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:
Bryan Bednarski
2026-06-03 06:36:46 -07:00
committed by Teknium
parent a78c73f3aa
commit 0d9b7132ff
26 changed files with 3188 additions and 140 deletions
+237 -3
View File
@@ -19,7 +19,7 @@ import os
import random
import threading
import time
from typing import Optional
from typing import Any, Optional
from agent.display import (
KawaiiSpinner,
@@ -58,6 +58,78 @@ def _ra():
return run_agent
def _emit_terminal_post_tool_call(
agent,
*,
function_name: str,
function_args: dict,
result: Any,
effective_task_id: str,
tool_call_id: str,
duration_ms: int = 0,
status: str | None = None,
error_type: str | None = None,
error_message: str | None = None,
) -> None:
try:
from model_tools import _emit_post_tool_call_hook, _tool_result_observer_fields
if status is None:
status, error_type, error_message = _tool_result_observer_fields(result)
_emit_post_tool_call_hook(
function_name=function_name,
function_args=function_args,
result=result,
task_id=effective_task_id or "",
session_id=getattr(agent, "session_id", "") or "",
tool_call_id=tool_call_id or "",
turn_id=getattr(agent, "_current_turn_id", "") or "",
api_request_id=getattr(agent, "_current_api_request_id", "") or "",
duration_ms=duration_ms,
status=status,
error_type=error_type,
error_message=error_message,
)
except Exception:
pass
def _cancelled_tool_result(reason: str = "user interrupt") -> str:
return json.dumps(
{
"error": f"Tool execution cancelled by {reason}",
"status": "cancelled",
},
ensure_ascii=False,
)
def _emit_cancelled_terminal_post_tool_call(
agent,
*,
function_name: str,
function_args: dict,
effective_task_id: str,
tool_call_id: str,
start_time: float,
reason: str = "user interrupt",
error_type: str = "keyboard_interrupt",
) -> str:
result = _cancelled_tool_result(reason)
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=result,
effective_task_id=effective_task_id,
tool_call_id=tool_call_id,
duration_ms=int((time.time() - start_time) * 1000),
status="cancelled",
error_type=error_type,
error_message=f"Tool execution cancelled by {reason}",
)
return result
def _tool_search_scoped_names(agent) -> frozenset:
"""Return the deferrable tool names the session may invoke via tool_call.
@@ -188,22 +260,61 @@ def execute_tool_calls_concurrent(agent, assistant_message, messages: list, effe
if _ts_scope_block is not None:
# Out-of-scope tool_call: reject before hooks/guardrails/dispatch.
block_result = _ts_scope_block
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=block_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
status="blocked",
error_type="tool_scope_block",
error_message=_ts_scope_block,
)
else:
try:
from hermes_cli.plugins import get_pre_tool_call_block_message
block_message = get_pre_tool_call_block_message(
function_name, function_args, task_id=effective_task_id or "",
function_name,
function_args,
task_id=effective_task_id or "",
session_id=getattr(agent, "session_id", "") or "",
tool_call_id=getattr(tool_call, "id", "") or "",
turn_id=getattr(agent, "_current_turn_id", "") or "",
api_request_id=getattr(agent, "_current_api_request_id", "") or "",
)
except Exception:
block_message = None
if block_message is not None:
block_result = json.dumps({"error": block_message}, ensure_ascii=False)
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=block_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
status="blocked",
error_type="plugin_block",
error_message=block_message,
)
else:
guardrail_decision = agent._tool_guardrails.before_call(function_name, function_args)
if not guardrail_decision.allows_execution:
block_result = agent._guardrail_block_result(guardrail_decision)
blocked_by_guardrail = True
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=block_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
status="blocked",
error_type="guardrail_block",
error_message=getattr(guardrail_decision, "message", None) or "Tool blocked by guardrail policy",
)
# ── Checkpoint preflight (only for tools that will execute) ──
if block_result is None:
@@ -315,6 +426,23 @@ def execute_tool_calls_concurrent(agent, assistant_message, messages: list, effe
messages=messages,
pre_tool_block_checked=True,
)
except KeyboardInterrupt:
try:
agent.interrupt("keyboard interrupt")
except Exception:
pass
result = _emit_cancelled_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
start_time=start,
)
duration = time.time() - start
logger.info("tool %s cancelled (%.2fs)", function_name, duration)
results[index] = (function_name, function_args, result, duration, True, False)
return
except Exception as tool_error:
result = f"Error executing tool '{function_name}': {tool_error}"
logger.error("_invoke_tool raised for %s: %s", function_name, tool_error, exc_info=True)
@@ -426,8 +554,30 @@ def execute_tool_calls_concurrent(agent, assistant_message, messages: list, effe
# Tool was cancelled (interrupt) or thread didn't return
if agent._interrupt_requested:
function_result = f"[Tool execution cancelled — {name} was skipped due to user interrupt]"
_emit_terminal_post_tool_call(
agent,
function_name=name,
function_args=args,
result=function_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tc, "id", "") or "",
status="cancelled",
error_type="keyboard_interrupt",
error_message="Tool execution cancelled by user interrupt",
)
else:
function_result = f"Error executing tool '{name}': thread did not return a result"
_emit_terminal_post_tool_call(
agent,
function_name=name,
function_args=args,
result=function_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tc, "id", "") or "",
status="error",
error_type="thread_missing_result",
error_message=function_result,
)
tool_duration = 0.0
else:
function_name, function_args, function_result, tool_duration, is_error, blocked = r
@@ -592,13 +742,21 @@ def execute_tool_calls_sequential(agent, assistant_message, messages: list, effe
