## 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.
256 lines
9.3 KiB
Python
256 lines
9.3 KiB
Python
"""Shared fixtures for everos memory integration tests.
|
|
|
|
These tests drive a **real** everos runtime (real LLM + embedding +
|
|
sqlite/lancedb), gated behind the ``real_llm`` marker; they skip when the
|
|
runtime isn't configured. Configure via ``~/.everos/everos.toml``, a
|
|
repo-root ``.env``, or ``EVEROS_*`` env vars, then::
|
|
|
|
uv run pytest tests/integration -m real_llm
|
|
|
|
Harness design (mirrors everos's own ``tests/e2e/conftest.py``):
|
|
|
|
- **Full lifespan init.** everos's schema (sqlite tables, lancedb
|
|
indexes) and its OME extraction engine are created by the FastAPI app
|
|
lifespan, not on first service call. The :func:`everos_env` fixture
|
|
drives ``create_app().router.lifespan_context`` so embedded service
|
|
calls (``memorize`` / ``search``) work in-process.
|
|
- **Per-test isolation.** ``EVEROS_MEMORY__ROOT`` points at a per-test
|
|
``tmp_path`` and everos's module-level singletons (writers, pipelines,
|
|
OME engine, llm client, strategy writers) are nulled so each test
|
|
rebuilds against its own store. The fixture is therefore function
|
|
scoped (real init per test — slow + billable, but correct).
|
|
- **Async extraction.** ``memorize`` enqueues work the OME engine runs in
|
|
the background; tests must ``await pipeline_drain()`` before searching
|
|
or the just-written memories won't be visible yet.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import asyncio
|
|
import importlib
|
|
import json
|
|
import time
|
|
import uuid
|
|
from pathlib import Path
|
|
from types import SimpleNamespace
|
|
from typing import Any
|
|
|
|
import pytest
|
|
import pytest_asyncio
|
|
|
|
_DATA_DIR = Path(__file__).parent / "data"
|
|
_CORPUS_PATH = _DATA_DIR / "everos_skill_corpus.json"
|
|
|
|
# Load a gitignored repo-root ``.env`` so secrets placed there reach both
|
|
# everos (via pydantic-settings) and the resolved-settings gate below.
|
|
# Real env vars still win — ``load_dotenv`` does not override.
|
|
try:
|
|
from dotenv import load_dotenv
|
|
|
|
load_dotenv(Path(__file__).resolve().parents[2] / ".env")
|
|
except ImportError: # python-dotenv absent — rely on real env vars only
|
|
pass
|
|
|
|
|
|
# everos module-level singletons that capture MemoryRoot / engine at first
|
|
# use and survive across tests; null them so each test rebuilds against
|
|
# its own ``EVEROS_MEMORY__ROOT``. Mirrors everos tests/e2e/conftest.py.
|
|
_MEMORIZE_SINGLETONS: tuple[str, ...] = (
|
|
"_episode_writer",
|
|
"_prompt_loader",
|
|
"_user_pipeline",
|
|
"_agent_pipeline",
|
|
"_ome_engine",
|
|
)
|
|
_STRATEGY_SINGLETONS: tuple[tuple[str, tuple[str, ...]], ...] = (
|
|
("everos.memory.strategies.extract_atomic_facts", ("_writer",)),
|
|
("everos.memory.strategies.extract_foresight", ("_writer",)),
|
|
("everos.memory.strategies.extract_user_profile", ("_writer", "_reader")),
|
|
("everos.memory.strategies.extract_agent_case", ("_writer",)),
|
|
("everos.memory.strategies.extract_agent_skill", ("_writer", "_reader")),
|
|
)
|
|
|
|
|
|
def _reset_everos_singletons(monkeypatch: pytest.MonkeyPatch) -> None:
|
|
svc = importlib.import_module("everos.service.memorize")
|
|
for attr in _MEMORIZE_SINGLETONS:
|
|
monkeypatch.setattr(svc, attr, None, raising=False)
|
|
client_mod = importlib.import_module("everos.component.llm.client")
|
|
monkeypatch.setattr(client_mod, "_llm_client", None, raising=False)
|
|
for mod_name, attrs in _STRATEGY_SINGLETONS:
|
|
mod = importlib.import_module(mod_name)
|
|
for attr in attrs:
|
|
monkeypatch.setattr(mod, attr, None, raising=False)
|
|
|
|
|
|
def _missing_runtime_config(settings: Any) -> list[str]:
|
|
"""Which everos LLM/embedding fields are unset across *all* config
|
|
sources (``~/.everos/everos.toml`` + ``.env`` + ``EVEROS_*`` env).
