test: add unit + integration test infrastructure for kernel components

.github/workflows/test-memory-integration.yml

@@ -0,0 +1,70 @@
+name: Test Memory Integration
+
+on:
+  pull_request:
+    branches: [ "main" ]
+  push:
+    branches: [ "main" ]
+
+permissions:
+  contents: read
+  actions: write
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  test-mem0:
+    runs-on: ubuntu-latest
+
+    env:
+      OLLAMA_MAX_WAIT: 60
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python 3.11
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install uv
+          uv venv
+          source .venv/bin/activate
+          uv pip install -r requirements.txt
+
+      - name: Create and patch config for Mem0 tests
+        run: |
+          source .venv/bin/activate
+          cp aios/config/config.yaml.example aios/config/config.yaml
+          python scripts/patch_config_for_memory_tests.py
+
+      - name: Install Ollama
+        run: |
+          curl -fsSL https://ollama.com/install.sh | sh
+
+          ollama serve > ollama.log 2>&1 &
+
+          for ((i=1;i<=${OLLAMA_MAX_WAIT};i++)); do
+            curl -s http://localhost:11434/api/version && echo "Ollama ready" && break
+            echo "Waiting for Ollama… ($i/${OLLAMA_MAX_WAIT})"; sleep 1
+          done
+
+          ollama pull nomic-embed-text
+          ollama pull qwen3:1.7b
+
+      - name: Run Mem0 provider integration tests
+        run: |
+          source .venv/bin/activate
+          pytest tests/modules/memory/test_mem0_provider.py -v -m integration
+
+      - name: Upload logs
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: memory-integration-logs
+          path: |
+            ollama.log

.github/workflows/test-ollama.yml

@@ -26,10 +26,10 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - name: Set up Python 3.10
+      - name: Set up Python 3.11
         uses: actions/setup-python@v5
         with:
-          python-version: "3.10"
+          python-version: "3.11"
 
       - name: Install dependencies
         run: |

.github/workflows/test-qdrant.yml

@@ -33,10 +33,10 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - name: Set up Python 3.10
+      - name: Set up Python 3.11
         uses: actions/setup-python@v5
         with:
-          python-version: "3.10"
+          python-version: "3.11"
 
       - name: Install dependencies
         run: |

.github/workflows/test-unit.yml

@@ -0,0 +1,55 @@
+name: Test Unit
+
+on:
+  pull_request:
+    branches: [ "main" ]
+  push:
+    branches: [ "main" ]
+
+permissions:
+  contents: read
+  actions: write
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  test-unit:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python 3.11
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install uv
+          uv venv
+          source .venv/bin/activate
+          uv pip install -r requirements.txt
+
+      - name: Create config from example
+        run: cp aios/config/config.yaml.example aios/config/config.yaml
+
+      - name: Run unit tests
+        run: |
+          source .venv/bin/activate
+          # NOTE: Known-broken files excluded from the unit run:
+          #   - tests/modules/llm/openai/test_concurrent.py (broken import)
+          #   - tests/modules/llm/openai/test_single.py (broken import)
+          #   - tests/modules/memory/test_memory.py (broken import)
+          #   - tests/modules/memory/test_memory_user_identity_isolation.py
+          #     (order-dependent ConfigManager singleton pollution — passes
+          #     in isolation, fails when run after other memory tests;
+          #     tracked separately)
+          # These are pre-existing issues unrelated to the CI workflow.
+          pytest tests/ -v -m "not integration" \
+            --ignore=tests/modules/llm/openai/test_concurrent.py \
+            --ignore=tests/modules/llm/openai/test_single.py \
+            --ignore=tests/modules/memory/test_memory.py \
+            --ignore=tests/modules/memory/test_memory_user_identity_isolation.py

README.md

@@ -379,6 +379,69 @@ Make sure you have installed a virtualized environment with GUI, then you can re
 | vLLM | [All Models](https://docs.vllm.ai/en/latest/) | ✅ | model-name | vllm | - |
 | Novita | [All Models](https://novita.ai/models/llm) | ✅ | model-name | novita | NOVITA_API_KEY |
 
