Merge pull request #506 from sylvesterkaczmarek/fix/inject-data-validation

fix(agent): reject invalid telemetry injection payloads

backend/routers/ai_intel.py

@@ -125,8 +125,8 @@ class LayerUpdate(BaseModel):
 
 class InjectRequest(BaseModel):
     layer: str
-    items: list[dict[str, Any]] = Field(..., max_length=200)
-    mode: str = "append"  # "append" or "replace"
+    items: list[Any] = Field(..., max_length=200)
+    mode: str = "append"  # validated by inject_layer_data
 
 
 # ---------------------------------------------------------------------------
@@ -641,39 +641,15 @@ INJECTABLE_LAYERS = {
 @router.post("/api/ai/inject", dependencies=[Depends(require_openclaw_or_local)])
 @limiter.limit("30/minute")
 async def inject_data(request: Request, body: InjectRequest):
-    """Inject custom data into ANY native ShadowBroker layer.
-    Items appear as real telemetry alongside automated feeds.
-    Tagged with _source='user:openclaw' so they can be filtered/removed."""
-    from services.fetchers._store import latest_data, _data_lock, bump_data_version
+    """Inject custom data through the shared validated layer helper."""
+    from services.ai_intel_store import inject_layer_data
 
-    if body.layer not in INJECTABLE_LAYERS:
-        raise HTTPException(400, f"Layer '{body.layer}' is not injectable. "
-                            f"Valid layers: {sorted(INJECTABLE_LAYERS)}")
+    result = inject_layer_data(body.layer, body.items, mode=body.mode)
+    if not result.get("ok"):
+        raise HTTPException(400, result.get("detail", "invalid injection payload"))
 
-    now = time.time()
-    items = body.items[:200]  # cap at 200
-
-    # Tag every injected item
-    for item in items:
-        item["_injected"] = True
-        item["_source"] = "user:openclaw"
-        item["_injected_at"] = now
-
-    with _data_lock:
-        existing = list(latest_data.get(body.layer) or [])
-        if body.mode == "replace":
-            existing = [x for x in existing if not x.get("_injected")]
-        existing.extend(items)
-        latest_data[body.layer] = existing
-        bump_data_version()
-
-    total = len(latest_data.get(body.layer, []))
-    return {
-        "ok": True,
-        "layer": body.layer,
-        "injected": len(items),
-        "total": total,
-    }
+    total = len(_latest_data.get(body.layer, []))
+    return {**result, "total": total}
 
 
 @router.delete("/api/ai/inject", dependencies=[Depends(require_openclaw_or_local)])

backend/services/ai_intel_store.py

@@ -121,6 +121,10 @@ def inject_layer_data(
     if layer not in _INJECTABLE_LAYERS:
         return {"ok": False, "detail": f"layer '{layer}' not injectable"}
 
+    mode = str(mode or "").strip().lower()
+    if mode not in {"append", "replace"}:
+        return {"ok": False, "detail": "mode must be 'append' or 'replace'"}
+
     items = list(items or [])[:200]
     if not items:
         return {"ok": False, "detail": "no items provided"}
@@ -136,6 +140,9 @@ def inject_layer_data(
         entry["_injected_at"] = now
         tagged.append(entry)
 
+    if not tagged:
+        return {"ok": False, "detail": "no valid items provided"}
+
     with _data_lock:
         current = latest_data.get(layer)
         existing = list(current) if isinstance(current, list) else []

backend/services/openclaw_channel.py

@@ -1362,7 +1362,9 @@ def _dispatch_command(cmd: str, args: dict[str, Any]) -> dict[str, Any]:
         if not layer or not items:
             return {"ok": False, "detail": "layer and items required"}
         from services.ai_intel_store import inject_layer_data
-        result = inject_layer_data(layer, items)
+        result = inject_layer_data(layer, items, mode=args.get("mode", "append"))
+        if not result.get("ok"):
+            return result
         return {"ok": True, "data": result}
 
     if cmd == "create_layer":

backend/tests/test_ai_intel_store_copy_on_write.py

@@ -1,4 +1,4 @@
-"""Regression coverage for copy-on-write OpenClaw layer injection."""
+"""Regression coverage for OpenClaw layer injection."""
 
 from services import ai_intel_store
 from services.fetchers import _store
@@ -13,47 +13,51 @@ def _publish_test_layer(monkeypatch, items):
 
 def test_append_does_not_mutate_previously_published_list(monkeypatch):
     before = _publish_test_layer(monkeypatch, [{"id": "existing"}])
-
     result = ai_intel_store.inject_layer_data(
-        "air_quality",
-        [{"id": "injected"}],
-        mode="append",
+        "air_quality", [{"id": "injected"}], mode="append"
     )
-
     after = _store.latest_data["air_quality"]
-    assert result == {
-        "ok": True,
-        "layer": "air_quality",
-        "injected": 1,
-        "mode": "append",
-    }
+    assert result["ok"] is True
     assert after is not before
     assert before == [{"id": "existing"}]
     assert [item["id"] for item in after] == ["existing", "injected"]
-    assert after[-1]["_injected"] is True
-    assert after[-1]["_source"] == "user:openclaw"
 
