Jaroslav Beneš
82a7ef5b58
file_write replaces a file entirely, so a model wanting to change one line either rewrote the whole thing from memory -- silently dropping everything it did not happen to recall -- or shelled out to sed. file_edit takes a unified diff instead, and services/agent/patch.py applies it. Four behaviours carry that module, and each exists because of how models actually write patches rather than how the format is specified. Fuzzy offset, exact content. A hunk header is a hint: models count from a truncated read or from the file as it was three edits ago and get the numbers wrong, and get the context lines right. So the hinted position is tried, then the file is scanned outward for an exact match of the context block. One match wins; more than one refuses, because guessing between two identical blocks is the one failure that silently corrupts a file. Line endings are normalised in and restored out, or every hunk on a CRLF file fails on context that looks identical in the error message. A blank context line that lost its leading space is read as blank, because trailing whitespace is stripped by half the things a model's output passes through. And nothing is written unless every hunk applies: a half-applied file is worse than a refused one, and the model cannot tell the difference without reading it again. It refuses a file this reply has not read, in those words. A patch written from memory either fails on context -- the good case -- or matches something it did not mean. AgentContext.read_paths records what was read; it lives there because runners never see a Generation and a read path is a fact about the machine, and it is shared with the approved copy because as_approved is dataclasses.replace, which copies field references. It resets each reply, and that is right rather than a limitation: tool_calls_json is never replayed, so on the next turn the model does not have the contents either. Writes and edits both render a git-style diff in the transcript now, escaped like everything else there and bounded at write time -- a generated file's diff can be larger than the file, and it sits on the row forever. That costs file_write one extra SFTP round trip to read the old contents, on the hottest agent operation, and it is a conscious trade: it is the difference between seeing what an agent did and having to go and look. It earns its keep twice, because that read also counts as having read the file. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
1001 lines
36 KiB
Python
1001 lines
36 KiB
Python
"""The agent tools: when they are offered, and what stops them running.
|
|
|
|
Everything here goes through a real SSH server on 127.0.0.1, so the gate and the
|
|
approval loop are exercised against something that genuinely executes rather
|
|
than a stub that always agrees.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import asyncio
|
|
import json as _json
|
|
import time
|
|
|
|
import pytest
|
|
|
|
from lembas.db.models import (
|
|
KIND_AGENT,
|
|
ROLE_ASSISTANT,
|
|
Chat,
|
|
Connection,
|
|
Message,
|
|
Model,
|
|
SshProfile,
|
|
User,
|
|
)
|
|
from lembas.services import generation as generation_service
|
|
from lembas.services import interaction, settings_store
|
|
from lembas.services import tools as tools_service
|
|
from lembas.services.agent import policy, session
|
|
from lembas.services.agent import ssh as ssh_service
|
|
|
|
asyncssh = pytest.importorskip("asyncssh")
|
|
|
|
|
|
# --- A machine to act on --------------------------------------------------------
|
|
class _Server(asyncssh.SSHServer):
|
|
def begin_auth(self, username: str) -> bool:
|
|
return False
|
|
|
|
|
|
async def _handler(process):
|
|
command = process.command or ""
|
|
process.stdout.write(f"ran: {command}\n")
|
|
process.exit(0)
|
|
|
|
|
|
@pytest.fixture
|
|
async def machine(tmp_path):
|
|
"""A real sshd, and a profile pointing at it with its key already pinned."""
|
|
project = tmp_path / "project"
|
|
project.mkdir()
|
|
|
|
server = await asyncssh.create_server(
|
|
_Server,
|
|
"127.0.0.1",
|
|
0,
|
|
server_host_keys=[asyncssh.generate_private_key("ssh-ed25519")],
|
|
process_factory=_handler,
|
|
sftp_factory=True,
|
|
)
|
|
port = next(iter(server.sockets)).getsockname()[1]
|
|
line, fingerprint = await ssh_service.capture_host_key("127.0.0.1", port)
|
|
try:
|
|
yield {"port": port, "host_key": line, "fingerprint": fingerprint, "dir": str(project)}
|
|
finally:
|
|
server.close()
|
|
await server.wait_closed()
|
|
|
|
|
|
def _setup(db, user_id, machine, *, mode=policy.MODE_MANUAL, enabled=True, kind=KIND_AGENT):
|
|
"""An agent chat pointed at the machine, with the feature switched on."""