|
|
|
|
Checking everos's resolved ``Settings`` rather than raw env vars
|
|
means a user who configures models/keys in ``config.toml`` (not env)
|
|
still un-skips the suite. Empty list == runtime ready.
|
|
"""
|
|
missing: list[str] = []
|
|
if settings.llm.api_key is None:
|
|
missing.append("llm.api_key")
|
|
if not settings.embedding.model:
|
|
missing.append("embedding.model")
|
|
if settings.embedding.api_key is None:
|
|
missing.append("embedding.api_key")
|
|
return missing
|
|
|
|
|
|
@pytest_asyncio.fixture
|
|
async def everos_env(
|
|
tmp_path: Path,
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> Any:
|
|
"""Isolated everos store + full lifespan, in the test's event loop.
|
|
|
|
Skips when everos isn't installed or its LLM / embedding runtime
|
|
isn't configured (in toml, .env, or env).
|
|
"""
|
|
pytest.importorskip("everos")
|
|
# Snapshot the real root BEFORE redirecting, so we can copy config
|
|
# files that carry LLM / embedding credentials into the isolated dir.
|
|
import shutil
|
|
|
|
from everos.config.settings import load_settings, resolve_root
|
|
|
|
load_settings.cache_clear()
|
|
real_root = resolve_root()
|
|
|
|
# Redirect data root to an isolated tmp dir so each test gets a
|
|
# fresh sqlite + lancedb.
|
|
monkeypatch.setenv("EVEROS_ROOT", str(tmp_path))
|
|
monkeypatch.setenv("EVEROS_MEMORY__ROOT", str(tmp_path))
|
|
|
|
for name in ("everos.toml", "ome.toml"):
|
|
src = real_root / name
|
|
if src.is_file():
|
|
shutil.copy2(src, tmp_path / name)
|
|
|
|
# mode=agent runs both user-memory and agent-memory pipelines.
|
|
monkeypatch.setenv("EVEROS_MEMORIZE__MODE", "agent")
|
|
# Tighten the boundary so the backend's accumulate-only store() path
|
|
# (no is_final flush) still crosses a boundary by volume in L3.
|
|
monkeypatch.setenv("EVEROS_BOUNDARY_DETECTION__HARD_MSG_LIMIT", "6")
|
|
load_settings.cache_clear()
|
|
|
|
missing = _missing_runtime_config(load_settings())
|
|
if missing:
|
|
load_settings.cache_clear()
|
|
pytest.skip(
|
|
"everos runtime not configured (missing: "
|
|
+ ", ".join(missing)
|
|
+ "); set them in ~/.everos/everos.toml, a repo-root .env, "
|
|
"or EVEROS_* env vars",
|
|
)
|
|
|
|
_reset_everos_singletons(monkeypatch)
|
|
|
|
# Bring up the everos runtime via the production path:
|
|
# EverosBackend.start() ensures the everos server is running.
|
|
from raven.plugins import PluginContext, ServiceLocator
|
|
from raven_everos.backend import EverosBackend
|
|
|
|
be = EverosBackend(
|
|
PluginContext(
|
|
config={},
|
|
services=ServiceLocator(workspace=tmp_path, user_id="default", agent_id="default"),
|
|
)
|
|
)
|
|
await be.start()
|
|
try:
|
|
yield SimpleNamespace(root=tmp_path, backend=be)
|
|
finally:
|
|
await be.stop()
|
|
load_settings.cache_clear()
|
|
|
|
|
|
@pytest.fixture
|
|
def pipeline_drain() -> Any:
|
|
"""Return an async waiter: blocks until the OME engine is idle AND
|
|
the md_change_state cascade queue is fully drained.