+## 🧪 Running Tests
+
+Tests use [pytest](https://docs.pytest.org/) with a custom `integration` marker defined in `pytest.ini`. Tests are organized under `tests/` by module.
+
+### Unit tests (fast, no external services)
+
+The CI command (see [`.github/workflows/test-unit.yml`](.github/workflows/test-unit.yml) for the source of truth):
+
+```bash
+pytest tests/ -v -m "not integration" \
+  --ignore=tests/modules/llm/openai/test_concurrent.py \
+  --ignore=tests/modules/llm/openai/test_single.py \
+  --ignore=tests/modules/memory/test_memory.py \
+  --ignore=tests/modules/memory/test_memory_user_identity_isolation.py
+```
+
+### Integration tests (requires external services)
+
+Integration tests require different services depending on the file:
+
+- **`test_shared_memory_e2e.py`** — requires Ollama + ChromaDB. Skips gracefully if Ollama is unreachable.
+- **`test_mem0_provider.py`** — requires Ollama (embedder + LLM) + ChromaDB. Does **not** skip gracefully; will fail/error if services aren't running.
+- **`test_zep_provider.py`** — requires a configured Zep Cloud API key. Does **not** skip gracefully; will fail/error without valid credentials.
+- **`test_qdrant_integration.py`** — requires a running Qdrant server. Skips via `unittest.SkipTest` if Qdrant is unreachable.
+
+Running `-m integration` without every service up will show failures for the files whose services are missing. This is expected in a partial local setup, not a broken test suite.
+
+```bash
+# For Mem0/ChromaDB tests only (most common local setup):
+#   ollama serve
+#   ollama pull nomic-embed-text
+python scripts/patch_config_for_memory_tests.py
+pytest tests/modules/memory/test_mem0_provider.py tests/modules/memory/test_shared_memory_e2e.py -v -m integration
+
+# Full integration suite (requires ALL services):
+pytest tests/ -v -m integration --continue-on-collection-errors
+```
+
+To force-test the graceful-skip path in `test_shared_memory_e2e.py`: `OLLAMA_SKIP=1 pytest tests/modules/memory/test_shared_memory_e2e.py -v -m integration`.
+
+### Running a specific module
+
+```bash
+pytest tests/modules/terminal/ -v
+pytest tests/modules/scheduler/ -v
+pytest tests/modules/syscall/ -v
+```
+
+### Marker system
+
+- `integration` — requires external services (Ollama, ChromaDB, Qdrant, Zep). Registered in `pytest.ini` and `tests/conftest.py`.
+- Unmarked tests are pure/mocked and run without any services.
+
+### Known pre-existing issues
+
+| File | Problem | Status |
+|------|---------|--------|
+| `tests/modules/llm/openai/test_concurrent.py` | Broken import (missing module) | Excluded via `--ignore` |
+| `tests/modules/llm/openai/test_single.py` | Broken import (missing module) | Excluded via `--ignore` |
+| `tests/modules/memory/test_memory.py` | Broken import (missing module) | Excluded via `--ignore` |
+| `tests/modules/memory/test_memory_user_identity_isolation.py` | Order-dependent failure — passes in isolation, fails in full suite due to `ConfigManager` singleton state pollution from earlier tests | Excluded via `--ignore` |
+| `tests/modules/memory/test_mem0_provider.py`, `test_zep_provider.py` | No graceful skip when backing services are unavailable — will produce failures/errors in a partial local setup | Expected; run only when services are configured |
+
 ## 🔧 Experimental Rust Rewrite (aios-rs)
 An early experimental Rust scaffold lives in `aios-rs/` providing trait definitions and minimal placeholder implementations (context, memory, storage, tool, scheduler, llm). This is NOT feature-parity yet; it's a foundation for incremental porting and performance-focused components.
 

pytest.ini

@@ -0,0 +1,7 @@
+[pytest]
+testpaths = tests
+python_files = test_*.py
+python_classes = Test*
+python_functions = test_*
+markers =
+    integration: requires external services (Ollama, ChromaDB, etc.)