|
|
settings_store.update(db, {"enabled": enabled}, key=settings_store.AGENTS)
|
|
|
|
profile = SshProfile(
|
|
owner_id=user_id,
|
|
name="Test box",
|
|
host="127.0.0.1",
|
|
port=machine["port"],
|
|
username="tester",
|
|
host_key=machine["host_key"],
|
|
host_fingerprint=machine["fingerprint"],
|
|
default_dir=machine["dir"],
|
|
)
|
|
db.add(profile)
|
|
db.commit()
|
|
|
|
connection = Connection(name="c", base_url="http://127.0.0.1:1", api_key_encrypted="")
|
|
db.add(connection)
|
|
db.commit()
|
|
db.add(
|
|
Model(
|
|
connection_id=connection.id,
|
|
model_id="m",
|
|
capabilities_json={"tools": True},
|
|
)
|
|
)
|
|
db.commit()
|
|
|
|
chat = Chat(
|
|
user_id=user_id,
|
|
model_id="m",
|
|
connection_id=connection.id,
|
|
kind=kind,
|
|
ssh_profile_id=profile.id,
|
|
project_dir=machine["dir"],
|
|
agent_mode=mode,
|
|
)
|
|
db.add(chat)
|
|
db.commit()
|
|
return chat, profile
|
|
|
|
|
|
def _offered(db, chat, user) -> set[str]:
|
|
return set(tools_service.resolve_tools(db, chat, user).by_name)
|
|
|
|
|
|
# --- The gate --------------------------------------------------------------------
|
|
async def test_the_agent_tools_are_offered_to_an_agent_chat(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine)
|
|
names = _offered(db, chat, db.get(User, user_id))
|
|
|
|
assert {"shell_run", "file_read", "file_write", "file_list"} <= names
|
|
|
|
|
|
async def test_an_ordinary_chat_gets_none_of_them(db, user_id, machine):
|
|
"""A plain conversation can never shell out, whatever it is asked."""
|
|
chat, _profile = _setup(db, user_id, machine, kind="chat")
|
|
assert "shell_run" not in _offered(db, chat, db.get(User, user_id))
|
|
|
|
|
|
async def test_nothing_is_offered_while_the_feature_is_off(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine, enabled=False)
|
|
assert "shell_run" not in _offered(db, chat, db.get(User, user_id))
|
|
|
|
|
|
async def test_nothing_is_offered_without_the_permission(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine)
|
|
user = db.get(User, user_id)
|
|
user.role = "user" # administrators pass everything
|
|
settings_store.update(db, {"default_permissions": {"tools.agent": False}})
|
|
db.commit()
|
|
|
|
assert "shell_run" not in _offered(db, chat, user)
|
|
|
|
|
|
async def test_nothing_is_offered_without_the_model_capability(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine)
|
|
model = db.scalar(tools_service.select(Model))
|
|
model.capabilities_json = {"tools": True, "tool_agent": False}
|
|
db.commit()
|
|
|
|
assert "shell_run" not in _offered(db, chat, db.get(User, user_id))
|
|
|
|
|
|
async def test_a_disabled_connection_takes_the_tools_away(db, user_id, machine):
|
|
"""Offering a tool that fails on its first call is worse than not offering
|
|
it, so every "no" collapses to an empty list."""
|
|
chat, profile = _setup(db, user_id, machine)
|
|
profile.enabled = False
|
|
db.commit()
|
|
|
|
assert "shell_run" not in _offered(db, chat, db.get(User, user_id))
|
|
|
|
|
|
async def test_somebody_elses_connection_is_not_reachable(db, user_id, machine):
|
|
from lembas.security.passwords import hash_password
|
|
|
|
chat, profile = _setup(db, user_id, machine)
|
|
intruder = User(name="Sam", email="s@shire.test", password_hash=hash_password("x"))
|
|
intruder.role = "user"
|
|
db.add(intruder)
|
|
db.commit()
|
|
|
|
assert session.resolve(db, chat, intruder) is None
|
|
|
|
|
|
# --- Running something -----------------------------------------------------------
|
|
async def test_a_command_runs_on_the_machine(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
outcome = await tools_service.run_tool(
|
|
context, "shell_run", '{"command": "echo hello"}'
|
|
)
|
|
assert outcome.event["status"] == "ok"
|
|
assert "echo hello" in outcome.content
|
|
|
|
|
|
async def test_files_are_written_and_read_back(db, user_id, machine, tmp_path):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
await tools_service.run_tool(
|
|
context, "file_write", '{"path": "note.txt", "content": "a mallorn tree"}'
|
|
)
|
|
assert (tmp_path / "project" / "note.txt").read_text() == "a mallorn tree"
|
|
|
|
read = await tools_service.run_tool(context, "file_read", '{"path": "note.txt"}')
|
|
assert "mallorn" in read.content
|
|
|
|
listed = await tools_service.run_tool(context, "file_list", "{}")
|
|
assert "note.txt" in listed.content
|
|
|
|
|
|
# --- Changing part of a file -------------------------------------------------------
|
|
def _context(db, user_id, machine, **kwargs):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO, **kwargs)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
return tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
|
|
async def test_editing_a_file_that_was_not_read_is_refused(db, user_id, machine, tmp_path):
|
|
"""Both halves matter. The wording is what the model acts on; that nothing
|
|
was written is the actual guarantee."""