# Check plugin hooks for a block directive before executing.
_block_msg: Optional[str] = None
_block_error_type = "plugin_block"
if _ts_scope_block is not None:
_block_msg = _ts_scope_block
_block_error_type = "tool_scope_block"
else:
try:
from hermes_cli.plugins import get_pre_tool_call_block_message
_block_msg = get_pre_tool_call_block_message(
function_name, function_args, task_id=effective_task_id or "",
function_name,
function_args,
task_id=effective_task_id or "",
session_id=getattr(agent, "session_id", "") or "",
tool_call_id=getattr(tool_call, "id", "") or "",
turn_id=getattr(agent, "_current_turn_id", "") or "",
api_request_id=getattr(agent, "_current_api_request_id", "") or "",
)
except Exception:
pass
@@ -687,11 +845,33 @@ def execute_tool_calls_sequential(agent, assistant_message, messages: list, effe
# Tool blocked by plugin policy — return error without executing.
function_result = json.dumps({"error": _block_msg}, ensure_ascii=False)
tool_duration = 0.0
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=function_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
status="blocked",
error_type=_block_error_type,
error_message=_block_msg,
)
elif _guardrail_block_decision is not None:
# Tool blocked by tool-loop guardrail — synthesize exactly one
# tool result for the original tool_call_id without executing.
function_result = agent._guardrail_block_result(_guardrail_block_decision)
tool_duration = 0.0
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=function_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
status="blocked",
error_type="guardrail_block",
error_message=getattr(_guardrail_block_decision, "message", None) or "Tool blocked by guardrail policy",
)
elif function_name == "todo":
from tools.todo_tool import todo_tool as _todo_tool
function_result = _todo_tool(
@@ -850,12 +1030,29 @@ def execute_tool_calls_sequential(agent, assistant_message, messages: list, effe
function_name, function_args, effective_task_id,
tool_call_id=tool_call.id,
session_id=agent.session_id or "",
turn_id=getattr(agent, "_current_turn_id", "") or "",
api_request_id=getattr(agent, "_current_api_request_id", "") or "",
enabled_tools=list(agent.valid_tool_names) if agent.valid_tool_names else None,
skip_pre_tool_call_hook=True,
enabled_toolsets=getattr(agent, "enabled_toolsets", None),
disabled_toolsets=getattr(agent, "disabled_toolsets", None),
)
_spinner_result = function_result
except KeyboardInterrupt:
function_result = _emit_cancelled_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
start_time=tool_start_time,
)
_spinner_result = function_result
try:
agent.interrupt("keyboard interrupt")
except Exception:
pass
raise
except Exception as tool_error:
function_result = f"Error executing tool '{function_name}': {tool_error}"
logger.error("handle_function_call raised for %s: %s", function_name, tool_error, exc_info=True)
@@ -872,11 +1069,27 @@ def execute_tool_calls_sequential(agent, assistant_message, messages: list, effe
function_name, function_args, effective_task_id,
tool_call_id=tool_call.id,
session_id=agent.session_id or "",
turn_id=getattr(agent, "_current_turn_id", "") or "",
api_request_id=getattr(agent, "_current_api_request_id", "") or "",
enabled_tools=list(agent.valid_tool_names) if agent.valid_tool_names else None,
skip_pre_tool_call_hook=True,
enabled_toolsets=getattr(agent, "enabled_toolsets", None),
disabled_toolsets=getattr(agent, "disabled_toolsets", None),
)
except KeyboardInterrupt:
_emit_cancelled_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
start_time=tool_start_time,
)
try:
agent.interrupt("keyboard interrupt")
except Exception:
pass
raise
except Exception as tool_error:
function_result = f"Error executing tool '{function_name}': {tool_error}"
logger.error("handle_function_call raised for %s: %s", function_name, tool_error, exc_info=True)
@@ -895,6 +1108,27 @@ def execute_tool_calls_sequential(agent, assistant_message, messages: list, effe
# Log tool errors to the persistent error log so [error] tags
# in the UI always have a corresponding detailed entry on disk.
_is_error_result, _ = _detect_tool_failure(function_name, function_result)
# The agent-runtime tools above (todo, session_search, memory,
# context-engine, memory-manager, clarify, delegate_task) are
# dispatched inline — they never reach handle_function_call, so the
# executor is the one that has to fire post_tool_call. For
# registry-dispatched tools the else-branch above invoked
# handle_function_call, which already fires the hook.
from agent.agent_runtime_helpers import agent_runtime_owns_post_tool_hook
_executor_must_emit_post_hook = (
not _execution_blocked
and agent_runtime_owns_post_tool_hook(agent, function_name)
)
if _executor_must_emit_post_hook:
_emit_terminal_post_tool_call(
agent,
function_name=function_name,
function_args=function_args,
result=function_result,
effective_task_id=effective_task_id,
tool_call_id=getattr(tool_call, "id", "") or "",
duration_ms=int(tool_duration * 1000),
)
if not _execution_blocked:
function_result = agent._append_guardrail_observation(
function_name,