add mcp support and update paths (#73)

issue: https://issues.redhat.com/browse/AAP-50401
- add mcp support
- update paths
- Fixes llama-stack integration
Note this is a first step, more work to be done to update and fix the container for a right
dependency management with lightspeed-stack.

Signed-off-by: Djebran Lezzoum <ldjebran@gmail.com>

Author: ldjebran

Containerfile

@@ -29,14 +29,16 @@ RUN uv sync --locked --no-install-project --no-dev
 # ======================================================
 # Final image without uv package manager
 # ------------------------------------------------------
-FROM quay.io/lightspeed-core/lightspeed-stack:dev-latest
+#FROM quay.io/lightspeed-core/lightspeed-stack:dev-latest
+# the lastest was broken, replace by the latest working image for now
+FROM quay.io/lightspeed-core/lightspeed-stack:dev-20250722-28abfaf
 
 USER 0
 
 # Re-declaring arguments without a value, inherits the global default one.
 ARG APP_ROOT
 ARG ANSIBLE_CHATBOT_VERSION
-RUN microdnf install -y --nodocs --setopt=keepcache=0 --setopt=tsflags=nodocs python3.11 jq
+RUN microdnf install -y --nodocs --setopt=keepcache=0 --setopt=tsflags=nodocs jq
 
 # PYTHONDONTWRITEBYTECODE 1 : disable the generation of .pyc
 # PYTHONUNBUFFERED 1 : force the stdout and stderr streams to be unbuffered
@@ -61,8 +63,6 @@ ENV LLAMA_STACK_CONFIG_DIR=/.llama/data
 # Data and configuration
 RUN mkdir -p /.llama/distributions/ansible-chatbot
 RUN mkdir -p /.llama/data/distributions/ansible-chatbot
-ADD lightspeed-stack.yaml /.llama/distributions/ansible-chatbot
-ADD ansible-chatbot-run.yaml /.llama/distributions/ansible-chatbot
 RUN echo -e "\
 {\n\
   \"version\": \"${ANSIBLE_CHATBOT_VERSION}\" \n\

Makefile

@@ -10,9 +10,9 @@ ANSIBLE_CHATBOT_INFERENCE_MODEL_FILTER ?=
 LLAMA_STACK_PORT ?= 8321
 LOCAL_DB_PATH ?= .
 CONTAINER_DB_PATH ?= /.llama/data/distributions/ansible-chatbot
-# quay.io/ansible/aap-rag-content:latest does not work with lightspeed-stack:latest
-# aap-rag-content uses llama-stack:0.2.14 whereas lightspeed-stack:latest uses 0.2.13.
-RAG_CONTENT_IMAGE ?= quay.io/ansible/aap-rag-content:1.0.1751985495
+RAG_CONTENT_IMAGE ?= quay.io/ansible/aap-rag-content:latest
+LIGHTSPEED_STACK_CONFIG ?= lightspeed-stack.yaml
+LLAMA_STACK_RUN_CONFIG ?= ansible-chatbot-run.yaml
 # Colors for terminal output
 RED := \033[0;31m
 NC := \033[0m # No Color
@@ -23,6 +23,8 @@ NC := \033[0m # No Color
 
 UV_HTTP_TIMEOUT=120
 
+PLATFORM ?= "linux/amd64"
+
 help:
 	@echo "Makefile for Ansible Chatbot Stack"
 	@echo "Available targets:"
@@ -61,7 +63,7 @@ setup-vector-db:
 	@echo "Setting up vector db and embedding image..."
 	rm -rf ./vector_db ./embeddings_model
 	mkdir -p ./vector_db
-	docker run -d --rm --name rag-content $(RAG_CONTENT_IMAGE) sleep infinity
+	docker run --platform $(PLATFORM) -d --rm --name rag-content $(RAG_CONTENT_IMAGE) sleep infinity
 	docker cp rag-content:/rag/llama_stack_vector_db/faiss_store.db.gz ./vector_db/aap_faiss_store.db.gz
 	docker cp rag-content:/rag/embeddings_model .
 	docker kill rag-content
@@ -76,8 +78,9 @@ check-env-build:
 
 build: check-env-build
 	@echo "Building customized Ansible Chatbot Stack image from lightspeed-core/lightspeed-stack..."
-	docker build -f ./Containerfile \
+	docker build --platform $(PLATFORM) -f ./Containerfile \
 		--build-arg ANSIBLE_CHATBOT_VERSION=$(ANSIBLE_CHATBOT_VERSION) \
+		--build-arg LLAMA_STACK_RUN_CONFIG=$(LLAMA_STACK_RUN_CONFIG) \
 		-t ansible-chatbot-stack:$(ANSIBLE_CHATBOT_VERSION) .
 	@printf "Custom image $(RED)ansible-chatbot-stack:$(ANSIBLE_CHATBOT_VERSION)$(NC) built successfully.\n"
 