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps({"path": "note.txt", "patch": "@@ -1,2 +1,2 @@\n alpha\n-beta\n+BETA\n"}),
|
|
)
|
|
|
|
assert outcome.content.startswith("Read the file first!")
|
|
assert outcome.event["status"] == "error"
|
|
assert (tmp_path / "project" / "note.txt").read_text() == "alpha\nbeta\n"
|
|
|
|
|
|
async def test_reading_then_editing_writes_the_new_text(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\ngamma\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(context, "file_read", '{"path": "note.txt"}')
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps(
|
|
{"path": "note.txt", "patch": "@@ -1,3 +1,3 @@\n alpha\n-beta\n+BETA\n gamma\n"}
|
|
),
|
|
)
|
|
|
|
assert outcome.event["status"] == "ok"
|
|
assert (tmp_path / "project" / "note.txt").read_text() == "alpha\nBETA\ngamma\n"
|
|
|
|
|
|
async def test_a_relative_and_an_absolute_path_are_the_same_file(db, user_id, machine, tmp_path):
|
|
"""`./note.txt` read and `note.txt` edited has to count as having read it,
|
|
or the check refuses the very thing it was meant to permit."""
|
|
target = tmp_path / "project" / "note.txt"
|
|
target.write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(context, "file_read", '{"path": "./note.txt"}')
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps({"path": str(target), "patch": "@@ -1,2 +1,2 @@\n alpha\n-beta\n+BETA\n"}),
|
|
)
|
|
|
|
assert outcome.event["status"] == "ok", outcome.content
|
|
|
|
|
|
async def test_a_failed_hunk_names_it_and_writes_nothing(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(context, "file_read", '{"path": "note.txt"}')
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps({"path": "note.txt", "patch": "@@ -1,2 +1,2 @@\n nope\n-wrong\n+x\n"}),
|
|
)
|
|
|
|
assert outcome.event["status"] == "error"
|
|
assert "Hunk 1" in outcome.content
|
|
assert (tmp_path / "project" / "note.txt").read_text() == "alpha\nbeta\n"
|
|
|
|
|
|
async def test_a_write_counts_as_having_read_it(db, user_id, machine, tmp_path):
|
|
"""`_run_write` reads the old content for its diff anyway, so write-then-edit
|
|
works in one reply without a second round trip."""
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(
|
|
context, "file_write", '{"path": "new.txt", "content": "one\\ntwo\\n"}'
|
|
)
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps({"path": "new.txt", "patch": "@@ -1,2 +1,2 @@\n one\n-two\n+TWO\n"}),
|
|
)
|
|
|
|
assert outcome.event["status"] == "ok", outcome.content
|
|
assert (tmp_path / "project" / "new.txt").read_text() == "one\nTWO\n"
|
|
|
|
|
|
async def test_the_read_set_survives_an_approval(db, user_id, machine, tmp_path):
|
|
"""`as_approved` is `dataclasses.replace`, which copies field *references*,
|
|
so the set is shared with the per-call copy a runner actually gets. That is
|
|
wanted, and it is not obvious enough to leave unpinned."""
|
|
from dataclasses import replace
|
|
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
approved = replace(context, agent=context.agent.as_approved())
|
|
|
|
await tools_service.run_tool(approved, "file_read", '{"path": "note.txt"}')
|
|
|
|
assert context.agent.read_paths, "the read done under approval is not visible"
|
|
|
|
|
|
async def test_an_edit_that_changes_nothing_says_so(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(context, "file_read", '{"path": "note.txt"}')
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps({"path": "note.txt", "patch": "@@ -1,2 +1,2 @@\n alpha\n-beta\n+beta\n"}),
|
|
)
|
|
|
|
assert outcome.event["status"] == "ok"
|
|
assert "changes nothing" in outcome.content
|
|
|
|
|
|
# --- The diff on the event ------------------------------------------------------------
|
|
async def test_an_edit_carries_a_diff(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\ngamma\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
await tools_service.run_tool(context, "file_read", '{"path": "note.txt"}')
|
|
outcome = await tools_service.run_tool(
|
|
context,
|
|
"file_edit",
|
|
_json.dumps(
|
|
{"path": "note.txt", "patch": "@@ -1,3 +1,3 @@\n alpha\n-beta\n+BETA\n gamma\n"}
|
|
),
|
|
)
|
|
|
|
diff = outcome.event["diff"]
|
|
assert "-beta" in diff
|
|
assert "+BETA" in diff
|
|
|
|
|
|
async def test_writing_a_new_file_shows_it_as_all_additions(db, user_id, machine):
|
|
"""Which is what git does, and the right display."""
