hermes-agent/tests/gateway/test_resume_command.py
claudlos c4f278c021 security(gateway): scope /resume and /sessions to the caller's origin (IDOR)
/resume resolved a persisted session id/title with no ownership check on any
adapter except Matrix, so an authorized caller could bind their gateway session
to another user's/room's transcript and read it. The titled-session listing and
numeric index were also globally enumerable on non-Matrix platforms, exposing
the ids and previews needed to target the IDOR.

Generalize the Matrix-only room guard to an adapter-agnostic ownership check
(live origin when active; DB row source + user_id for persisted-only sessions,
the only fields available), applied to the direct-id/title path and the
listing/numeric paths on every platform. An explicit admin --all override is
honored. The Matrix path is preserved unchanged.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-01 05:38:03 -07:00

615 lines
27 KiB
Python

"""Tests for /resume gateway slash command.
Tests the _handle_resume_command handler (switch to a previously-named session)
across gateway messenger platforms.
"""
from types import SimpleNamespace
from unittest.mock import AsyncMock, MagicMock
import pytest
from gateway.config import Platform
from gateway.platforms.base import MessageEvent
from gateway.session import SessionSource, build_session_key
def _make_event(text="/resume", platform=Platform.TELEGRAM,
user_id="12345", chat_id="67890"):
"""Build a MessageEvent for testing."""
source = SessionSource(
platform=platform,
user_id=user_id,
chat_id=chat_id,
user_name="testuser",
)
return MessageEvent(text=text, source=source)
def _session_key_for_event(event):
"""Get the session key that build_session_key produces for an event."""
return build_session_key(event.source)
def _make_runner(session_db=None, current_session_id="current_session_001",
event=None):
"""Create a bare GatewayRunner with a mock session_store and optional session_db."""
from gateway.run import GatewayRunner
runner = object.__new__(GatewayRunner)
runner.adapters = {}
runner.config = SimpleNamespace(platforms={})
runner._voice_mode = {}
# Gateway holds the async facade; the slash handlers await it.
if session_db is not None:
from hermes_state import AsyncSessionDB
session_db = AsyncSessionDB(session_db)
runner._session_db = session_db
runner._running_agents = {}
runner._is_user_authorized = lambda _source: True
# Compute the real session key if an event is provided
session_key = build_session_key(event.source) if event else "agent:main:telegram:dm"
# Mock session_store that returns a session entry with a known session_id
mock_session_entry = MagicMock()
mock_session_entry.session_id = current_session_id
mock_session_entry.session_key = session_key
mock_store = MagicMock()
mock_store.get_or_create_session.return_value = mock_session_entry
mock_store.load_transcript.return_value = []
mock_store.switch_session.return_value = mock_session_entry
runner.session_store = mock_store
return runner
# ---------------------------------------------------------------------------
# _handle_resume_command
# ---------------------------------------------------------------------------
class TestHandleResumeCommand:
"""Tests for GatewayRunner._handle_resume_command."""
@pytest.mark.asyncio
async def test_no_session_db(self):
"""Returns error when session database is unavailable."""
