LCORE-2547: finish OGX runtime naming cleanup

.github/workflows/e2e_tests.yaml

@@ -54,8 +54,7 @@ jobs:
       E2E_OPENAI_MODEL: ${{ vars.E2E_OPENAI_MODEL }}
       E2E_DEPLOYMENT_MODE: ${{ matrix.mode }}
       FAISS_VECTOR_STORE_ID: ${{ vars.FAISS_VECTOR_STORE_ID }}
-      # Override via repo Actions variable E2E_LLAMA_HOSTNAME; default matches server-mode lightspeed-stack.yaml
-      E2E_LLAMA_HOSTNAME: ${{ vars.E2E_LLAMA_HOSTNAME || 'llama-stack' }}
+      E2E_OGX_HOSTNAME: ogx
 
     steps:
       - uses: actions/checkout@v7
@@ -289,8 +288,8 @@ jobs:
           echo "=== Test failure logs ==="
 
           if [ "${{ matrix.mode }}" == "server" ]; then
-            echo "=== llama-stack logs ==="
-            docker compose logs llama-stack
+            echo "=== OGX logs ==="
+            docker compose logs ogx
             echo ""
             echo "=== lightspeed-stack logs ==="
             docker compose logs lightspeed-stack

.github/workflows/e2e_tests_providers.yaml

@@ -355,8 +355,8 @@ jobs:
           echo "=== Test failure logs ==="
 
           if [ "${{ matrix.mode }}" == "server" ]; then
-            echo "=== llama-stack logs ==="
-            docker compose logs llama-stack
+            echo "=== OGX logs ==="
+            docker compose logs ogx
             echo ""
             echo "=== lightspeed-stack logs ==="
             docker compose logs lightspeed-stack

.github/workflows/e2e_tests_rhaiis.yaml

@@ -249,8 +249,8 @@ jobs:
           echo "=== Test failure logs ==="
 
           if [ "${{ matrix.mode }}" == "server" ]; then
-            echo "=== llama-stack logs ==="
-            docker compose logs llama-stack
+            echo "=== OGX logs ==="
+            docker compose logs ogx
             echo ""
             echo "=== lightspeed-stack logs ==="
             docker compose logs lightspeed-stack

.gitignore

@@ -183,7 +183,7 @@ dev/
 # VSCode
 .vscode/
 
-# Llama related - when running the stack as lib client
+# OGX related - when running the stack as lib client
 .llama
 
 # Database files
@@ -195,7 +195,7 @@ requirements.*.backup
 # Local run files
 local-run.yaml
 
-# Synthesized Llama Stack run.yaml written by unified library mode (LCORE-2336)
+# Synthesized OGX run.yaml written by unified library mode (LCORE-2336)
 .generated/
 
 # Sisyphus planning files

.tekton/integration-tests/pipeline/lightspeed-stack-integration-test.yaml

@@ -15,7 +15,7 @@ spec:
       default: '{"components": [{"name":"lightspeed-stack-0-8", "containerImage": "quay.io/example/lightspeed-stack-0-8:latest"}]}'
       type: string
     - name: llama-stack-image
-      description: 'Llama Stack runs from source on UBI (init container clones repo and installs deps). Kept for logging/backwards compatibility.'
+      description: 'OGX runs from source on UBI (init container clones repo and installs deps). Parameter name kept for backwards compatibility.'
       default: 'run-from-source (UBI)'
       type: string
     - name: test-name
@@ -81,7 +81,7 @@ spec:
               - name: instanceType
                 value: "m5.large"
     - name: get-stack-images
-      description: Extract lightspeed-stack image and commit from SNAPSHOT (Llama Stack runs from source in-pod)
+      description: Extract lightspeed-stack image and commit from SNAPSHOT (OGX runs from source in-pod)
       runAfter:
         - provision-cluster
       params:

deploy/ogx/README.md

@@ -1,9 +1,9 @@
-# Llama Stack container image
+# OGX container image
 
-`test.containerfile` builds the Llama Stack server image used by
-`docker-compose.yaml` (server mode, e.g. for the e2e suite). Besides the
-Llama Stack distribution itself, the image bundles the pieces needed to
-generate its run configuration at container start:
+`test.containerfile` builds the OGX server image used by `docker-compose.yaml`
+(server mode, e.g. for the e2e suite). Besides the OGX distribution itself, the
+image bundles the pieces needed to generate its run configuration at container
+start:
 