|
|
context = _context(db, user_id, machine)
|
|
|
|
outcome = await tools_service.run_tool(
|
|
context, "file_write", '{"path": "fresh.txt", "content": "one\\ntwo\\n"}'
|
|
)
|
|
|
|
body = [
|
|
line
|
|
for line in outcome.event["diff"].split("\n")
|
|
if line and not line.startswith(("@@", "+++", "---"))
|
|
]
|
|
assert body and all(line.startswith("+") for line in body), body
|
|
|
|
|
|
async def test_overwriting_a_file_shows_what_changed(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "note.txt").write_text("alpha\nbeta\n")
|
|
context = _context(db, user_id, machine)
|
|
|
|
outcome = await tools_service.run_tool(
|
|
context, "file_write", '{"path": "note.txt", "content": "alpha\\nBETA\\n"}'
|
|
)
|
|
|
|
assert "-beta" in outcome.event["diff"]
|
|
assert "+BETA" in outcome.event["diff"]
|
|
|
|
|
|
async def test_a_file_too_big_to_read_is_written_without_a_diff(db, user_id, machine, tmp_path):
|
|
"""A truncated original would invent deletions of the tail, which is worse
|
|
than showing no diff at all."""
|
|
from lembas.services import settings_store as store
|
|
|
|
store.update(db, {"max_output_bytes": 1024}, key=store.AGENTS)
|
|
(tmp_path / "project" / "big.txt").write_text("x" * 4000)
|
|
context = _context(db, user_id, machine)
|
|
|
|
outcome = await tools_service.run_tool(
|
|
context, "file_write", '{"path": "big.txt", "content": "small"}'
|
|
)
|
|
|
|
assert outcome.event["status"] == "ok"
|
|
assert "diff" not in outcome.event
|
|
|
|
|
|
async def test_writing_a_file_drops_the_project_listing(db, user_id, machine):
|
|
"""Otherwise the model is shown a five-minute-old tree that it knows is
|
|
wrong, and concludes the file it has just created does not exist.
|
|
|
|
The TTL is for drift nobody can see coming. This is not that: it is this
|
|
process changing the tree it has just described.
|
|
"""
|
|
from lembas.services.agent import index as index_service
|
|
|
|
chat, profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
index_service._CACHE[(profile.id, machine["dir"])] = index_service.ProjectIndex(
|
|
paths=("stale.txt",), total=1, source="git", built_at=time.monotonic()
|
|
)
|
|
|
|
await tools_service.run_tool(
|
|
context, "file_write", '{"path": "fresh.txt", "content": "hi"}'
|
|
)
|
|
|
|
assert index_service.cached(profile.id, machine["dir"]) is None
|
|
|
|
|
|
# --- The runner backstop ----------------------------------------------------------
|
|
async def test_a_runner_refuses_what_the_mode_forbids(db, user_id, machine):
|
|
"""`_authorise` is the real gate and runs first. This is the belt to that
|
|
brace: a call arriving by some other path -- a retry, a re-run button --
|
|
must not walk past it."""
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_PLAN)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
outcome = await tools_service.run_tool(context, "shell_run", '{"command": "rm -rf /"}')
|
|
assert outcome.event["status"] == "error"
|
|
assert "not allowed" in outcome.content.lower()
|
|
|
|
|
|
async def test_plan_mode_still_reads(db, user_id, machine, tmp_path):
|
|
(tmp_path / "project" / "readme.txt").write_text("contents")
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_PLAN)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
outcome = await tools_service.run_tool(context, "file_read", '{"path": "readme.txt"}')
|
|
assert outcome.event["status"] == "ok"
|
|
assert "contents" in outcome.content
|
|
|
|
|
|
async def test_edit_mode_writes_but_will_not_run(db, user_id, machine):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_EDIT)
|
|
user = db.get(User, user_id)
|
|
resolved = tools_service.resolve_tools(db, chat, user)
|
|
context = tools_service.context_for(db, user, chat, tools=resolved)
|
|
|
|
wrote = await tools_service.run_tool(
|
|
context, "file_write", '{"path": "x.txt", "content": "hi"}'
|
|
)
|
|
assert wrote.event["status"] == "ok"
|
|
|
|
ran = await tools_service.run_tool(context, "shell_run", '{"command": "ls"}')
|
|
assert ran.event["status"] == "error"
|
|
|
|
|
|
# --- The approval card in the loop -------------------------------------------------
|
|
def _chunk(name: str, arguments: str, *, call_id="c1", index=0):
|
|
return {
|
|
"choices": [
|
|
{
|
|
"delta": {
|
|
"tool_calls": [
|
|
{
|
|
"index": index,
|
|
"id": call_id,
|
|
"function": {"name": name, "arguments": arguments},
|
|
}
|
|
]
|
|
}
|
|
}
|
|
]
|
|
}
|
|
|
|
|
|
def _text(text: str) -> dict:
|
|
return {"choices": [{"delta": {"content": text}}]}
|
|
|
|
|
|
def _stub_stream(rounds, seen):
|
|
async def stream_chat(_endpoint, payload):
|
|
seen.append(payload)
|
|
for chunk in rounds[min(len(seen) - 1, len(rounds) - 1)]:
|
|
yield chunk
|
|
|
|
return stream_chat
|
|
|
|
|
|
async def _until_paused(generation, *, timeout: float = 3.0):
|
|
deadline = asyncio.get_running_loop().time() + timeout
|
|
while asyncio.get_running_loop().time() < deadline:
|
|
if generation.pending is not None:
|
|
return generation.pending
|
|
await asyncio.sleep(0.01)
|
|
raise AssertionError("the reply never paused for approval")
|
|
|
|
|
|
def _pending_reply(db, chat):
|
|
db.add(Message(chat_id=chat.id, role="user", content="do it", complete=True))
|
|
db.commit()
|
|
assistant = Message(chat_id=chat.id, role=ROLE_ASSISTANT, content="", complete=False)
|
|
db.add(assistant)
|
|
db.commit()
|
|
return assistant.id
|
|
|
|
|
|
async def test_a_command_waits_for_approval_and_the_card_shows_it(
|
|
db, user_id, machine, monkeypatch
|
|
):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_MANUAL)
|
|
message_id = _pending_reply(db, chat)
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[[_chunk("shell_run", '{"command": "rm -rf /tmp/x"}')], [_text("Done.")]], []
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
task = asyncio.create_task(generation_service._run(generation))
|
|
pending = await _until_paused(generation)
|
|
|
|
assert pending.kind == interaction.KIND_APPROVAL
|
|
item = pending.items[0]