runner = _make_runner(session_db=None)
event = _make_event(text="/resume My Project")
result = await runner._handle_resume_command(event)
assert "not available" in result.lower()
@pytest.mark.asyncio
async def test_list_named_sessions_when_no_arg(self, tmp_path):
"""With no argument, lists recently titled sessions."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("sess_001", "telegram", user_id="12345")
db.create_session("sess_002", "telegram", user_id="12345")
db.set_session_title("sess_001", "Research")
db.set_session_title("sess_002", "Coding")
event = _make_event(text="/resume")
runner = _make_runner(session_db=db, event=event)
result = await runner._handle_resume_command(event)
assert "Research" in result
assert "Coding" in result
assert "Named Sessions" in result
assert "1." in result
assert "2." in result
assert "/resume 1" in result
db.close()
@pytest.mark.asyncio
async def test_list_shows_usage_when_no_titled(self, tmp_path):
"""With no arg and no titled sessions, shows instructions."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("sess_001", "telegram", user_id="12345") # No title
event = _make_event(text="/resume")
runner = _make_runner(session_db=db, event=event)
result = await runner._handle_resume_command(event)
assert "No named sessions" in result
assert "/title" in result
db.close()
@pytest.mark.asyncio
async def test_resume_by_index(self, tmp_path):
"""Numeric argument resumes the indexed titled session from the list."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("sess_001", "telegram", user_id="12345")
db.create_session("sess_002", "telegram", user_id="12345")
db.set_session_title("sess_001", "Research")
db.set_session_title("sess_002", "Coding")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume 2")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
assert "Resumed" in result
runner.session_store.switch_session.assert_called_once()
call_args = runner.session_store.switch_session.call_args
assert call_args[0][1] == "sess_001"
db.close()
@pytest.mark.asyncio
async def test_resume_index_out_of_range(self, tmp_path):
"""Out-of-range numeric arguments show a helpful error."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("sess_001", "telegram", user_id="12345")
db.set_session_title("sess_001", "Research")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume 9")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
assert "out of range" in result.lower()
assert "/resume" in result
runner.session_store.switch_session.assert_not_called()
db.close()
@pytest.mark.asyncio
async def test_resume_by_name(self, tmp_path):
"""Resolves a title and switches to that session."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("old_session_abc", "telegram", user_id="12345")
db.set_session_title("old_session_abc", "My Project")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume My Project")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
assert "Resumed" in result
assert "My Project" in result
# Verify switch_session was called with the old session ID
runner.session_store.switch_session.assert_called_once()
call_args = runner.session_store.switch_session.call_args
assert call_args[0][1] == "old_session_abc"
db.close()
@pytest.mark.asyncio
async def test_resume_clears_session_model_overrides(self, tmp_path):
"""Resume must not carry a previous session's /model override into the
restored conversation, while leaving other chats' overrides intact (#10702)."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("old_session_abc", "telegram")
db.set_session_title("old_session_abc", "My Project")
db.create_session("current_session_001", "telegram")
event = _make_event(text="/resume My Project")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
key = _session_key_for_event(event)
runner._session_model_overrides = {
key: {"model": "gpt-5", "provider": "openai"},
"agent:main:telegram:dm:other": {"model": "keep-me"},
}
runner._pending_model_notes = {
key: "[Note: switched to gpt-5]",
"agent:main:telegram:dm:other": "[Note: keep-me]",
}
result = await runner._handle_resume_command(event)
assert "Resumed" in result
# The resumed chat's override + pending note are cleared...
assert key not in runner._session_model_overrides
assert key not in runner._pending_model_notes
# ...but an unrelated chat's state is untouched.
assert runner._session_model_overrides["agent:main:telegram:dm:other"] == {"model": "keep-me"}
assert runner._pending_model_notes["agent:main:telegram:dm:other"] == "[Note: keep-me]"
db.close()
@pytest.mark.asyncio
async def test_resume_nonexistent_name(self, tmp_path):
"""Returns error for unknown session name."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume Nonexistent Session")
runner = _make_runner(session_db=db, event=event)
result = await runner._handle_resume_command(event)
assert "No session found" in result
db.close()
@pytest.mark.asyncio
async def test_resume_already_on_session(self, tmp_path):
"""Returns friendly message when already on the requested session."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("current_session_001", "telegram", user_id="12345")
db.set_session_title("current_session_001", "Active Project")
event = _make_event(text="/resume Active Project")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
assert "Already on session" in result
db.close()
@pytest.mark.asyncio
async def test_resume_auto_lineage(self, tmp_path):
