CubeLoop — a Pythonic, async-native agent harness core designed for high performance, readability, and production-grade persistence.
Python agent-frameworkagent-harnessagent-runtimeagentic-aiai-agentsai-frameworkanthropicharnessharness-engineeringharness-frameworklanggraphlanggraph-alternativellmmcpopenaipi-agentpi-agent-corepythonpython-agent
from __future__ import annotations
import json
from cubeloop.cli.trace.follow import format_event, is_run_complete, iter_new_spans
from cubeloop.cli.trace.model import Span
def _line(span_id, parent_id, name):
return json.dumps(
{
"name": name,
"context": {"trace_id": "0xt", "span_id": span_id},
"parent_id": parent_id,
"start_time": "2026-05-20T00:00:00Z",
"end_time": "2026-05-20T00:00:00.1Z",
"status": {"status_code": "UNSET"},
"attributes": {"gen_ai.operation.name": name},
}
)
def test_iter_new_spans_incremental(tmp_path):
f = tmp_path / "r1.jsonl"
f.write_text(_line("0x2", "0x1", "chat") + "\n")
state = {"offset": 0, "buffer": ""}
spans = list(iter_new_spans(f, state))
assert [s.name for s in spans] == ["chat"]
# Append more; only the new span is yielded.
with f.open("a") as fh:
fh.write(_line("0x3", "0x1", "execute_tool") + "\n")
spans = list(iter_new_spans(f, state))
assert [s.name for s in spans] == ["execute_tool"]
def test_partial_line_held_until_newline(tmp_path):
f = tmp_path / "r1.jsonl"
payload = _line("0x2", "0x1", "chat")
f.write_text(payload) # no trailing newline yet
state = {"offset": 0, "buffer": ""}
assert list(iter_new_spans(f, state)) == [] # held in buffer
with f.open("a") as fh:
fh.write("\n")
spans = list(iter_new_spans(f, state))
assert [s.name for s in spans] == ["chat"]
def test_is_run_complete_on_root():
root = Span(json.loads(_line("0x1", None, "invoke_agent")))
child = Span(json.loads(_line("0x2", "0x1", "chat")))
assert is_run_complete(child) is False
assert is_run_complete(root) is True
def test_format_event_contains_name():
sp = Span(json.loads(_line("0x2", "0x1", "chat")))
assert "chat" in format_event(sp)