Metadata/add more fields

.env.example

@@ -16,6 +16,7 @@ POSTGRES_USER=postgres
 POSTGRES_PASSWORD=postgres
 
 SENTRY_DSN=
+OTEL_ENABLED=false
 
 # Configure these with your own Docker registry images
 
@@ -26,8 +27,9 @@ GUARDRAILS_HUB_API_KEY="<ADD-KEY>"
 HF_TOKEN="<ADD-HF-KEY>"
 # SHA-256 hex digest of your bearer token (64 lowercase hex chars)
 AUTH_TOKEN="<ADD-HASH-TOKEN>"
-KAAPI_AUTH_URL="<ADD-KAAPI-AUTH-URL>"
-KAAPI_AUTH_TIMEOUT=5
+# Comma-separated source IPs allowed to call this service (kaapi-backend).
+# Empty disables the check; required when ENVIRONMENT=production.
+ALLOWED_IPS="127.0.0.1"
 
 # URL for the guardrails API — required for the multiple_validators evaluation script
 GUARDRAILS_API_URL="http://localhost:8001/api/v1/guardrails/"

.env.test.example

@@ -25,5 +25,6 @@ OPENAI_API_KEY="<ADD-KEY>"
 GUARDRAILS_HUB_API_KEY="<ADD-KEY>"
 # SHA-256 hex digest of your bearer token (64 lowercase hex chars)
 AUTH_TOKEN="<ADD-HASH-TOKEN>"
-KAAPI_AUTH_URL="<ADD-KAAPI-AUTH-URL>"
-KAAPI_AUTH_TIMEOUT=5
+# Comma-separated source IPs allowed to call this service (kaapi-backend).
+# Empty disables the check; required when ENVIRONMENT=production.
+ALLOWED_IPS="127.0.0.1"

.github/workflows/continuous-integration.yml

@@ -42,7 +42,6 @@ jobs:
             AUTH_TOKEN_HASH="$(echo -n "ci-test-token" | sha256sum | awk '{print $1}')"
           fi
           sed -i "s|AUTH_TOKEN=\"<ADD-HASH-TOKEN>\"|AUTH_TOKEN=\"${AUTH_TOKEN_HASH}\"|" .env.test
-          sed -i "s|GUARDRAILS_HUB_API_KEY=\"<ADD-KEY>\"|GUARDRAILS_HUB_API_KEY=\"${{ secrets.GUARDRAILS_HUB_API_KEY }}\"|" .env.test
           cp .env.test .env
 
       - name: Install uv
@@ -55,15 +54,6 @@ jobs:
         run: uv sync
         working-directory: backend
 
-      - name: Install Guardrails hub validators
-        env:
-          GUARDRAILS_HUB_API_KEY: ${{ secrets.GUARDRAILS_HUB_API_KEY }}
-        run: |
-          source .venv/bin/activate
-          chmod +x scripts/install_guardrails_from_hub.sh
-          scripts/install_guardrails_from_hub.sh
-        working-directory: backend
-
       - name: Activate virtual environment and run Alembic migrations
         run: |
           source .venv/bin/activate