|
|
|
|
everos extraction is async/background — ``memorize`` returns before
|
|
the strategies (and their md -> lancedb propagation) finish. Call
|
|
``await pipeline_drain()`` after memorize/store and before search.
|
|
"""
|
|
|
|
async def _wait(*, deadline_seconds: float = 180.0, interval: float = 0.5) -> None:
|
|
from everos.infra.persistence.sqlite import md_change_state_repo
|
|
from everos.service.memorize import _get_engine
|
|
|
|
engine = _get_engine()
|
|
async with asyncio.timeout(deadline_seconds):
|
|
while True:
|
|
if await engine.wait_idle(timeout=0.5):
|
|
summary = await md_change_state_repo.queue_summary()
|
|
if summary.pending == 0:
|
|
return
|
|
await asyncio.sleep(interval)
|
|
|
|
return _wait
|
|
|
|
|
|
@pytest.fixture
|
|
def ids() -> Any:
|
|
"""Fresh identifiers per test for data isolation.
|
|
|
|
``user_id`` / ``agent_id`` are the bare everos owner identities: what
|
|
``sender_id`` is stamped with on store, what search filters on, and
|
|
what the host passes to ``backend.recall(user_id=...)`` /
|
|
``recall(agent_id=...)``.
|
|
"""
|
|
tag = uuid.uuid4().hex[:8]
|
|
user_id = f"u-{tag}"
|
|
agent_id = f"a-{tag}"
|
|
return SimpleNamespace(
|
|
user_id=user_id,
|
|
agent_id=agent_id,
|
|
session=f"sess-{tag}",
|
|
)
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def corpus() -> dict[str, Any]:
|
|
"""The seeded conversation corpus (user facts + skill demonstrations)."""
|
|
return json.loads(_CORPUS_PATH.read_text(encoding="utf-8"))
|
|
|
|
|
|
def as_everos_payload(
|
|
session_id: str,
|
|
messages: list[dict[str, Any]],
|
|
*,
|
|
user_id: str,
|
|
agent_id: str,
|
|
base_ts_ms: int | None = None,
|
|
) -> dict[str, Any]:
|
|
"""Convert AgentLoop-shape messages into the everos memorize payload.
|
|
|
|
Stamps ``sender_id`` by role to match everos's owner model (the L2
|
|
direct-service path mirrors what the backend does on L3):
|
|
``user`` -> ``user_id``; ``assistant`` / ``tool`` -> ``agent_id``.
|
|
So a later search with ``user_id`` / ``agent_id`` finds these rows.
|
|
Passes through ``tool_calls`` / ``tool_call_id`` when present (agent
|
|
case/skill extraction needs real tool round-trips).
|
|
"""
|
|
base = base_ts_ms if base_ts_ms is not None else int(time.time() * 1000)
|
|
out: list[dict[str, Any]] = []
|
|
for i, m in enumerate(messages):
|
|
role = m["role"]
|
|
entry: dict[str, Any] = {
|
|
"sender_id": user_id if role == "user" else agent_id,
|
|
"role": role,
|
|
"timestamp": base + i * 1000,
|
|
"content": m["content"],
|
|
}
|
|
if m.get("tool_calls") is not None:
|
|
entry["tool_calls"] = m["tool_calls"]
|
|
if m.get("tool_call_id") is not None:
|
|
entry["tool_call_id"] = m["tool_call_id"]
|
|
out.append(entry)
|
|
return {"session_id": session_id, "messages": out}
|