## 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.
198 lines
7.4 KiB
Python
198 lines
7.4 KiB
Python
"""Unit tests for RoutineValidator (Sentinel Stage 1)."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
from raven.contracts.llm_provider import LLMResponse, ToolCallRequest
|
|
from raven.proactive_engine.sentinel.predictor.routine_validator import (
|
|
VALIDATOR_TOOL,
|
|
RoutineValidator,
|
|
)
|
|
from raven.proactive_engine.sentinel.types import Routine
|
|
|
|
# ── stubs ─────────────────────────────────────────────────────────────
|
|
|
|
|
|
class _StubProvider:
|
|
"""Minimal LLMProvider-shaped stub. Records call history; replays a
|
|
queued response (or raises)."""
|
|
|
|
def __init__(self, *, response: LLMResponse | None = None, exc: Exception | None = None):
|
|
self._response = response
|
|
self._exc = exc
|
|
self.calls: list[dict] = []
|
|
|
|
async def chat_with_retry(self, **kwargs): # noqa: ANN003
|
|
self.calls.append(kwargs)
|
|
if self._exc is not None:
|
|
raise self._exc
|
|
assert self._response is not None
|
|
return self._response
|
|
|
|
|
|
def _ok_response(*, is_routine: bool, confidence: float, reason: str = "ok") -> LLMResponse:
|
|
return LLMResponse(
|
|
content=None,
|
|
tool_calls=[
|
|
ToolCallRequest(
|
|
id="tc1",
|
|
name="routine_validation",
|
|
arguments={
|
|
"is_routine": is_routine,
|
|
"confidence": confidence,
|
|
"reason": reason,
|
|
},
|
|
)
|
|
],
|
|
finish_reason="stop",
|
|
)
|
|
|
|
|
|
def _routine(**kw) -> Routine:
|
|
defaults = dict(
|
|
id="dow1-h09-meeting",
|
|
pattern="Tuesday 09:00-12:00 — meeting · standup",
|
|
keywords=["meeting", "standup"],
|
|
day_of_week=1,
|
|
time_slot=(9, 12),
|
|
occurrence_count=4,
|
|
)
|
|
defaults.update(kw)
|
|
return Routine(**defaults)
|
|
|
|
|
|
_HISTORY = "\n".join(f"[2026-04-{d:02d} 09:30] standup meeting notes" for d in (7, 14, 21, 28))
|
|
_NOW = 1_700_000_000_000
|
|
|
|
|
|
# ── tests ─────────────────────────────────────────────────────────────
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_happy_path_returns_validation():
|
|
provider = _StubProvider(response=_ok_response(is_routine=True, confidence=0.85, reason="4 Tuesdays"))
|
|
v = RoutineValidator(provider, model="test-model")
|
|
out = await v.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out is not None
|
|
assert out.is_routine is True
|
|
assert out.confidence == 0.85
|
|
assert "Tuesday" in out.reason or "4" in out.reason
|
|
assert out.validated_at_ms == _NOW
|
|
# Check provider was called with our tool schema, not someone else's
|
|
assert provider.calls[0]["tools"] == [VALIDATOR_TOOL]
|
|
assert provider.calls[0]["model"] == "test-model"
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_no_tool_call_returns_none():
|
|
"""LLM returned text content instead of calling the tool — must fail safely."""
|
|
provider = _StubProvider(response=LLMResponse(content="just chatter", tool_calls=[]))
|
|
v = RoutineValidator(provider, model="m")
|
|
assert await v.validate(_routine(), _HISTORY, now_ms=_NOW) is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_provider_raises_returns_none():
|
|
"""LLM call exception (network, timeout, etc.) must not propagate."""