.gitignore

@@ -16,4 +16,6 @@ backend/app/evaluation/outputs/**/*.json
 
 # marimo related artefacts
 backend/notebooks/*.csv
-**/__marimo__/
+**/__marimo__/
+
+backend/logs/app.log

backend/.guardrails/hub_registry.json

@@ -6,39 +6,39 @@
       "exports": [
         "BanList"
       ],
-      "installed_at": "2026-05-12T05:28:55.036165+00:00",
+      "installed_at": "2026-08-26T05:41:11.480494+00:00",
       "package_name": "guardrails-grhub-ban-list"
     },
     "guardrails/llm_critic": {
       "import_path": "guardrails_grhub_llm_critic",
       "exports": [
         "LLMCritic"
       ],
-      "installed_at": "2026-05-12T05:29:05.596003+00:00",
+      "installed_at": "2026-08-26T05:41:14.741349+00:00",
       "package_name": "guardrails-grhub-llm-critic"
     },
     "guardrails/llamaguard_7b": {
       "import_path": "guardrails_grhub_llamaguard_7b",
       "exports": [
         "LlamaGuard7B"
       ],
-      "installed_at": "2026-05-12T05:29:16.995441+00:00",
+      "installed_at": "2026-08-26T05:41:18.742264+00:00",
       "package_name": "guardrails-grhub-llamaguard-7b"
     },
     "guardrails/profanity_free": {
       "import_path": "guardrails_grhub_profanity_free",
       "exports": [
         "ProfanityFree"
       ],
-      "installed_at": "2026-05-12T05:29:28.079410+00:00",
+      "installed_at": "2026-08-26T05:41:26.279899+00:00",
       "package_name": "guardrails-grhub-profanity-free"
     },
     "guardrails/nsfw_text": {
       "import_path": "guardrails_grhub_nsfw_text",
       "exports": [
         "NSFWText"
       ],
-      "installed_at": "2026-05-12T05:29:50.654888+00:00",
+      "installed_at": "2026-08-26T05:41:36.879459+00:00",
       "package_name": "guardrails-grhub-nsfw-text"
     }
   }

backend/Dockerfile

@@ -58,22 +58,15 @@ ENV HF_HOME=/app/hf_cache
 RUN --mount=type=secret,id=HF_TOKEN \
     HF_TOKEN="$(cat /run/secrets/HF_TOKEN 2>/dev/null || true)" \
     /app/.venv/bin/python -c "from transformers import AutoTokenizer, AutoModelForSequenceClassification; \
-AutoTokenizer.from_pretrained('textdetox/xlmr-large-toxicity-classifier', cache_dir='/app/hf_cache'); \
-AutoModelForSequenceClassification.from_pretrained('textdetox/xlmr-large-toxicity-classifier', cache_dir='/app/hf_cache')"
+AutoTokenizer.from_pretrained('michellejieli/NSFW_text_classifier', cache_dir='/app/hf_cache'); \
+AutoModelForSequenceClassification.from_pretrained('michellejieli/NSFW_text_classifier', cache_dir='/app/hf_cache')"
 
 RUN --mount=type=secret,id=HF_TOKEN \
     HF_TOKEN="$(cat /run/secrets/HF_TOKEN 2>/dev/null || true)" \
     /app/.venv/bin/python -c "from transformers import AutoTokenizer, AutoModelForTokenClassification; \
 AutoTokenizer.from_pretrained('dslim/bert-base-NER-uncased', cache_dir='/app/hf_cache'); \
 AutoModelForTokenClassification.from_pretrained('dslim/bert-base-NER-uncased', cache_dir='/app/hf_cache')"
 
-# Pre-install Guardrails hub validators so container startup is not blocked by downloads
-RUN --mount=type=secret,id=GUARDRAILS_HUB_API_KEY \
-    GUARDRAILS_HUB_API_KEY="$(cat /run/secrets/GUARDRAILS_HUB_API_KEY 2>/dev/null || true)" \
-    ENABLE_METRICS="false" \
-    ENABLE_REMOTE_INFERENCING="true" \
-    /app/scripts/install_guardrails_from_hub.sh
-
 # -------------------------------
 # Entrypoint (runtime setup)
 # -------------------------------

backend/README.md

@@ -191,19 +191,32 @@ echo -n "your-plain-text-token" | shasum -a 256
 
 Set the resulting digest as `AUTH_TOKEN` in your `.env` / `.env.test`.
 
-## Multi-tenant API Key Configuration
+## Caller Authentication
 
-Ban List and LLM Prompt Config APIs use `X-API-KEY` auth instead of bearer token auth.
+kaapi-guardrails is an internal service. Its only caller is kaapi-backend, which authenticates the
+end user and resolves the tenant before calling.
 