 
 def test_replace_does_not_mutate_previously_published_list(monkeypatch):
     before = _publish_test_layer(
         monkeypatch,
-        [
-            {"id": "native"},
-            {"id": "old-injected", "_injected": True},
-        ],
+        [{"id": "native"}, {"id": "old-injected", "_injected": True}],
     )
-
     result = ai_intel_store.inject_layer_data(
-        "air_quality",
-        [{"id": "new-injected"}],
-        mode="replace",
+        "air_quality", [{"id": "new-injected"}], mode="replace"
     )
-
     after = _store.latest_data["air_quality"]
     assert result["ok"] is True
     assert after is not before
     assert before == [
-        {"id": "native"},
-        {"id": "old-injected", "_injected": True},
+        {"id": "native"}, {"id": "old-injected", "_injected": True}
     ]
     assert [item["id"] for item in after] == ["native", "new-injected"]
+
+
+def test_replace_rejects_invalid_items_without_clearing_existing_data(monkeypatch):
+    before = _publish_test_layer(
+        monkeypatch,
+        [{"id": "native"}, {"id": "old-injected", "_injected": True}],
+    )
+    result = ai_intel_store.inject_layer_data(
+        "air_quality", ["not-a-mapping"], mode="replace"
+    )
+    assert result == {"ok": False, "detail": "no valid items provided"}
+    assert _store.latest_data["air_quality"] is before
+
+
+def test_unknown_mode_is_rejected_without_mutating_layer(monkeypatch):
+    before = _publish_test_layer(monkeypatch, [{"id": "existing"}])
+    result = ai_intel_store.inject_layer_data(
+        "air_quality", [{"id": "injected"}], mode="overwrite"
+    )
+    assert result == {
+        "ok": False, "detail": "mode must be 'append' or 'replace'"
+    }
+    assert _store.latest_data["air_quality"] is before

backend/tests/test_injection_entrypoints.py

@@ -0,0 +1,102 @@
+"""Regression coverage for public and OpenClaw telemetry injection paths."""
+
+import pytest
+from fastapi import HTTPException
+
+from routers import ai_intel
+from services.fetchers import _store
+from services.openclaw_channel import _dispatch_command
+
+
+def _publish(monkeypatch, items):
+    published = list(items)
+    monkeypatch.setitem(_store.latest_data, "air_quality", published)
+    monkeypatch.setattr(_store, "bump_data_version", lambda: None)
+    return published
+
+
+def test_openclaw_rejects_unknown_mode_without_mutation(monkeypatch):
+    before = _publish(monkeypatch, [{"id": "existing"}])
+    result = _dispatch_command(
+        "inject_data",
+        {"layer": "air_quality", "items": [{"id": "new"}], "mode": "overwrite"},
+    )
+    assert result == {"ok": False, "detail": "mode must be 'append' or 'replace'"}
+    assert _store.latest_data["air_quality"] is before
+
+
+def test_openclaw_all_invalid_replace_preserves_existing(monkeypatch):
+    before = _publish(
+        monkeypatch,
+        [{"id": "native"}, {"id": "old", "_injected": True}],
+    )
+    result = _dispatch_command(
+        "inject_data",
+        {"layer": "air_quality", "items": ["invalid"], "mode": "replace"},
+    )
+    assert result == {"ok": False, "detail": "no valid items provided"}
+    assert _store.latest_data["air_quality"] is before
+
+
+def test_openclaw_valid_replace_preserves_native_non_mapping_entries(monkeypatch):
+    _publish(
+        monkeypatch,
+        ["native-marker", {"id": "native"}, {"id": "old", "_injected": True}],
+    )
+    result = _dispatch_command(
+        "inject_data",
+        {"layer": "air_quality", "items": [{"id": "new"}], "mode": "replace"},
+    )
+    assert result["ok"] is True
+    after = _store.latest_data["air_quality"]
+    assert after[0] == "native-marker"
+    assert after[1] == {"id": "native"}
+    assert after[2]["id"] == "new"
+
+
+@pytest.mark.asyncio
+async def test_rest_rejects_unknown_mode_without_mutation(monkeypatch):
+    before = _publish(monkeypatch, [{"id": "existing"}])
+    body = ai_intel.InjectRequest(
+        layer="air_quality", items=[{"id": "new"}], mode="overwrite"
+    )
+    with pytest.raises(HTTPException) as exc_info:
+        await ai_intel.inject_data(None, body)
+    assert exc_info.value.status_code == 400
+    assert _store.latest_data["air_quality"] is before
+
+
+@pytest.mark.asyncio
+async def test_rest_empty_and_all_invalid_replace_preserve_existing(monkeypatch):
+    for invalid_items in ([], ["invalid"]):
+        before = _publish(
+            monkeypatch,
+            [{"id": "native"}, {"id": "old", "_injected": True}],
+        )
+        body = ai_intel.InjectRequest(
+            layer="air_quality", items=invalid_items, mode="replace"
+        )
+        with pytest.raises(HTTPException) as exc_info:
+            await ai_intel.inject_data(None, body)
+        assert exc_info.value.status_code == 400
+        assert _store.latest_data["air_quality"] is before
+
+
+@pytest.mark.asyncio
+async def test_rest_valid_replace_uses_shared_helper(monkeypatch):
+    _publish(
+        monkeypatch,
+        ["native-marker", {"id": "native"}, {"id": "old", "_injected": True}],
+    )
+    body = ai_intel.InjectRequest(
+        layer="air_quality", items=[{"id": "new"}], mode="replace"
+    )
+    result = await ai_intel.inject_data(None, body)
+    assert result["ok"] is True
+    assert result["mode"] == "replace"
+    assert result["injected"] == 1
+    assert result["total"] == 3
+    after = _store.latest_data["air_quality"]
+    assert after[0] == "native-marker"
+    assert after[1] == {"id": "native"}
+    assert after[2]["id"] == "new"