|
|
# The exact command, verbatim. A card that paraphrased it would be
|
|
# approving something other than what runs.
|
|
assert item.detail == "rm -rf /tmp/x"
|
|
assert "Test box" in item.title
|
|
assert "Manual" in item.reason
|
|
|
|
pending.resolve(interaction.DENY)
|
|
await task
|
|
|
|
|
|
async def test_denying_reaches_the_model_as_words_and_runs_nothing(
|
|
db, user_id, machine, monkeypatch, tmp_path
|
|
):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_MANUAL)
|
|
message_id = _pending_reply(db, chat)
|
|
payloads: list[dict] = []
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[
|
|
[_chunk("file_write", '{"path": "never.txt", "content": "nope"}')],
|
|
[_text("Understood.")],
|
|
],
|
|
payloads,
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
task = asyncio.create_task(generation_service._run(generation))
|
|
pending = await _until_paused(generation)
|
|
pending.resolve(interaction.DENY)
|
|
await task
|
|
|
|
turns = [m for m in payloads[1]["messages"] if m.get("role") == "tool"]
|
|
assert "declined" in turns[0]["content"].lower()
|
|
# And it says not to look for another way round, because a model reads
|
|
# "not allowed" as "not allowed like that" otherwise.
|
|
assert "another way" in turns[0]["content"]
|
|
assert not (tmp_path / "project" / "never.txt").exists()
|
|
|
|
|
|
async def test_allowing_runs_it(db, user_id, machine, monkeypatch, tmp_path):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_MANUAL)
|
|
message_id = _pending_reply(db, chat)
|
|
payloads: list[dict] = []
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[
|
|
[_chunk("file_write", '{"path": "yes.txt", "content": "written"}')],
|
|
[_text("Done.")],
|
|
],
|
|
payloads,
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
task = asyncio.create_task(generation_service._run(generation))
|
|
pending = await _until_paused(generation)
|
|
pending.resolve(interaction.ALLOW)
|
|
await task
|
|
|
|
assert (tmp_path / "project" / "yes.txt").read_text() == "written"
|
|
|
|
|
|
async def test_auto_mode_never_pauses(db, user_id, machine, monkeypatch, tmp_path):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
message_id = _pending_reply(db, chat)
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[
|
|
[_chunk("file_write", '{"path": "auto.txt", "content": "no asking"}')],
|
|
[_text("Done.")],
|
|
],
|
|
[],
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await asyncio.wait_for(generation_service._run(generation), timeout=10)
|
|
|
|
assert generation.pending is None
|
|
assert (tmp_path / "project" / "auto.txt").read_text() == "no asking"
|
|
|
|
|
|
async def test_the_credential_is_cleared_when_the_reply_ends(db, user_id, machine, monkeypatch):
|
|
"""A finished Generation lingers five minutes so late followers get the
|
|
final frames. A private key should not linger with it."""
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
message_id = _pending_reply(db, chat)
|
|
|
|
captured = {}
|
|
original = tools_service.context_for
|
|
|
|
def capture(*args, **kwargs):
|
|
context = original(*args, **kwargs)
|
|
captured["context"] = context
|
|
return context
|
|
|
|
monkeypatch.setattr(tools_service, "context_for", capture)
|
|
monkeypatch.setattr(
|
|
generation_service, "stream_chat", _stub_stream([[_text("Nothing to do.")]], [])
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await generation_service._run(generation)
|
|
|
|
assert captured["context"].agent is not None
|
|
assert captured["context"].agent.spec == {}, "the decrypted credential is dropped"
|
|
|
|
|
|
# --- The harness has to be able to name the machine ---------------------------
|
|
def test_the_registry_maps_the_agent_tools_to_their_family(db):
|
|
"""`harness._families` maps an offered tool *name* back to a family to
|
|
decide which fragments apply, and it has no chat to resolve against. Without
|
|
the agent tools listed here, `shell_run` resolves to no family and an agent
|
|
chat is told nothing about the machine it is working on -- the same omission
|
|
that cost custom tools their guidance once already."""