"""Asking for 'My Project' when 'My Project #2' exists gets the latest."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("sess_v1", "telegram", user_id="12345")
db.set_session_title("sess_v1", "My Project")
db.create_session("sess_v2", "telegram", user_id="12345")
db.set_session_title("sess_v2", "My Project #2")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume My Project")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
assert "Resumed" in result
# Should resolve to #2 (latest in lineage)
call_args = runner.session_store.switch_session.call_args
assert call_args[0][1] == "sess_v2"
db.close()
@pytest.mark.asyncio
async def test_resume_follows_compression_continuation(self, tmp_path):
"""Gateway /resume should reopen the live descendant after compression."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("compressed_root", "telegram", user_id="12345")
db.set_session_title("compressed_root", "Compressed Work")
db.end_session("compressed_root", "compression")
db.create_session("compressed_child", "telegram", user_id="12345", parent_session_id="compressed_root")
db.append_message("compressed_child", "user", "hello from continuation")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume Compressed Work")
runner = _make_runner(
session_db=db,
current_session_id="current_session_001",
event=event,
)
runner.session_store.load_transcript.side_effect = (
lambda session_id: [{"role": "user", "content": "hello from continuation"}]
if session_id == "compressed_child"
else []
)
result = await runner._handle_resume_command(event)
assert "Resumed session" in result
assert "(1 message)" in result
call_args = runner.session_store.switch_session.call_args
assert call_args[0][1] == "compressed_child"
runner.session_store.load_transcript.assert_called_with("compressed_child")
db.close()
@pytest.mark.asyncio
async def test_resume_clears_running_agent(self, tmp_path):
"""Switching sessions clears any cached running agent."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("old_session", "telegram", user_id="12345")
db.set_session_title("old_session", "Old Work")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume Old Work")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
# Simulate a running agent using the real session key
real_key = _session_key_for_event(event)
runner._running_agents[real_key] = MagicMock()
await runner._handle_resume_command(event)
assert real_key not in runner._running_agents
db.close()
@pytest.mark.asyncio
async def test_resume_evicts_cached_agent(self, tmp_path):
"""Gateway /resume evicts the cached AIAgent so the next message
rebuilds with the correct session_id end-to-end — mirrors /branch
and /reset. Without this, the cached agent's memory provider keeps
writing into the wrong session. See #6672.
"""
import threading
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("old_session", "telegram", user_id="12345")
db.set_session_title("old_session", "Old Work")
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume Old Work")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
# Seed the cache with a fake agent
real_key = _session_key_for_event(event)
runner._agent_cache = {real_key: (MagicMock(), object())}
runner._agent_cache_lock = threading.RLock()
await runner._handle_resume_command(event)
assert real_key not in runner._agent_cache
db.close()
@pytest.mark.asyncio
async def test_resume_strips_outer_brackets(self, tmp_path):
"""Users may copy `<session_id>` from the usage hint literally.
The gateway should strip outer ``<>``, ``[]``, ``""``, and ``''``
before lookup so ``/resume <abc123>`` works the same as
``/resume abc123``.
"""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("abc123", "telegram", user_id="12345")
db.set_session_title("abc123", "Bracketed")
db.create_session("current_session_001", "telegram", user_id="12345")
for raw in ("<abc123>", "[abc123]", '"abc123"', "'abc123'"):
event = _make_event(text=f"/resume {raw}")
runner = _make_runner(
session_db=db,
current_session_id="current_session_001",
event=event,
)
result = await runner._handle_resume_command(event)
# Either the session was resumed (and we get a "Resumed" / "Already on" reply)
# or it was found-then-redirected. Failure mode = "No session found matching '<abc123>'".
assert "abc123" not in str(result) or "not found" not in str(result).lower(), (
f"bracket stripping failed for {raw!r}: gateway returned {result!r}"
)
db.close()
@pytest.mark.asyncio
async def test_resume_resolves_by_session_id(self, tmp_path):
"""The gateway should accept a bare session ID, not just a title.
Before this fix, /resume in the gateway only called
``resolve_session_by_title``, so ``/resume <session_id>`` always
returned "Session not found" even for valid IDs.
"""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("unnamed_session_xyz", "telegram", user_id="12345")
# Deliberately no title set — this session can ONLY be resolved by ID.
db.create_session("current_session_001", "telegram", user_id="12345")
event = _make_event(text="/resume unnamed_session_xyz")
runner = _make_runner(
session_db=db,
current_session_id="current_session_001",
event=event,
)
result = await runner._handle_resume_command(event)
# Should NOT be the not-found error.
assert "not found" not in str(result).lower(), (
f"session-id lookup failed: {result!r}"
)
db.close()
class TestHandleSessionsCommand:
"""Tests for GatewayRunner._handle_sessions_command."""
