## Summary
Reported as "deliver_files is broken". It is not: **the tool never ran.**
Every `deliver_files` call on one install was refused before dispatch -- five in a row in a single session -- and nothing was ever delivered:
```
Error: Invalid parameters for tool 'deliver_files': files should be array
```
What the model sent was the array, spelled as a string:
```json
{"files": "[{\"path\": \"...poster.png\", \"title\": \"...\"}]", "message": "..."}
```
`cast_params` already forgives that spelling for every scalar type -- `"3"` for an integer, `"true"` for a boolean -- and did not for arrays or objects, which are exactly the types a model is most likely to stringify.
## Type
- [x] Fix
- [ ] Feature
## The edge of the forgiving is the design
When the value is a string, try `json.loads`, and take the result **only if it is the type the schema asked for**. A string that is not JSON, or is JSON of the wrong shape, is returned unchanged so the validator still names the real problem -- the point is to widen what is accepted, not to swallow a type error.
**Nothing here can raise.** Turning "your argument has the wrong type" into a traceback would be worse than the refusal it replaces, so the parse is wrapped and failure falls through.
**A `type: "string"` field is never parsed.** JSON is ordinary text in a message body or a file's content, and replacing it would hand the tool something it never asked for. Two independent things prevent it today (the string branch returns before the parse is reached, and the parse only wins on a type match), which is why the case pinning it is a tripwire rather than a load-bearing assertion -- said so in its own docstring rather than left to look like coverage.
## Not just deliver_files
Six built-in tools declare an array or object parameter and were all exposed:
| tool | parameter |
|---|---|
| `deliver_files` | `files` |
| `ask_user` | `questions` |
| `image_generate` | `images` |
| `message` | `media` |
| `create_playbook` | `skills` |
| `video_generate` | `params` |
The same failure was already seen from the other side: `acp_client/acp_agent.py` records `update_task_state` and `operations should be array` from a real dispatch.
## Verification
```
make lint-python ruff check + ruff format --check, 1850 files
make lint-imports 9 contracts kept, 0 broken
pytest tests/ 19,699 passed, 7 failed (all pre-existing, see below)
```
- [x] Relevant tests pass locally
- [x] Relevant lint / type checks pass locally
- [ ] User-facing docs or screenshots -- no user-facing string
**End to end, through the real tool**, with the exact payload that failed five times in production:
```
files sent as: str -> [{"path": ".../poster.png", "title": "..."}]
returned: Delivered 1 file(s): poster.png (72 B).
still "should be array": False
registered in the store: True
```
Twelve cases, five mutations, each killing at least one:
| mutation | red |
|---|---|
| drop the whole branch (the old behaviour) | 3 |
| raise instead of falling through on a parse failure | 5 |
| accept the parse whatever type it yielded | 1 |
| do not re-cast the parsed value, so items are not cast | 1 |
| widen the tuple to include `string` | **survived -- unreachable** |
The last is worth stating rather than hiding: that mutation cannot change behaviour, because the `type == "string"` branch returns earlier and the type check would refuse the result anyway. I also tried moving the new block above that branch, alone and combined with widening the tuple, and neither reddens anything either. The invariant holds by construction; the case that documents it says so.
**Pre-existing failures, checked against the base.** 7 failed in the full run, all in `tests/test_security_network.py` (`test_a_redirect_to_a_dangerous_address_is_refused` and siblings). Running that file alone on this same tree gives 4 -- the count is order-dependent -- and the change touches only `raven/agent/tools/params.py` plus a new test file, neither of which `raven/security/network.py` imports. The failures are the box's resolver: it returns poisoned answers roughly half the time, which I measured separately (`en.wikipedia.org` came back as `2001::1` on 3 of 6 lookups). Not introduced here, and worth someone's attention on its own.
## Risk
- [x] Security impact considered -- widens what reaches a tool's own validation, not what bypasses it. A parsed value is still validated against the same schema, and a parse that yields the wrong type is refused exactly as before
- [x] Backward compatibility considered -- a call that worked before takes the identical path; the new branch is reached only by a `str` where an array or object is declared, which previously always failed
- [x] Rollback path is clear -- single-commit revert
## Related Issues
N/A
## Swept
`mr-review-patterns` pre-submit sweep, over `git diff origin/main...HEAD`.
