feat(observability): observer-grade telemetry hooks + NeMo-Relay plugin
Adds backend-neutral observer hooks for plugins: session, turn, API request, tool, approval, and subagent lifecycle events with stable correlation IDs (session_id, task_id, turn_id, api_request_id, tool_call_id, parent/child subagent ids). Extends VALID_HOOKS with api_request_error and subagent_start. Hot path is zero-cost when no plugin subscribes: has_hook()/presence checks gate all payload construction, request payloads are returned by reference when no middleware rewrites, and the sanitized response payload no longer embeds raw response objects. Bundles the optional NeMo-Relay observability plugin (plugins/observability/nemo_relay) as an in-repo consumer of the new hooks, peer to the existing langfuse plugin. Fails open when the optional nemo-relay package is not installed. Authored-by: Bryan Bednarski <bbednarski@nvidia.com> Salvaged from #29722 onto current main.
This commit is contained in:
@@ -1,4 +1,5 @@
|
||||
from unittest.mock import MagicMock, patch
|
||||
from types import SimpleNamespace
|
||||
from hermes_cli.plugins import VALID_HOOKS, PluginManager
|
||||
from cli import HermesCLI
|
||||
|
||||
@@ -20,12 +21,18 @@ def test_session_finalize_on_reset(mock_invoke_hook):
|
||||
cli.new_session(silent=True)
|
||||
|
||||
# Check if on_session_finalize was called for the old session
|
||||
mock_invoke_hook.assert_any_call(
|
||||
"on_session_finalize", session_id="test-session-id", platform="cli"
|
||||
assert any(
|
||||
c.args == ("on_session_finalize",)
|
||||
and c.kwargs["session_id"] == "test-session-id"
|
||||
and c.kwargs["platform"] == "cli"
|
||||
for c in mock_invoke_hook.call_args_list
|
||||
)
|
||||
# Check if on_session_reset was called for the new session
|
||||
mock_invoke_hook.assert_any_call(
|
||||
"on_session_reset", session_id=cli.session_id, platform="cli"
|
||||
assert any(
|
||||
c.args == ("on_session_reset",)
|
||||
and c.kwargs["session_id"] == cli.session_id
|
||||
and c.kwargs["platform"] == "cli"
|
||||
for c in mock_invoke_hook.call_args_list
|
||||
)
|
||||
|
||||
|
||||
@@ -41,11 +48,45 @@ def test_session_finalize_on_cleanup(mock_invoke_hook):
|
||||
|
||||
cli_mod._run_cleanup()
|
||||
|
||||
mock_invoke_hook.assert_any_call(
|
||||
"on_session_finalize", session_id="cleanup-session-id", platform="cli"
|
||||
assert any(
|
||||
c.args == ("on_session_finalize",)
|
||||
and c.kwargs["session_id"] == "cleanup-session-id"
|
||||
and c.kwargs["platform"] == "cli"
|
||||
and c.kwargs["reason"] == "shutdown"
|
||||
for c in mock_invoke_hook.call_args_list
|
||||
)
|
||||
|
||||
|
||||
@patch("hermes_cli.plugins.invoke_hook")
|
||||
def test_interrupted_session_end_helper_emits_observer_shape(mock_invoke_hook):
|
||||
"""Verify quiet single-query interruption emits a correlated session end."""
|
||||
import cli as cli_mod
|
||||
|
||||
mock_agent = MagicMock()
|
||||
mock_agent.session_id = "agent-session-id"
|
||||
mock_agent.model = "test-model"
|
||||
mock_agent.platform = "cli"
|
||||
mock_agent._current_task_id = "task-1"
|
||||
mock_agent._current_turn_id = "turn-1"
|
||||
mock_agent._current_api_request_id = "api-1"
|
||||
cli = SimpleNamespace(agent=mock_agent, session_id="cli-session-id")
|
||||
|
||||
cli_mod._emit_interrupted_session_end(cli, reason="keyboard_interrupt")
|
||||
|
||||
mock_agent.interrupt.assert_called_once_with("keyboard interrupt")
|
||||
assert cli.session_id == "agent-session-id"
|
||||
mock_invoke_hook.assert_called_once()
|
||||
call = mock_invoke_hook.call_args
|
||||
assert call.args == ("on_session_end",)
|
||||
assert call.kwargs["session_id"] == "agent-session-id"
|
||||
assert call.kwargs["task_id"] == "task-1"
|
||||
assert call.kwargs["turn_id"] == "turn-1"
|
||||
assert call.kwargs["api_request_id"] == "api-1"
|
||||
assert call.kwargs["completed"] is False
|
||||
assert call.kwargs["interrupted"] is True
|
||||
assert call.kwargs["reason"] == "keyboard_interrupt"
|
||||
|
||||
|
||||
@patch("hermes_cli.plugins.invoke_hook")
|
||||
def test_hook_errors_are_caught(mock_invoke_hook):
|
||||
"""Verify hook exceptions are caught and don't crash the agent."""
|
||||
|
||||
Reference in New Issue
Block a user