 - `/opt/app-root/ogx_configuration.py` — the config-generation
   script (copied from `src/ogx_configuration.py`).
@@ -21,19 +21,20 @@ shape:
 
 - **Unified mode** — the `lightspeed-stack.yaml` carries a *synthesis
   input* (a non-empty `inference.providers` or `vector_store.providers`,
-  or a `llama_stack.config` block). The full `run.yaml` is synthesized
-  from it; no external `run.yaml` mount is needed.
+  or an `ogx.config` block). The full `run.yaml` is synthesized from it;
+  no external `run.yaml` mount is needed.
 - **Legacy mode** — no synthesis input present. The mounted `run.yaml`
-  (`$LLAMA_STACK_CONFIG`, default `/opt/app-root/run.yaml`) is enriched
-  with lightspeed dynamic values (BYOK RAG, Solr/OKP, Azure Entra ID).
+  (`$OGX_CONFIG`, with deprecated fallback `$LLAMA_STACK_CONFIG`, default
+  `/opt/app-root/run.yaml`) is enriched with lightspeed dynamic values
+  (BYOK RAG, Solr/OKP, Azure Entra ID).
 
 The repository `docker-compose.yaml` mounts both files and works for
 either mode — with a unified `lightspeed-stack.yaml` the `run.yaml`
 mount is simply ignored. A unified-only deployment needs just:
 
 ```yaml
 services:
-  llama-stack:
+  ogx:
     build:
       context: .
       dockerfile: deploy/ogx/test.containerfile
@@ -48,5 +49,5 @@ services:
 The compose file also mounts host copies of the script, the baseline
 data directory, and the entrypoint over their baked-in counterparts, so
 `docker compose up` picks up local changes to any of them without an
-image rebuild. Rebuild (`docker compose build llama-stack`) when
+image rebuild. Rebuild (`docker compose build ogx`) when
 dependencies (`pyproject.toml` / `uv.lock`) or the providers change.

docker-compose.yaml

@@ -1,11 +1,11 @@
 services:
   # Red Hat OGX distribution with FAISS
-  llama-stack:
+  ogx:
     build:
       context: .
       dockerfile: deploy/ogx/test.containerfile
     platform: linux/amd64
-    container_name: llama-stack
+    container_name: ogx
     ports:
       - "8321:8321"  # Expose OGX on 8321 (adjust if needed)
     depends_on:
@@ -14,7 +14,7 @@ services:
     volumes:
       # Used in legacy mode only: with a unified lightspeed-stack.yaml (one
       # carrying inference.providers / vector_store.providers / a
-      # llama_stack.config block) the entrypoint synthesizes the run config
+      # ogx.config block) the entrypoint synthesizes the run config
       # from lightspeed-stack.yaml and this mount is ignored
       - ./run.yaml:/opt/app-root/run.yaml:z
       # Host copies so `docker compose up` picks up script changes without rebuilding OGX
@@ -24,7 +24,7 @@ services:
       - ${GCP_KEYS_PATH:-./tmp/.gcp-keys-dummy}:/opt/app-root/.gcp-keys:ro
       - ./lightspeed-stack.yaml:/opt/app-root/lightspeed-stack.yaml:ro,z
       # Writable OGX storage (rag/ work copy is re-seeded by entrypoint each start)
-      - llama-storage:/opt/app-root/src/.llama/storage
+      - ogx-storage:/opt/app-root/src/.llama/storage
       # Read-only e2e FAISS fixtures — never mount these as the live KV_RAG_PATH
       - ./tests/e2e/rag:/opt/app-root/src/.llama/storage/.e2e-rag-seed:ro,z
       - mock-tls-certs:/certs:ro
@@ -80,8 +80,8 @@ services:
       - SOLR_CONTENT_FIELD=${SOLR_CONTENT_FIELD:-}
       - SOLR_EMBEDDING_MODEL=${SOLR_EMBEDDING_MODEL:-}
       - SOLR_EMBEDDING_DIM=${SOLR_EMBEDDING_DIM:-}
-      #required for enrichment
-      - E2E_LLAMA_HOSTNAME=${E2E_LLAMA_HOSTNAME:-llama-stack}