- **1 (a write location moved)** -- nothing moved; a branch was added to an existing dispatch on `target_type`. Grepped every caller of `cast_params`: the registry is the only one, and it runs `validate_params` right after, which is the reader this affects.
- **2 (a value with several entry points)** -- the cast is the single funnel every tool argument passes through, which is why the fix belongs here and not in `deliver_files`.
- **3 (a guard)** -- checked both directions: what it wrongly lets through (JSON of the wrong type, non-JSON text -- both refused, with cases) and what it wrongly refuses (a real list, unaffected, with a case).
- **4 (tests)** -- five mutations above, one honestly reported as unreachable. Failing-ID sets diffed against the base rather than counts.
- **6 (stale text)** -- the module docstring enumerated what is coerced ("a string that should be an integer"); it now names arrays and objects too. Grepped the tracked tree for other statements of the rule: none.
- **9 (error handling)** -- the new path adds one way to end and it is caught: `json.loads` raises `ValueError` on bad text and `TypeError` on a non-string, and both fall through to the untouched value.
- **11** -- `git ls-files | grep -c node_modules` is 0; the commit-message gate was run, not eyeballed.
Not applicable, with reasons: **3b, 3c, 3d, 3e** (no tool schema produced, no state moved, one lane, no assertion lost its symbol); **4b, 4c, 4d** (no platform branch, no shell, no rendered number); **5, 5b** (no dependency; `json` is stdlib and already imported across this package); **6b, 6c** (no backup, no generated artifact); **7, 7b** (nothing on disk); **8** (no asset, no secret); **10** (no prompt text).
## Deliberately not fixed
The error text still ends with `[Analyze the error above and try a different approach.]`, and the model plainly did not -- it sent the same shape five times. A message naming the expected spelling (`files must be a JSON array, not a string containing one`) would give it something to act on. That is a change to the registry's error surface, affects every tool, and deserves its own argument.
## Moved from !594
Opened on a compliant source branch. AGENTS.md 2.1 asks for a 3-5 word snake_case description and the previous branch had seven; GitLab cannot change an existing merge request's source branch, so the change moves rather than the name.
Same commits, rebased onto `c2a9768f`. Both code findings from the review on !594 are already in them -- `RecursionError` caught with its regression case, and the implementation comment reduced to the two non-obvious reasons. Nothing else changed, so that review carries.
271 lines
8.7 KiB
Python
271 lines
8.7 KiB
Python
"""Tests for AgentLoop's response_modifier hook.
|
|
|
|
The hook is a generic content transform applied to the final assistant
|
|
content just before it's sent as the reply. Sentinel's NudgeInjector
|
|
is the intended primary user, but the hook itself is agnostic.
|
|
|
|
Contract this file pins:
|
|
- modifier=None → content passes through unchanged (regression safety)
|
|
- modifier set → modifier is called with (session_key, content) and its return
|
|
value replaces the content
|
|
- a SENTINEL-origin turn skips the modifier (anti-cascade)
|
|
- modifier exception does not crash the loop (logged, original content preserved)
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import tempfile
|
|
from pathlib import Path
|
|
from typing import Any, Callable
|
|
|
|
import pytest
|
|
|
|
from raven.agent.hook.adapters import OnUserInboundAdapter, ResponseModifierAdapter
|
|
from raven.agent.loop import AgentLoop
|
|
from raven.agent.loop.bundles import HostWiring, ToolWiring, TurnPolicy
|
|
from raven.providers.base import LLMProvider, LLMResponse
|
|
from raven.spine.message import ChatType, Source
|
|
from raven.spine.turn import Origin, TurnRequest
|
|
|
|
|
|
class StubProvider(LLMProvider):
|
|
"""Always returns a fixed assistant message. No tool calls."""