|
|
book = tools_service.registry(db)
|
|
for name in ("shell_run", "file_read", "file_write", "file_list"):
|
|
assert name in book, name
|
|
assert book[name].family == "agent"
|
|
|
|
|
|
async def test_the_harness_says_where_and_under_what_rules(db, user_id, machine):
|
|
from lembas.services import harness
|
|
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_MANUAL)
|
|
user = db.get(User, user_id)
|
|
offered = tools_service.resolve_tools(db, chat, user).schemas
|
|
text = harness.compose(db, user, offered, chat)
|
|
|
|
assert "Test box" in text, "which machine"
|
|
assert machine["dir"] in text, "which directory"
|
|
assert "Manual" in text, "what the mode permits"
|
|
# The single most likely cause of "the agent seems stupid": `cd build`
|
|
# followed by `make` fails silently otherwise.
|
|
assert "fresh shell" in text
|
|
|
|
|
|
async def test_an_ordinary_chat_is_told_none_of_it(db, user_id, machine):
|
|
from lembas.services import harness
|
|
|
|
chat, _profile = _setup(db, user_id, machine, kind="chat")
|
|
user = db.get(User, user_id)
|
|
offered = tools_service.resolve_tools(db, chat, user).schemas
|
|
text = harness.compose(db, user, offered, chat)
|
|
|
|
assert "Test box" not in text
|
|
assert "fresh shell" not in text
|
|
|
|
|
|
async def test_an_agent_chat_is_told_its_real_round_budget(db, user_id, machine):
|
|
"""MAX_ROUNDS is three. An agent chat gets forty, and telling it three would
|
|
be a false fact about its own budget on every turn."""
|
|
from lembas.services import harness
|
|
|
|
settings_store.update(db, {"max_steps": 25}, key=settings_store.AGENTS)
|
|
chat, _profile = _setup(db, user_id, machine)
|
|
user = db.get(User, user_id)
|
|
values = harness.context_variables(
|
|
db, user, tools_service.resolve_tools(db, chat, user).schemas, chat
|
|
)
|
|
assert values["max_rounds"] == "25"
|
|
|
|
|
|
async def test_an_agent_chat_gets_the_rounds_it_was_promised(db, user_id, machine, monkeypatch):
|
|
"""And is then allowed to use them, which is the half that was missing.
|
|
|
|
The budget sizes the loop and names itself in the out-of-rounds message, and
|
|
the harness above tells the model the same number. But the comparison that
|
|
ends the loop read the global `MAX_ROUNDS` of three. So an agent chat
|
|
allowed forty rounds stopped after three and reported that it had taken
|
|
forty: two wrong answers to "why did it stop", with no way to tell them
|
|
apart from the outside.
|
|
"""
|
|
settings_store.update(db, {"max_steps": 5}, key=settings_store.AGENTS)
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
assistant = Message(chat_id=chat.id, role=ROLE_ASSISTANT, content="", complete=False)
|
|
db.add(assistant)
|
|
db.commit()
|
|
|
|
payloads: list[dict] = []
|
|
|
|
async def stream_chat(_endpoint, payload):
|
|
payloads.append(payload)
|
|
yield {
|
|
"choices": [
|
|
{"delta": {"tool_calls": [{"index": 0, "id": "c1", "function": {
|
|
"name": "file_list", "arguments": '{"path": "."}'}}]}}
|
|
]
|
|
}
|
|
|
|
monkeypatch.setattr(generation_service, "stream_chat", stream_chat)
|
|
|
|
async def _no_title(*_args, **_kwargs):
|
|
return ""
|
|
|
|
monkeypatch.setattr("lembas.services.chat.generate_title", _no_title)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=assistant.id)
|
|
await generation_service._run(generation)
|
|
|
|
# Five rounds that may call tools, then the one that gives up.
|
|
assert len(payloads) == 6
|
|
assert "after 5 rounds" in generation.tool_events[-1]["error"]
|
|
|
|
|
|
# --- Interjecting while it works --------------------------------------------------
|
|
async def test_a_queued_prompt_is_taken_in_between_rounds(db, user_id, machine, monkeypatch):
|
|
"""The point of queueing in an agent chat: steering work already under way.
|
|
|
|
An agent that has just finished one loop and is about to start another is
|
|
exactly when "actually, do it the other way" is worth having, and making it
|
|
wait for the whole reply would mean it arrives after the thing it was meant
|
|
to change.