+      # Required for enrichment; E2E_LLAMA_HOSTNAME is the external/GitHub Actions name
+      - E2E_OGX_HOSTNAME=${E2E_OGX_HOSTNAME:-${E2E_LLAMA_HOSTNAME:-ogx}}
     networks:
       - lightspeednet
     healthcheck:
@@ -111,17 +111,17 @@ services:
       - CLIENT_SECRET=${CLIENT_SECRET:-}
       # FAISS vector store ID (used by inline RAG config)
       - FAISS_VECTOR_STORE_ID=${FAISS_VECTOR_STORE_ID:-}
-      # Substituted in mounted lightspeed-stack.yaml (llama_stack.url); GitHub Actions sets via vars
-      - E2E_LLAMA_HOSTNAME=${E2E_LLAMA_HOSTNAME:-llama-stack}
-      - E2E_LLAMA_PORT=${E2E_LLAMA_PORT:-8321}
+      # Substituted in mounted lightspeed-stack.yaml (ogx.url); GitHub Actions sets E2E_LLAMA_HOSTNAME
+      - E2E_OGX_HOSTNAME=${E2E_OGX_HOSTNAME:-${E2E_LLAMA_HOSTNAME:-ogx}}
+      - E2E_OGX_PORT=${E2E_OGX_PORT:-${E2E_LLAMA_PORT:-8321}}
       # OpenTelemetry configuration (tracing disabled by default)
       - OTEL_EXPORTER_OTLP_ENDPOINT=${OTEL_EXPORTER_OTLP_ENDPOINT:-}
       - OTEL_EXPORTER_OTLP_PROTOCOL=${OTEL_EXPORTER_OTLP_PROTOCOL:-}
       - OTEL_SERVICE_NAME=${OTEL_SERVICE_NAME:-}
       - OTEL_ANONYMIZATION_SECRET=${OTEL_ANONYMIZATION_SECRET:-lightspeed-stack-otel-anonymization-dev-default}
       - OTEL_SDK_DISABLED=${OTEL_SDK_DISABLED:-true}
     depends_on:
-        llama-stack:
+        ogx:
           condition: service_healthy
         mock-mcp:
           condition: service_healthy
@@ -186,7 +186,7 @@ services:
 
 
 volumes:
-  llama-storage:
+  ogx-storage:
   mock-tls-certs:
 
 networks:

docs/design/ogx-config-merge/ogx-config-merge.md

@@ -351,7 +351,7 @@ None at the REST API surface. Internal API additions in
 - `migrate_config_dumb(run_yaml_path, lightspeed_yaml_path, output_path)
   -> None` — dumb-mode migration (lossless round-trip).
 - `deep_merge_list_replace(base, overlay) -> dict` — merge helper.
-- `apply_high_level_inference(ls_config, inference)` — high-level expansion.
+- `apply_high_level_inference(ogx_config, inference)` — high-level expansion.
 - `load_default_baseline() -> dict` — loads `src/data/default_run.yaml`.
 
 CLI additions in `src/lightspeed_stack.py`:
@@ -474,11 +474,11 @@ not have had a full release with a working migration path. Releases:
 3. Run `dedupe_providers_vector_io` on the baseline.
 4. Apply existing enrichment: `enrich_byok_rag`, `enrich_solr` (Azure
    Entra ID intentionally stays separate because it's a `.env`
-   side-effect, not an `ls_config` mutation).
+   side-effect, not an `ogx_config` mutation).
 5. If top-level `inference.providers` is non-empty →
-   `apply_high_level_inference(ls_config, lcs_config["inference"])`.
+   `apply_high_level_inference(ogx_config, lcs_config["inference"])`.
 6. If `unified` and `unified.native_override` non-empty →
-   `deep_merge_list_replace(ls_config, native_override)`.
+   `deep_merge_list_replace(ogx_config, native_override)`.
 7. `dedupe_providers_vector_io` again for good measure.
 8. Return the final dict.
 

docs/devel_doc/conversations_api.md

@@ -115,17 +115,17 @@ When a user makes a query **without** providing a `conversation_id`:
 ```python
 # No conversation_id provided - create a new conversation first
 conversation = await client.conversations.create(metadata={})
-llama_stack_conv_id = conversation.id
+ogx_conv_id = conversation.id
 # Store the normalized version
-conversation_id = normalize_conversation_id(llama_stack_conv_id)
+conversation_id = normalize_conversation_id(ogx_conv_id)
 