|
|
|
|
def __init__(self, content: str = "stub response"):
|
|
super().__init__(api_key="test")
|
|
self._content = content
|
|
|
|
async def chat(
|
|
self,
|
|
messages,
|
|
tools=None,
|
|
model=None,
|
|
max_tokens=4096,
|
|
temperature=0.7,
|
|
reasoning_effort=None,
|
|
tool_choice=None,
|
|
):
|
|
return LLMResponse(content=self._content, finish_reason="stop")
|
|
|
|
def get_default_model(self) -> str:
|
|
return "stub"
|
|
|
|
|
|
@pytest.fixture
|
|
def workspace():
|
|
with tempfile.TemporaryDirectory() as td:
|
|
yield Path(td)
|
|
|
|
|
|
def _make_agent(
|
|
workspace: Path,
|
|
modifier: Callable[[str, str], str] | None = None,
|
|
provider: LLMProvider | None = None,
|
|
) -> AgentLoop:
|
|
return AgentLoop(
|
|
provider=provider or StubProvider(),
|
|
workspace=workspace,
|
|
model="stub",
|
|
policy=TurnPolicy(max_iterations=2),
|
|
host=HostWiring(hooks=[ResponseModifierAdapter(modifier)]),
|
|
tools=ToolWiring(restrict_to_workspace=True),
|
|
)
|
|
|
|
|
|
def _make_msg(content: str = "hello", metadata: dict[str, Any] | None = None) -> TurnRequest:
|
|
return TurnRequest(
|
|
origin=Origin.USER,
|
|
source=Source(
|
|
channel="test",
|
|
chat_id="chat1",
|
|
sender_id="user",
|
|
chat_type=ChatType.DM,
|
|
extras=metadata or {},
|
|
),
|
|
text=content,
|
|
)
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_no_modifier_content_passes_through(workspace):
|
|
agent = _make_agent(workspace, modifier=None)
|
|
out = await agent._process_message(_make_msg())
|
|
assert out is not None
|
|
assert out[0] == "stub response"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_modifier_transforms_content(workspace):
|
|
calls: list[tuple[str, str]] = []
|
|
|
|
def mod(session_key: str, content: str) -> str:
|
|
calls.append((session_key, content))
|
|
return content + " [MODIFIED]"
|
|
|
|
agent = _make_agent(workspace, modifier=mod)
|
|
out = await agent._process_message(_make_msg())
|
|
assert out is not None
|
|
assert out[0] == "stub response [MODIFIED]"
|
|
assert len(calls) == 1
|
|
assert calls[0][0] == "test:chat1"
|
|
assert calls[0][1] == "stub response"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_modifier_skipped_for_sentinel_origin(workspace):
|
|
# The after_send gate reads origin. A SENTINEL-origin turn (e.g. the
|
|
# supersede notice) skips the modifier.
|
|
from raven.spine import Origin
|
|
|
|
calls: list[tuple[str, str]] = []
|
|
|
|
def mod(session_key: str, content: str) -> str:
|
|
calls.append((session_key, content))
|
|
return content + " [MODIFIED]"
|
|
|
|
agent = _make_agent(workspace, modifier=mod)
|
|
out = await agent._process_message(_make_msg(), origin=Origin.SENTINEL)
|
|
assert out is not None and out[0] == "stub response"
|
|
assert calls == []
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_modifier_runs_for_user_origin(workspace):
|
|
# A menu pick is origin=USER (its reply IS the user's intent) — the modifier
|
|
# runs.
|
|
from raven.spine import Origin
|
|
|
|
calls: list[tuple[str, str]] = []
|
|
|
|
def mod(session_key: str, content: str) -> str:
|
|
calls.append((session_key, content))
|
|
return content + " [MODIFIED]"
|
|
|
|
agent = _make_agent(workspace, modifier=mod)
|
|
out = await agent._process_message(_make_msg(), origin=Origin.USER)
|
|
assert out is not None and out[0] == "stub response [MODIFIED]"
|
|
assert len(calls) == 1
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_modifier_skipped_for_subagent_origin(workspace):