@pytest.mark.asyncio
async def test_sessions_command_lists_current_platform_sessions(self, tmp_path):
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("tg_session", "telegram", user_id="12345")
db.set_session_title("tg_session", "Telegram Work")
db.create_session("discord_session", "discord")
db.set_session_title("discord_session", "Discord Work")
event = _make_event(text="/sessions")
runner = _make_runner(session_db=db, event=event)
result = await runner._handle_sessions_command(event)
assert "Sessions" in result
assert "Telegram Work" in result
assert "tg_session" in result
assert "Discord Work" not in result
db.close()
@pytest.mark.asyncio
async def test_sessions_all_does_not_leak_cross_origin_for_non_admin(self, tmp_path):
"""`/sessions all` from a non-admin caller must stay scoped to the
caller's own origin — it must NOT enumerate other origins' sessions
(the enumeration half of the /resume IDOR). Cross-origin listing is
gated behind an explicitly-configured admin, which the default test
config is not."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("tg_named", "telegram", user_id="12345")
db.set_session_title("tg_named", "Telegram Work")
db.create_session("discord_unnamed", "discord") # other origin
db.append_message("discord_unnamed", "user", "discord first prompt")
event = _make_event(text="/sessions all full")
runner = _make_runner(session_db=db, event=event)
result = await runner._handle_sessions_command(event)
# Caller's own (telegram) session is shown; the cross-origin (discord)
# session is NOT leaked even with `all`.
assert "Telegram Work" in result
assert "discord_unnamed" not in result
assert "Discord" not in result
db.close()
@pytest.mark.asyncio
async def test_resume_blocks_cross_user_and_unowned_rows(self, tmp_path):
"""An identity-bearing caller cannot resume a session it can't prove it
owns: a row owned by a different user, or a same-platform row with no
recorded owner (NULL user_id) must both be denied (IDOR)."""
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("victim_other_uid", "telegram", user_id="99999")
db.set_session_title("victim_other_uid", "Other User")
db.create_session("victim_missing_uid", "telegram") # NULL owner
db.set_session_title("victim_missing_uid", "Unowned")
db.create_session("current_session_001", "telegram", user_id="12345")
for name in ("Other User", "victim_other_uid", "Unowned", "victim_missing_uid"):
event = _make_event(text=f"/resume {name}")
runner = _make_runner(session_db=db, current_session_id="current_session_001",
event=event)
result = await runner._handle_resume_command(event)
runner.session_store.switch_session.assert_not_called()
assert "Resumed" not in result, name
db.close()
@pytest.mark.asyncio
async def test_gateway_dispatches_sessions_command(self, tmp_path):
from hermes_state import SessionDB
db = SessionDB(db_path=tmp_path / "state.db")
db.create_session("tg_session", "telegram", user_id="12345")
db.set_session_title("tg_session", "Telegram Work")
event = _make_event(text="/sessions")
runner = _make_runner(session_db=db, event=event)
runner._handle_sessions_command = AsyncMock(return_value="sessions output")
result = await runner._handle_message(event)
assert result == "sessions output"
runner._handle_sessions_command.assert_awaited_once_with(event)
db.close()
class TestSameOriginChatGroupScoping:
"""Live group sessions are per-user by default (group_sessions_per_user=True),
so a co-member must not be able to resume another member's live group session
via the live-origin branch of _resume_target_allowed (IDOR)."""
@staticmethod
def _src(user_id, *, chat_type="group", chat_id="guild-123",
platform=Platform.DISCORD, user_id_alt=None, thread_id=None):
return SessionSource(platform=platform, chat_id=chat_id,
chat_type=chat_type, user_id=user_id,
user_id_alt=user_id_alt, thread_id=thread_id)
def test_blocks_cross_user_live_group_by_default(self):
runner = _make_runner()
assert runner._same_origin_chat(self._src("alice"), self._src("bob")) is False
def test_allows_same_user_live_group(self):
runner = _make_runner()
assert runner._same_origin_chat(self._src("alice"), self._src("alice")) is True
def test_allows_cross_user_when_group_explicitly_shared(self):
runner = _make_runner()
runner.config.group_sessions_per_user = False
assert runner._same_origin_chat(self._src("alice"), self._src("bob")) is True
def test_dm_cross_user_still_blocked(self):
runner = _make_runner()
a = self._src("alice", chat_type="dm", chat_id="dm-1")
b = self._src("bob", chat_type="dm", chat_id="dm-1")
assert runner._same_origin_chat(a, b) is False
def test_resume_target_allowed_blocks_cross_user_live_group(self):
"""End-to-end via the live-origin branch: Alice cannot resume Bob's
active group session in the same chat."""