|
|
provider = _StubProvider(exc=RuntimeError("connection reset"))
|
|
v = RoutineValidator(provider, model="m")
|
|
assert await v.validate(_routine(), _HISTORY, now_ms=_NOW) is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_finish_reason_error_returns_none():
|
|
provider = _StubProvider(response=LLMResponse(content="upstream 500", finish_reason="error"))
|
|
v = RoutineValidator(provider, model="m")
|
|
assert await v.validate(_routine(), _HISTORY, now_ms=_NOW) is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_malformed_args_returns_none():
|
|
"""Tool call with missing required field must return None, not raise."""
|
|
bad = LLMResponse(
|
|
content=None,
|
|
tool_calls=[ToolCallRequest(id="x", name="routine_validation", arguments={"confidence": 0.7})],
|
|
finish_reason="stop",
|
|
)
|
|
provider = _StubProvider(response=bad)
|
|
v = RoutineValidator(provider, model="m")
|
|
assert await v.validate(_routine(), _HISTORY, now_ms=_NOW) is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_args_not_dict_returns_none():
|
|
bad = LLMResponse(
|
|
content=None,
|
|
tool_calls=[ToolCallRequest(id="x", name="routine_validation", arguments=[1, 2, 3])], # type: ignore[arg-type]
|
|
finish_reason="stop",
|
|
)
|
|
provider = _StubProvider(response=bad)
|
|
v = RoutineValidator(provider, model="m")
|
|
assert await v.validate(_routine(), _HISTORY, now_ms=_NOW) is None
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_confidence_clamped_to_unit_range():
|
|
"""LLMs occasionally return out-of-range confidence; clamp instead of reject."""
|
|
provider = _StubProvider(response=_ok_response(is_routine=True, confidence=1.7))
|
|
v = RoutineValidator(provider, model="m")
|
|
out = await v.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out is not None
|
|
assert out.confidence == 1.0
|
|
|
|
provider2 = _StubProvider(response=_ok_response(is_routine=False, confidence=-0.3))
|
|
v2 = RoutineValidator(provider2, model="m")
|
|
out2 = await v2.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out2 is not None
|
|
assert out2.confidence == 0.0
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_truncates_long_history():
|
|
"""Long history should be tail-truncated before going to the LLM."""
|
|
long_history = "x" * 50_000 # well beyond _HISTORY_MAX_CHARS
|
|
provider = _StubProvider(response=_ok_response(is_routine=False, confidence=0.1))
|
|
v = RoutineValidator(provider, model="m")
|
|
await v.validate(_routine(), long_history, now_ms=_NOW)
|
|
user_content = provider.calls[0]["messages"][1]["content"]
|
|
# The injected history should be much shorter than the input
|
|
assert len(user_content) < 12_000
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_reason_truncated_to_300_chars():
|
|
long_reason = "因为" * 500
|
|
provider = _StubProvider(response=_ok_response(is_routine=True, confidence=0.9, reason=long_reason))
|
|
v = RoutineValidator(provider, model="m")
|
|
out = await v.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out is not None
|
|
assert len(out.reason) <= 300
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_nan_confidence_normalized_to_zero():
|
|
"""LLM returning NaN would slip through max/min clamping (Python's
|
|
max(0.0, NaN) == NaN). Must be caught explicitly."""
|
|
import math
|
|
|
|
provider = _StubProvider(response=_ok_response(is_routine=True, confidence=math.nan))
|
|
v = RoutineValidator(provider, model="m")
|
|
out = await v.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out is not None
|
|
assert out.confidence == 0.0
|
|
assert math.isfinite(out.confidence)
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_validate_inf_confidence_normalized_to_zero():
|
|
"""+inf / -inf are also non-finite and must be rejected before clamping."""
|
|
import math
|
|
|
|
for bad in (math.inf, -math.inf):
|
|
provider = _StubProvider(response=_ok_response(is_routine=True, confidence=bad))
|
|
v = RoutineValidator(provider, model="m")
|
|
out = await v.validate(_routine(), _HISTORY, now_ms=_NOW)
|
|
assert out is not None
|
|
assert out.confidence == 0.0
|