|
|
"""
|
|
from lembas.services import chat as chat_service
|
|
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
message_id = _pending_reply(db, chat)
|
|
chat_service.create_message(
|
|
db, chat, "user", "actually, check the other directory first", queued=True
|
|
)
|
|
|
|
payloads: list[dict] = []
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[[_chunk("file_list", '{"path": "."}')], [_text("Done.")]],
|
|
payloads,
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await generation_service._run(generation)
|
|
|
|
# Verbatim, in the user role, with nothing wrapped around it: this genuinely
|
|
# is the person at the keyboard, and quoting it would teach the model that a
|
|
# user turn can be a quotation -- the distinction `execute_plan` relies on.
|
|
assert payloads[1]["messages"][-1] == {
|
|
"role": "user",
|
|
"content": "actually, check the other directory first",
|
|
}
|
|
|
|
|
|
async def test_an_interjection_is_delivered_only_once(db, user_id, machine, monkeypatch):
|
|
"""Marked delivered before the request goes out, so a crash loses it rather
|
|
than asking the same thing twice and letting an agent act on it twice."""
|
|
from lembas.services import chat as chat_service
|
|
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
message_id = _pending_reply(db, chat)
|
|
waiting = chat_service.create_message(db, chat, "user", "one more thing", queued=True)
|
|
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[
|
|
[_chunk("file_list", '{"path": "."}')],
|
|
[_chunk("file_list", '{"path": "src"}')],
|
|
[_text("Done.")],
|
|
],
|
|
[],
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await generation_service._run(generation)
|
|
|
|
db.expire_all()
|
|
assert db.get(Message, waiting.id).queued is False
|
|
assert generation.injected_ids == [waiting.id]
|
|
|
|
|
|
async def test_the_reply_sorts_before_the_prompt_it_took_in(db, user_id, machine, monkeypatch):
|
|
"""Otherwise the next request reads "answer, then the question it answered",
|
|
and a small model dutifully answers it a second time."""
|
|
from lembas.services import chat as chat_service
|
|
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_AUTO)
|
|
message_id = _pending_reply(db, chat)
|
|
waiting = chat_service.create_message(db, chat, "user", "and this", queued=True)
|
|
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream([[_chunk("file_list", '{"path": "."}')], [_text("Done.")]], []),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await generation_service._run(generation)
|
|
|
|
db.expire_all()
|
|
reply = db.get(Message, message_id)
|
|
assert reply.created_at > db.get(Message, waiting.id).created_at
|
|
|
|
|
|
# --- Plan mode's artifact ---------------------------------------------------------
|
|
def test_plan_submit_is_offered_only_in_plan_mode(db, user_id, machine):
|
|
"""It ends the reply. A model in Auto mode that proposed a plan instead of
|
|
doing the work would be obeying the wrong instinct at the wrong moment."""
|
|
user = db.get(User, user_id)
|
|
planning, _p = _setup(db, user_id, machine, mode=policy.MODE_PLAN)
|
|
assert "plan_submit" in _offered(db, planning, user)
|
|
|
|
planning.agent_mode = policy.MODE_AUTO
|
|
db.commit()
|
|
assert "plan_submit" not in _offered(db, planning, user)
|
|
|
|
|
|
async def test_a_plan_ends_the_reply_and_lands_on_the_message(
|
|
db, user_id, machine, monkeypatch
|
|
):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_PLAN)
|
|
message_id = _pending_reply(db, chat)
|
|
|
|
plan = {"title": "Tidy the logs", "steps": ["Read the log", "Rotate it", "Restart"]}
|
|
rounds = [
|
|
[_chunk("plan_submit", _json.dumps(plan))],
|
|
# The model gets one wordless round to explain itself. If it tries to
|
|
# act in it -- as this one does -- there are no tools to act with.
|
|
[_chunk("shell_run", '{"command": "rm -rf /"}'), _text("Here is what I would do.")],
|
|
]
|
|
seen: list[dict] = []
|
|
monkeypatch.setattr(generation_service, "stream_chat", _stub_stream(rounds, seen))
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await asyncio.wait_for(generation_service._run(generation), timeout=10)
|
|
|
|
assert generation.plan == plan
|
|
assert len(seen) == 2, "one round to plan, one to say what it proposed"
|
|
assert "tools" not in seen[1], "the second round is offered nothing to act with"
|
|
assert len(generation.tool_events) == 1, "the shell call had no tool to reach"
|
|
|
|
db.expire_all()
|
|
message = db.get(Message, message_id)
|
|
assert message.plan_json == plan
|
|
|
|
|
|
async def test_a_plan_with_no_steps_is_sent_back(db, user_id, machine, monkeypatch):
|
|
chat, _profile = _setup(db, user_id, machine, mode=policy.MODE_PLAN)
|
|
message_id = _pending_reply(db, chat)
|
|
monkeypatch.setattr(
|
|
generation_service,
|
|
"stream_chat",
|
|
_stub_stream(
|
|
[[_chunk("plan_submit", '{"title": "Nothing", "steps": []}')], [_text("Sorry.")]],
|
|
[],
|
|
),
|
|
)
|
|
|
|
generation = generation_service.Generation(chat_id=chat.id, message_id=message_id)
|
|
await asyncio.wait_for(generation_service._run(generation), timeout=10)
|
|
|
|
assert generation.plan is None
|
|
assert generation.tool_events[0]["status"] == "error"
|
|
|
|
|
|
# --- Carrying a plan out ------------------------------------------------------------
|
|
def test_executing_a_plan_switches_to_edit_and_quotes_it(client, db, registered, machine):
|
|
"""Edit, never Auto. The plan was written under a mode where every command
|
|
stopped for approval, and a button that also removed the asking is not the
|
|
button anybody pressed."""
