feat(plugins): pre_api_request/post_api_request with narrow payloads

- Rename per-LLM-call hooks from pre_llm_request/post_llm_request for clarity vs pre_llm_call
- Emit summary kwargs only (counts, usage dict from normalize_usage); keep env_var_enabled for HERMES_DUMP_REQUESTS
- Add is_truthy_value/env_var_enabled to utils; wire hermes_cli.plugins._env_enabled through it
- Update Langfuse local setup doc; add scripts/langfuse_smoketest.py and optional ~/.hermes plugin tests

Made-with: Cursor
This commit is contained in:
kshitijk4poor
2026-04-06 10:33:13 +05:30
committed by Teknium
parent 9e820dda37
commit f530ef1835
7 changed files with 637 additions and 25 deletions

View File

@@ -0,0 +1,102 @@
"""Smoke tests for the user-installed Langfuse plugin (when present).
The canonical plugin lives under ``~/.hermes/plugins/langfuse_tracing/``.
These tests are skipped in CI unless that directory exists locally.
"""
from __future__ import annotations
import importlib.util
import sys
from pathlib import Path
from unittest.mock import MagicMock, patch
import pytest
PLUGIN_INIT = Path.home() / ".hermes" / "plugins" / "langfuse_tracing" / "__init__.py"
needs_user_plugin = pytest.mark.skipif(
not PLUGIN_INIT.is_file(),
reason="langfuse_tracing plugin not installed at ~/.hermes/plugins/langfuse_tracing/",
)
def _load_user_plugin():
name = "langfuse_tracing_user_plugin"
if name in sys.modules:
return sys.modules[name]
spec = importlib.util.spec_from_file_location(name, PLUGIN_INIT)
if spec is None or spec.loader is None:
raise RuntimeError("cannot load langfuse plugin")
mod = importlib.util.module_from_spec(spec)
sys.modules[name] = mod
spec.loader.exec_module(mod)
return mod
@needs_user_plugin
def test_langfuse_plugin_registers_api_request_hooks():
mod = _load_user_plugin()
ctx = MagicMock()
ctx.manifest.name = "langfuse_tracing"
mod.register(ctx)
registered = [c[0][0] for c in ctx.register_hook.call_args_list]
assert "pre_api_request" in registered
assert "post_api_request" in registered
assert "pre_llm_call" in registered
@needs_user_plugin
def test_pre_post_api_request_smoke_with_mock_langfuse():
mod = _load_user_plugin()
mod._TRACE_STATE.clear()
gen_obs = MagicMock()
root_obs = MagicMock()
root_obs.start_observation.return_value = gen_obs
client = MagicMock()
client.create_trace_id.return_value = "trace-smoke-test"
client.start_observation.return_value = root_obs
with patch.object(mod, "_get_langfuse", return_value=client):
mod.on_pre_api_request(
task_id="t1",
session_id="s1",
platform="cli",
model="test/model",
provider="openrouter",
base_url="https://openrouter.ai/api/v1",
api_mode="chat_completions",
api_call_count=1,
message_count=3,
tool_count=5,
approx_input_tokens=100,
request_char_count=400,
max_tokens=4096,
)
mod.on_post_api_request(
task_id="t1",
session_id="s1",
provider="openrouter",
base_url="https://openrouter.ai/api/v1",
api_mode="chat_completions",
model="test/model",
api_call_count=1,
api_duration=0.05,
finish_reason="stop",
usage={
"input_tokens": 10,
"output_tokens": 20,
"total_tokens": 30,
"reasoning_tokens": 0,
"cache_read_tokens": 0,
"cache_write_tokens": 0,
},
assistant_content_chars=42,
assistant_tool_call_count=0,
response_model="test/model",
)
gen_obs.update.assert_called()
gen_obs.end.assert_called()

View File

@@ -196,9 +196,9 @@ class TestPluginLoading:
class TestPluginHooks:
"""Tests for lifecycle hook registration and invocation."""
def test_valid_hooks_include_request_scoped_llm_hooks(self):
assert "pre_llm_request" in VALID_HOOKS
assert "post_llm_request" in VALID_HOOKS
def test_valid_hooks_include_request_scoped_api_hooks(self):
assert "pre_api_request" in VALID_HOOKS
assert "post_api_request" in VALID_HOOKS
def test_register_and_invoke_hook(self, tmp_path, monkeypatch):
"""Registered hooks are called on invoke_hook()."""
@@ -270,7 +270,11 @@ class TestPluginHooks:
plugins_dir = tmp_path / "hermes_test" / "plugins"
_make_plugin_dir(
plugins_dir, "request_hook",
register_body='ctx.register_hook("pre_llm_request", lambda **kw: {"seen": kw.get("api_call_count")})',
register_body=(
'ctx.register_hook("pre_api_request", '
'lambda **kw: {"seen": kw.get("api_call_count"), '
'"mc": kw.get("message_count"), "tc": kw.get("tool_count")})'
),
)
monkeypatch.setenv("HERMES_HOME", str(tmp_path / "hermes_test"))
@@ -278,15 +282,18 @@ class TestPluginHooks:
mgr.discover_and_load()
results = mgr.invoke_hook(
"pre_llm_request",
"pre_api_request",
session_id="s1",
task_id="t1",
model="test",
api_call_count=2,
messages=[],
tools=[],
message_count=5,
tool_count=3,
approx_input_tokens=100,
request_char_count=400,
max_tokens=8192,
)
assert results == [{"seen": 2}]
assert results == [{"seen": 2, "mc": 5, "tc": 3}]
def test_invalid_hook_name_warns(self, tmp_path, monkeypatch, caplog):
"""Registering an unknown hook name logs a warning."""

View File

@@ -1454,7 +1454,7 @@ class TestRunConversation:
assert mock_handle_function_call.call_args.kwargs["tool_call_id"] == "c1"
assert mock_handle_function_call.call_args.kwargs["session_id"] == agent.session_id
def test_request_scoped_llm_hooks_fire_for_each_api_call(self, agent):
def test_request_scoped_api_hooks_fire_for_each_api_call(self, agent):
self._setup_agent(agent)
tc = _mock_tool_call(name="web_search", arguments="{}", call_id="c1")
resp1 = _mock_response(content="", finish_reason="tool_calls", tool_calls=[tc])
@@ -1477,13 +1477,15 @@ class TestRunConversation:
result = agent.run_conversation("search something")
assert result["final_response"] == "Done searching"
pre_request_calls = [kw for name, kw in hook_calls if name == "pre_llm_request"]
post_request_calls = [kw for name, kw in hook_calls if name == "post_llm_request"]
pre_request_calls = [kw for name, kw in hook_calls if name == "pre_api_request"]
post_request_calls = [kw for name, kw in hook_calls if name == "post_api_request"]
assert len(pre_request_calls) == 2
assert len(post_request_calls) == 2
assert [call["api_call_count"] for call in pre_request_calls] == [1, 2]
assert [call["api_call_count"] for call in post_request_calls] == [1, 2]
assert all(call["session_id"] == agent.session_id for call in pre_request_calls)
assert all("message_count" in c and "messages" not in c for c in pre_request_calls)
assert all("usage" in c and "response" not in c for c in post_request_calls)
def test_interrupt_breaks_loop(self, agent):
self._setup_agent(agent)