all_results = _loop.run_until_complete(_run_all())
finally:
if not verbose:
if progress is None:
@@ -914,6 +917,8 @@ def _test_new_agents(
f" [green bold]{job['agent_name']}: PASSED all {len(results)} cases >= {job['threshold']}[/green bold]"
)
_loop.close()
if len(agents_tested) == 0:
click.secho("No agents completed successfully.", fg="yellow")
raise SystemExit(1)