|
|
from sqlalchemy import select as _select
|
|
|
|
user = db.scalar(_select(User))
|
|
chat, _profile = _setup(db, user.id, machine, mode=policy.MODE_PLAN)
|
|
message = Message(
|
|
chat_id=chat.id,
|
|
role=ROLE_ASSISTANT,
|
|
content="Here is what I would do.",
|
|
complete=True,
|
|
plan_json={"title": "Tidy the logs", "steps": ["Read the log", "Rotate it"]},
|
|
)
|
|
db.add(message)
|
|
db.commit()
|
|
|
|
response = client.post(f"/api/chats/{chat.id}/messages/{message.id}/execute-plan")
|
|
assert response.status_code == 200
|
|
|
|
db.refresh(chat)
|
|
assert chat.agent_mode == policy.MODE_EDIT
|
|
|
|
sent = db.scalars(
|
|
_select(Message).where(Message.chat_id == chat.id, Message.role == "user")
|
|
).all()[-1]
|
|
assert "Tidy the logs" in sent.content
|
|
assert "Read the log" in sent.content
|
|
# Quoted rather than stated. A plan whose text came out of a file the model
|
|
# read must not arrive wearing the reader's authority.
|
|
assert sent.content.lstrip().startswith("Carry out the plan you proposed above")
|
|
assert "> " in sent.content
|
|
|
|
|
|
def test_a_message_with_no_plan_cannot_be_executed(client, db, registered, machine):
|
|
from sqlalchemy import select as _select
|
|
|
|
user = db.scalar(_select(User))
|
|
chat, _profile = _setup(db, user.id, machine)
|
|
message = Message(chat_id=chat.id, role=ROLE_ASSISTANT, content="hi", complete=True)
|
|
db.add(message)
|
|
db.commit()
|
|
|
|
assert (
|
|
client.post(f"/api/chats/{chat.id}/messages/{message.id}/execute-plan").status_code
|
|
== 404
|
|
)
|
|
|
|
|
|
# --- Rewind ---------------------------------------------------------------------------
|
|
def test_editing_a_turn_records_that_the_machine_did_not_rewind(
|
|
client, db, registered, machine
|
|
):
|
|
"""The transcript goes back; the project directory does not. Deleting
|
|
somebody's real working tree to match would be far worse than the
|
|
inconsistency, so the model is told instead."""
|
|
from sqlalchemy import select as _select
|
|
|
|
user = db.scalar(_select(User))
|
|
chat, _profile = _setup(db, user.id, machine)
|
|
first = Message(chat_id=chat.id, role="user", content="do a thing", complete=True)
|
|
db.add(first)
|
|
db.commit()
|
|
|
|
assert chat.rewound_at is None
|
|
client.post(
|
|
f"/api/chats/{chat.id}/messages/{first.id}/edit", data={"content": "do another thing"}
|
|
)
|
|
db.refresh(chat)
|
|
assert chat.rewound_at is not None
|
|
|
|
|
|
def test_an_ordinary_chat_records_no_rewind(client, db, registered, machine):
|
|
from sqlalchemy import select as _select
|
|
|
|
user = db.scalar(_select(User))
|
|
chat, _profile = _setup(db, user.id, machine, kind="chat")
|
|
first = Message(chat_id=chat.id, role="user", content="hello", complete=True)
|
|
db.add(first)
|
|
db.commit()
|
|
|
|
client.post(f"/api/chats/{chat.id}/messages/{first.id}/edit", data={"content": "hi"})
|
|
db.refresh(chat)
|
|
assert chat.rewound_at is None
|
|
|
|
|
|
async def test_the_harness_warns_after_a_rewind(db, user_id, machine):
|
|
from datetime import UTC, datetime
|
|
|
|
from lembas.services import harness
|
|
|
|
chat, _profile = _setup(db, user_id, machine)
|
|
user = db.get(User, user_id)
|
|
offered = tools_service.resolve_tools(db, chat, user).schemas
|
|
|
|
assert "was rewound" not in harness.compose(db, user, offered, chat)
|
|
|
|
chat.rewound_at = datetime.now(UTC)
|
|
db.commit()
|
|
text = harness.compose(db, user, offered, chat)
|
|
assert "was rewound" in text
|
|
assert "still there" in text
|