 # Use the conversation in responses.create()
 response = await client.responses.create(
     input=input_text,
     model=model_id,
     instructions=system_prompt,
     store=True,
-    conversation=llama_stack_conv_id,  # Use OGX format
+    conversation=ogx_conv_id,  # Use OGX format
     # ... other parameters
 )
 ```
@@ -151,7 +151,7 @@ ogx_conv_id = to_ogx_conversation_id(conversation_id)
 response = await client.responses.create(
     input=input_text,
     model=model_id,
-    conversation=llama_stack_conv_id,  # Existing conversation
+    conversation=ogx_conv_id,  # Existing conversation
     # ... other parameters
 )
 ```
@@ -166,7 +166,7 @@ Conversations are stored in **two databases**:
 - `openai_conversations`: Stores conversation metadata
 - `conversation_items`: Stores individual messages/turns in conversations
 
-**Configuration (in `config/llama_stack_client_config.yaml`):**
+**Configuration (in OGX `run.yaml` / library client config):**
 ```yaml
 storage:
   stores:

docs/testing/e2e_testing.md

@@ -142,10 +142,10 @@ uv run behave tests/e2e/features/health.feature --tags=-skip-in-library-mode
 | `E2E_DEPLOYMENT_MODE`           | `server`    | `server` or `library`. Drives config paths and which scenarios run (e.g. `@skip-in-library-mode`).                                        |
 | `E2E_LSC_HOSTNAME`              | `localhost` | Host of the Lightspeed Core Stack API.                                                                                                    |
 | `E2E_LSC_PORT`                  | `8080`      | Port of the Lightspeed Core Stack API.                                                                                                    |
-| `E2E_LLAMA_HOSTNAME`            | `localhost` | Host of the OGX service (server mode).                                                                                            |
-| `E2E_LLAMA_PORT`                | `8321`      | Port of the OGX service.                                                                                                          |
-| `E2E_LLAMA_STACK_URL`           | —           | Full base URL for OGX (overrides host/port if set). Used by shield helpers.                                                       |
-| `E2E_LLAMA_STACK_API_KEY`       | `xyzzy`     | API key for OGX client (e.g. shield API).                                                                                         |
+| `E2E_OGX_HOSTNAME`            | `localhost` | Host of the OGX service (server mode). GitHub Actions repo variable: `E2E_LLAMA_HOSTNAME` (compose maps it here). |
+| `E2E_OGX_PORT`                | `8321`      | Port of the OGX service.                                                                                                          |
+| `E2E_OGX_STACK_URL`           | —           | Full base URL for OGX (overrides host/port if set). Used by shield helpers.                                                       |
+| `E2E_OGX_STACK_API_KEY`       | `xyzzy`     | API key for OGX client (e.g. shield API).                                                                                         |
 | `E2E_DEFAULT_MODEL_OVERRIDE`    | —           | Override default LLM model id (e.g. `gpt-4o-mini`).                                                                                       |
 | `E2E_DEFAULT_PROVIDER_OVERRIDE` | —           | Override default provider id (e.g. `openai`).                                                                                             |
 | `FAISS_VECTOR_STORE_ID`         | —           | Vector store id for FAISS-related scenarios.                                                                                              |

lightspeed-stack.yaml

@@ -7,11 +7,11 @@ service:
   workers: 1
   color_log: true
   access_log: true
-# llama_stack configuration
+# ogx configuration
 # When using 'make run', a container is ALWAYS launched at http://localhost:8321 (hardcoded in Makefile).
-# This llama_stack section controls where lightspeed-core connects to OGX.
-# To use a different port: override with 'make run LLAMA_STACK_PORT=<port>' and update the url below,
-# or run ogx manually and don't use 'make run'.
+# This ogx section controls where lightspeed-core connects to OGX.
+# To use a different port: override with 'make run OGX_PORT=<port>' and update the url below,