-Required environment variables:
-- `KAAPI_AUTH_URL`: Base URL of the Kaapi auth service used to verify API keys.
-- `KAAPI_AUTH_TIMEOUT`: Timeout in seconds for auth verification calls.
+Every request (except the health check) must carry:
 
-At runtime, the backend calls:
-- `GET {KAAPI_AUTH_URL}/apikeys/verify`
-- Header: `X-API-KEY: <token>`
+- `Authorization: Bearer <plain-text-token>`
+- `X-ORGANIZATION-ID: <int>` and `X-PROJECT-ID: <int>` — the tenant, resolved by kaapi-backend
+- and originate from an IP listed in `ALLOWED_IPS`
 
-If verification succeeds, tenant's scope (`organization_id`, `project_id`) is resolved from the auth response and applied to tenant-scoped CRUD operations (for example Ban Lists and LLM Prompt Configs).
+`ALLOWED_IPS` is a comma-separated list of source IPs allowed to reach this service:
+
+```bash
+ALLOWED_IPS="10.0.3.14"
+```
+
+Leave it empty to disable the check for local development. It is required when
+`ENVIRONMENT=production` — the service will not start without it.
+
+Rejections:
+- Source IP not whitelisted → `403` (checked before the token)
+- Missing or invalid token → `401`
+- Missing or non-integer tenant headers → `422`
+
+Tenant scope is never read from the query string or request body.
 
 ## Guardrails AI Setup
 
@@ -339,3 +352,5 @@ class GenderAssumptionBiasSafetyValidatorConfig(BaseValidatorConfig):
 ```
 
 3. In `backend/app/schemas/guardrail_config.py`, add the newly created config class to `ValidatorConfigItem`.
+
+# test

backend/app/alembic/versions/010_add_traceability_columns.py

@@ -0,0 +1,84 @@
+"""Add traceability columns to request_log and validator_log
+
+Revision ID: 010
+Revises: 009
+Create Date: 2026-08-25 00:00:00.000000
+
+"""
+
+from typing import Sequence, Union
+
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects.postgresql import JSONB
+
+revision: str = "010"
+down_revision = "009"
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "request_log",
+        sa.Column(
+            "metadata",
+            JSONB,
+            nullable=True,
+            comment="Full run_guardrails request payload",
+        ),
+    )
+    op.add_column(
+        "validator_log",
+        sa.Column(
+            "order",
+            sa.Integer(),
+            nullable=True,
+            comment="1-based execution order of the validator within the request",
+        ),
+    )
+    op.add_column(
+        "validator_log",
+        sa.Column(
+            "duration_ms",
+            sa.Integer(),
+            nullable=True,
+            comment="Wall-clock execution time of the validator in milliseconds",
+        ),
+    )
+    op.add_column(
+        "validator_log",
+        sa.Column(
+            "stage",
+            sa.String(),
+            nullable=True,
+            comment="Stage the validator checked (input or output)",
+        ),
+    )
+    op.add_column(
+        "validator_log",
+        sa.Column(
+            "type",
+            sa.String(),
+            nullable=True,
+            comment="Validator type (ValidatorType enum value)",
+        ),
+    )
+    op.add_column(
+        "validator_log",
+        sa.Column(
+            "metadata",
+            JSONB,
+            nullable=True,
+            comment="Full resolved validator config used for this run",
+        ),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("validator_log", "metadata")
+    op.drop_column("validator_log", "type")
+    op.drop_column("validator_log", "stage")
+    op.drop_column("validator_log", "duration_ms")
+    op.drop_column("validator_log", "order")
+    op.drop_column("request_log", "metadata")

backend/app/alembic/versions/011_rename_validator_log_request_id.py

@@ -0,0 +1,35 @@
+"""Rename validator_log.request_id to request_log_id
+
+The column is a foreign key to request_log.id (the surrogate PK), not to
+request_log.request_id (the caller-supplied business id) - the old name
+made it easy to join on the wrong column.
+
+Revision ID: 011
+Revises: 010
+Create Date: 2026-09-04 00:00:00.000000
+
+"""
+
+from typing import Sequence, Union
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision: str = "011"
+down_revision = "010"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.alter_column("validator_log", "request_id", new_column_name="request_log_id")
+    op.execute(
+        "ALTER INDEX idx_validator_log_request_id RENAME TO idx_validator_log_request_log_id"
+    )
+
+
+def downgrade() -> None:
+    op.execute(
+        "ALTER INDEX idx_validator_log_request_log_id RENAME TO idx_validator_log_request_id"
+    )
+    op.alter_column("validator_log", "request_log_id", new_column_name="request_id")