requirements.txt

@@ -19,6 +19,7 @@ redis>=4.5.1
 sentence-transformers
 nltk
 hypothesis
+pytest
 scikit-learn
 pulp
 gdown

scripts/patch_config_for_memory_tests.py

@@ -0,0 +1,79 @@
+#!/usr/bin/env python3
+"""
+Patch aios/config/config.yaml for memory integration tests.
+
+Sets:
+  - memory.provider = "mem0"
+  - memory.auto_extract = true
+  - memory.auto_inject = true
+  - memory.mem0.embedder uses ollama/nomic-embed-text
+  - memory.mem0.vector_store uses chroma with persistence path
+
+Reads from and writes to aios/config/config.yaml by default.
+Pass a path argument to override.
+
+Usage:
+    python scripts/patch_config_for_memory_tests.py
+    python scripts/patch_config_for_memory_tests.py path/to/config.yaml
+"""
+import sys
+
+import yaml
+
+
+def patch_config(config_path: str = "aios/config/config.yaml") -> None:
+    with open(config_path) as f:
+        cfg = yaml.safe_load(f)
+
+    memory = cfg.setdefault("memory", {})
+
+    # Switch provider to mem0
+    memory["provider"] = "mem0"
+
+    # Enable personalization pipeline
+    memory["auto_extract"] = True
+    memory["auto_inject"] = True
+
+    # Ensure mem0 section is present and correct
+    mem0 = memory.setdefault("mem0", {})
+    mem0["user_id"] = "default"
+
+    # LLM: ollama (avoids needing an OpenAI API key in CI)
+    mem0["llm"] = {
+        "provider": "ollama",
+        "config": {
+            "model": "qwen3:1.7b",
+            "ollama_base_url": "http://localhost:11434",
+        },
+    }
+
+    # Embedder: ollama + nomic-embed-text
+    mem0["embedder"] = {
+        "provider": "ollama",
+        "config": {
+            "model": "nomic-embed-text",
+            "ollama_base_url": "http://localhost:11434",
+        },
+    }
+
+    # Vector store: chroma with persistence
+    mem0["vector_store"] = {
+        "provider": "chroma",
+        "config": {
+            "collection_name": "mem0_memories",
+            "path": ".mem0/chroma",
+        },
+    }
+
+    with open(config_path, "w") as f:
+        yaml.dump(cfg, f, default_flow_style=False, sort_keys=False)
+
+    print(f"Patched {config_path}: memory.provider=mem0, "
+          f"llm=ollama/qwen3:1.7b, "
+          f"embedder=ollama/nomic-embed-text, "
+          f"vector_store=chroma (.mem0/chroma)")
+
+
+if __name__ == "__main__":
+    path = sys.argv[1] if len(sys.argv) > 1 else "aios/config/config.yaml"
+    patch_config(path)

tests/conftest.py

@@ -0,0 +1,32 @@
+"""
+Root conftest for the AIOS test suite.
+
+Responsibilities:
+    1. Register the ``integration`` marker programmatically so that
+       pytest never emits an "unknown marker" warning — even if
+       pytest.ini is not picked up for any reason.
+    2. Ensure the project root is importable (sys.path fix).
+"""
+
+import os
+import sys
+
+import pytest
+
+# ── sys.path fix ──────────────────────────────────────────────────────
+# Many existing test files already do their own
+#     sys.path.insert(0, <project_root>)
+# This ensures the project root is always available regardless of the
+# working directory pytest is invoked from, without duplicating entries.
+_PROJECT_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+if _PROJECT_ROOT not in sys.path:
+    sys.path.insert(0, _PROJECT_ROOT)
+
+
+# ── Marker registration ──────────────────────────────────────────────
+def pytest_configure(config: pytest.Config) -> None:
+    """Register custom markers as a safety-net backup to pytest.ini."""
+    config.addinivalue_line(
+        "markers",
+        "integration: requires external services (Ollama, ChromaDB, etc.)",
+    )