+# or run OGX manually and don't use 'make run'.
 ogx:
   use_as_library_client: false
   url: http://localhost:8321

scripts/ogx-entrypoint.sh

@@ -6,12 +6,12 @@
 # configuration shape:
 #   - unified mode: the lightspeed config carries a synthesis input (a
 #     non-empty inference.providers or vector_store.providers, or a
-#     llama_stack.config / ogx.config block). The full run.yaml is synthesized
+#     ogx.config block (deprecated alias: llama_stack.config). The full run.yaml is synthesized
 #     from it — no external run.yaml mount is needed, and $OGX_CONFIG /
-#     $LLAMA_STACK_CONFIG is ignored. The shipped default baseline is read from
+#     $LLAMA_STACK_CONFIG (deprecated) is ignored. The shipped default baseline is read from
 #     /opt/app-root/data/default_run.yaml.
-#   - legacy mode: the mounted run.yaml ($OGX_CONFIG, falling back to
-#     $LLAMA_STACK_CONFIG) is enriched with lightspeed dynamic values
+#   - legacy mode: the mounted run.yaml ($OGX_CONFIG, falling back to the
+#     deprecated $LLAMA_STACK_CONFIG) is enriched with lightspeed dynamic values
 #     (BYOK RAG, Solr/OKP, Azure Entra ID).
 
 set -e

src/client/ogx.py

@@ -159,34 +159,34 @@ def _enrich_library_config(self, input_config_path: str) -> str:
         """Enrich OGX config with BYOK RAG and OKP Solr settings."""
         try:
             with open(input_config_path, encoding="utf-8") as f:
-                ls_config = yaml.safe_load(f)
+                ogx_config = yaml.safe_load(f)
         except (OSError, yaml.YAMLError) as e:
             logger.warning("Failed to read OGX config: %s", e)
             return input_config_path
 
         config = configuration.configuration
 
         # Enrichment: BYOK RAG
-        enrich_byok_rag(ls_config, [s.model_dump() for s in config.rag.byok.stores])
+        enrich_byok_rag(ogx_config, [s.model_dump() for s in config.rag.byok.stores])
 
         # Enrichment: Solr - enabled when "okp" appears in either inline or tool list
         rag_config_for_solr = {
             "inline": config.rag.retrieval.inline.sources,
             "tool": config.rag.retrieval.tool.sources,
         }
-        enrich_solr(ls_config, rag_config_for_solr, config.rag.okp.model_dump())
+        enrich_solr(ogx_config, rag_config_for_solr, config.rag.okp.model_dump())
 
         # Enrichment: Azure Entra ID deferred auth
         entra_id_config = (
             config.azure_entra_id.model_dump() if config.azure_entra_id else None
         )
-        enrich_azure_entra_id_inference(ls_config, entra_id_config)
+        enrich_azure_entra_id_inference(ogx_config, entra_id_config)
 
         enriched_path = os.path.join(tempfile.gettempdir(), "ogx_enriched_config.yaml")
 
         try:
             with open(enriched_path, "w", encoding="utf-8") as f:
-                yaml.dump(ls_config, f, Dumper=YamlDumper, default_flow_style=False)
+                yaml.dump(ogx_config, f, Dumper=YamlDumper, default_flow_style=False)
             logger.info("Wrote enriched OGX config to %s", enriched_path)
             return enriched_path
         except OSError as e:

src/configuration.py

@@ -73,14 +73,14 @@ def replace_env_vars_preserving_native_override(
     ogx_section = config_dict.get("ogx")
     if ogx_section is None:
         ogx_section = config_dict.get("llama_stack")
-    ls_config = ogx_section.get("config") if isinstance(ogx_section, dict) else None
-    if not (isinstance(ls_config, dict) and "native_override" in ls_config):
+    ogx_config = ogx_section.get("config") if isinstance(ogx_section, dict) else None
+    if not (isinstance(ogx_config, dict) and "native_override" in ogx_config):
         return replace_env_vars(config_dict)
 
-    raw_override = ls_config["native_override"]
-    ls_config["native_override"] = {}  # keep secrets out of env resolution
+    raw_override = ogx_config["native_override"]
+    ogx_config["native_override"] = {}  # keep secrets out of env resolution
     resolved = replace_env_vars(config_dict)
-    ls_config["native_override"] = raw_override  # restore source dict if reused
+    ogx_config["native_override"] = raw_override  # restore source dict if reused
     resolved_ogx = (resolved.get("ogx") or resolved.get("llama_stack") or {}).get(
         "config"
     )