|
|
# parity: a subagent result re-injection returned from the
|
|
# system-message branch before reaching after_send today, so it never ran
|
|
# the modifier. On the spine path (origin=SUBAGENT, non-system) it must still
|
|
# skip — else the summary gets a nudge layered on (regression).
|
|
from raven.spine import Origin
|
|
|
|
calls: list[tuple[str, str]] = []
|
|
|
|
def mod(session_key: str, content: str) -> str:
|
|
calls.append((session_key, content))
|
|
return content + " [MODIFIED]"
|
|
|
|
agent = _make_agent(workspace, modifier=mod)
|
|
out = await agent._process_message(_make_msg(), origin=Origin.SUBAGENT)
|
|
assert out is not None and out[0] == "stub response"
|
|
assert calls == []
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_personalization_skipped_for_subagent_only(workspace, monkeypatch):
|
|
# parity (asymmetry lock): a subagent result re-injection must NOT be
|
|
# personalized (the system-message branch never ran personalization, and the
|
|
# announce is system-generated, not user input). The skip is SUBAGENT-only —
|
|
# a SENTINEL turn reaches and runs personalization today, so it must keep it.
|
|
import raven.agent.personalizer as personalizer_mod
|
|
from raven.spine import Origin
|
|
|
|
classified: list[str] = []
|
|
|
|
class _SpyPersonalizer:
|
|
def __init__(self, *a, **k):
|
|
pass
|
|
|
|
async def classify(self, content, history=None):
|
|
classified.append(content)
|
|
return {"needs_clarification": False}
|
|
|
|
async def post_learn(self, *a, **k):
|
|
pass
|
|
|
|
monkeypatch.setattr(personalizer_mod, "Personalizer", _SpyPersonalizer)
|
|
|
|
agent = _make_agent(workspace)
|
|
agent.configure_personalization(True)
|
|
|
|
await agent._process_message(_make_msg("subagent announce"), origin=Origin.SUBAGENT)
|
|
assert classified == [] # SUBAGENT skips personalization
|
|
|
|
await agent._process_message(_make_msg("sentinel notice"), origin=Origin.SENTINEL)
|
|
assert classified == ["sentinel notice"] # SENTINEL still personalized
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_modifier_exception_preserves_content(workspace):
|
|
def bad_mod(session_key: str, content: str) -> str:
|
|
raise RuntimeError("boom")
|
|
|
|
agent = _make_agent(workspace, modifier=bad_mod)
|
|
out = await agent._process_message(_make_msg())
|
|
# Exception logged but original content survives.
|
|
assert out is not None
|
|
assert out[0] == "stub response"
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# on_user_inbound hook — Sentinel engagement tracking entry point
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_on_user_inbound_called_for_user_message(workspace):
|
|
received = []
|
|
|
|
agent = AgentLoop(
|
|
provider=StubProvider(),
|
|
workspace=workspace,
|
|
model="stub",
|
|
policy=TurnPolicy(max_iterations=2),
|
|
host=HostWiring(hooks=[OnUserInboundAdapter(lambda msg: received.append(msg))]),
|
|
tools=ToolWiring(restrict_to_workspace=True),
|
|
)
|
|
await agent._process_message(_make_msg("hello"))
|
|
assert len(received) == 1
|
|
assert received[0].text == "hello"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_on_user_inbound_skipped_for_sentinel_origin(workspace):
|
|
from raven.spine import Origin
|
|
|
|
received = []
|
|
|
|
agent = AgentLoop(
|
|
provider=StubProvider(),
|
|
workspace=workspace,
|
|
model="stub",
|
|
policy=TurnPolicy(max_iterations=2),
|
|
host=HostWiring(hooks=[OnUserInboundAdapter(lambda msg: received.append(msg))]),
|
|
tools=ToolWiring(restrict_to_workspace=True),
|
|
)
|
|
await agent._process_message(_make_msg("sentinel-originated"), origin=Origin.SENTINEL)
|
|
assert received == []
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_on_user_inbound_exception_does_not_crash(workspace):
|
|
def bad_cb(msg):
|
|
raise RuntimeError("boom")
|
|
|
|
agent = AgentLoop(
|
|
provider=StubProvider(),
|
|
workspace=workspace,
|
|
model="stub",
|
|
policy=TurnPolicy(max_iterations=2),
|
|
host=HostWiring(hooks=[OnUserInboundAdapter(bad_cb)]),
|
|
tools=ToolWiring(restrict_to_workspace=True),
|
|
)
|
|
out = await agent._process_message(_make_msg())
|
|
assert out is not None
|
|
assert out[0] == "stub response"
|