runner = _make_runner()
bob = self._src("bob")
runner._gateway_session_origin_for_id = lambda sid: bob
assert runner._resume_target_allowed(
self._src("alice"), "bobs_live_sid", allow_override=False
) is False
# --- thread scoping: thread_id is part of the session key, so a session in
# one thread must never match a caller in another thread of the same chat,
# even when threads are shared among participants by default. ---
def test_blocks_cross_thread_same_user_same_chat(self):
"""Same user, same parent chat, different thread → different session."""
runner = _make_runner()
a = self._src("alice", thread_id="thread-A")
b = self._src("alice", thread_id="thread-B")
assert runner._same_origin_chat(a, b) is False
def test_allows_same_thread_shared_participants(self):
"""Threads are shared by default (thread_sessions_per_user=False), so
co-members in the SAME thread share the session."""
runner = _make_runner()
a = self._src("alice", thread_id="thread-A")
b = self._src("bob", thread_id="thread-A")
assert runner._same_origin_chat(a, b) is True
def test_blocks_cross_thread_even_when_shared(self):
"""Cross-thread is blocked regardless of thread-sharing: sharing only
applies WITHIN a thread, never across threads."""
runner = _make_runner()
a = self._src("alice", thread_id="thread-A")
b = self._src("bob", thread_id="thread-B")
assert runner._same_origin_chat(a, b) is False
def test_blocks_thread_vs_no_thread(self):
"""A threaded origin must not match a non-threaded caller in the same
parent chat (and vice versa)."""
runner = _make_runner()
threaded = self._src("alice", thread_id="thread-A")
parent = self._src("alice", thread_id=None)
assert runner._same_origin_chat(parent, threaded) is False
assert runner._same_origin_chat(threaded, parent) is False
class TestResumeRowVisibleMatrixAllScoping:
"""Non-admin Matrix `/resume --all` must NOT enumerate every Matrix titled
session: the cross-room listing short-circuit is admin-only, mirroring the
non-Matrix branch. A non-admin `--all` falls back to same-room scoping."""
@staticmethod
def _matrix_src(chat_id="!room-a:hs", user_id="@alice:hs"):
return SessionSource(platform=Platform.MATRIX, chat_id=chat_id,
chat_type="group", user_id=user_id)
def test_non_admin_all_does_not_expose_other_room(self):
runner = _make_runner()
runner._resume_caller_is_admin = lambda src: False
# Titled row whose live origin is a DIFFERENT Matrix room.
other_room = SessionSource(platform=Platform.MATRIX, chat_id="!room-b:hs",
chat_type="group", user_id="@bob:hs")
runner._gateway_session_origin_for_id = lambda sid: other_room
row = {"id": "sid_other_room"}
assert runner._resume_row_visible(self._matrix_src(), row, allow_all=True) is False
def test_non_admin_all_still_shows_same_room(self):
runner = _make_runner()
runner._resume_caller_is_admin = lambda src: False
same_room = SessionSource(platform=Platform.MATRIX, chat_id="!room-a:hs",
chat_type="group", user_id="@bob:hs")
runner._gateway_session_origin_for_id = lambda sid: same_room
row = {"id": "sid_same_room"}
assert runner._resume_row_visible(self._matrix_src(), row, allow_all=True) is True
def test_admin_all_exposes_cross_room(self):
runner = _make_runner()
runner._resume_caller_is_admin = lambda src: True
other_room = SessionSource(platform=Platform.MATRIX, chat_id="!room-b:hs",
chat_type="group", user_id="@bob:hs")
runner._gateway_session_origin_for_id = lambda sid: other_room
row = {"id": "sid_other_room"}
assert runner._resume_row_visible(self._matrix_src(), row, allow_all=True) is True
def test_non_admin_all_fails_closed_on_unknown_origin(self):
runner = _make_runner()
runner._resume_caller_is_admin = lambda src: False
runner._gateway_session_origin_for_id = lambda sid: None
row = {"id": "sid_unknown"}
assert runner._resume_row_visible(self._matrix_src(), row, allow_all=True) is False