@@ -104,10 +107,11 @@ run: check-env-run
 	@echo "Running Ansible Chatbot Stack container..."
 	@echo "Using vLLM URL: $(ANSIBLE_CHATBOT_VLLM_URL)"
 	@echo "Using inference model: $(ANSIBLE_CHATBOT_INFERENCE_MODEL)"
-	docker run --security-opt label=disable -it -p $(LLAMA_STACK_PORT):8080 \
+	docker run --platform $(PLATFORM) --security-opt label=disable -it -p $(LLAMA_STACK_PORT):8080 \
 	  -v ./embeddings_model:/.llama/data/embeddings_model \
 	  -v ./vector_db/aap_faiss_store.db:$(CONTAINER_DB_PATH)/aap_faiss_store.db \
-	  -v ./lightspeed-stack.yaml:/.llama/data/lightspeed-stack.yaml \
+	  -v ./$(LIGHTSPEED_STACK_CONFIG):/.llama/distributions/ansible-chatbot/config/lightspeed-stack.yaml \
+	  -v ./$(LLAMA_STACK_RUN_CONFIG):/.llama/distributions/llama-stack/config/ansible-chatbot-run.yaml \
 	  -v ./ansible-chatbot-system-prompt.txt:/.llama/distributions/ansible-chatbot/system-prompts/default.txt \
 	  --env VLLM_URL=$(ANSIBLE_CHATBOT_VLLM_URL) \
 	  --env VLLM_API_TOKEN=$(ANSIBLE_CHATBOT_VLLM_API_TOKEN) \
@@ -140,11 +144,13 @@ run-local-db: check-env-run-local-db
 	@echo "Using inference model: $(ANSIBLE_CHATBOT_INFERENCE_MODEL)"
 	@echo "Using inference model for tools filtering : $(ANSIBLE_CHATBOT_INFERENCE_MODEL_FILTER)"
 	@echo "Mapping local DB from $(LOCAL_DB_PATH) to $(CONTAINER_DB_PATH)"
-	docker run --security-opt label=disable -it -p $(LLAMA_STACK_PORT):8080 \
+	docker run --platform $(PLATFORM) --security-opt label=disable -it -p $(LLAMA_STACK_PORT):8080 \
 	  -v $(LOCAL_DB_PATH):$(CONTAINER_DB_PATH) \
 	  -v ./embeddings_model:/app/embeddings_model \
 	  -v ./vector_db/aap_faiss_store.db:$(CONTAINER_DB_PATH)/aap_faiss_store.db \
-	  -v ./lightspeed-stack.yaml:/.llama/data/lightspeed-stack.yaml \
+	  -v ./$(LIGHTSPEED_STACK_CONFIG):/.llama/distributions/ansible-chatbot/config/lightspeed-stack.yaml \
+	  -v ./$(LLAMA_STACK_RUN_CONFIG):/.llama/distributions/llama-stack/config/ansible-chatbot-run.yaml \
+	  -v ./ansible-chatbot-system-prompt.txt:/.llama/distributions/ansible-chatbot/system-prompts/default.txt \
 	  --env VLLM_URL=$(ANSIBLE_CHATBOT_VLLM_URL) \
 	  --env VLLM_API_TOKEN=$(ANSIBLE_CHATBOT_VLLM_API_TOKEN) \
 	  --env INFERENCE_MODEL=$(ANSIBLE_CHATBOT_INFERENCE_MODEL) \

README.md

@@ -95,10 +95,13 @@ Builds the image `ansible-chatbot-stack:$ANSIBLE_CHATBOT_VERSION`.
 ```commandline
 └── .llama/
     ├── distributions/
+    │   └── llama-stack/
+    │       └── config
+    │           └── ansible-chatbot-run.yaml
     │   └── ansible-chatbot/
-    │       ├── lightspeed-stack.yaml
-    │       ├── ansible-chatbot-run.yaml
-    │       ├── ansible-chatbot-version-info.json
+    │       ├── ansible-chatbot-version-info.json    
+    │       └── config
+    │           └── lightspeed-stack.yaml
     │       └── system-prompts/
     │           └── default.txt
     └── providers.d

ansible-chatbot-mcp-run.yaml

@@ -0,0 +1,117 @@
+version: '2'
+image_name: ansible-chatbot
+container_image: ansible-chatbot
+apis:
+- inference
+- vector_io
+- safety
+- agents
+- datasetio
+- telemetry
+- tool_runtime
+providers:
+  inference:
+  - provider_id: my_rhoai_dev
+    provider_type: remote::vllm
+    config:
+      url: ${env.VLLM_URL}
+      max_tokens: ${env.VLLM_MAX_TOKENS:=4096}
+      api_token: ${env.VLLM_API_TOKEN:=fake}
+      tls_verify: ${env.VLLM_TLS_VERIFY:=true}
+  - provider_id: inline_sentence-transformer
+    provider_type: inline::sentence-transformers
+    config: {}
+  vector_io:
+  - provider_id: aap_faiss
+    provider_type: inline::faiss
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.VECTOR_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/aap_faiss_store.db
+  safety:
+  - provider_id: llama-guard
+    provider_type: inline::llama-guard
+    config:
+      excluded_categories: []
+  agents:
+  - provider_id: lightspeed_inline_agent
+    provider_type: inline::lightspeed_inline_agent
+    config:
+      persistence_store:
+        type: sqlite
+        namespace: null
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/agents_store.db
+      responses_store:
+        type: sqlite
+        namespace: null
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/responses_store.db
+      tools_filter:
+        enabled: true
+        model_id: ${env.INFERENCE_MODEL_FILTER:=}
+  datasetio:
+  - provider_id: localfs
+    provider_type: inline::localfs
+    config:
+      kvstore:
+        type: sqlite
+        namespace: null
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/localfs_datasetio.db
+  telemetry:
+  - provider_id: meta-reference
+    provider_type: inline::meta-reference
+    config:
+      service_name: ${env.OTEL_SERVICE_NAME:=ansible-chatbot-stack}
+      sinks: ${env.TELEMETRY_SINKS:=console,sqlite}
+      sqlite_db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/trace_store.db
+  tool_runtime:
+  - provider_id: rag-runtime
+    provider_type: inline::rag-runtime
+    config: {}
+  - provider_id: model-context-protocol
+    provider_type: remote::model-context-protocol
+    config: {}
+metadata_store:
+  namespace: null
+  type: sqlite
+  db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/registry.db
+models:
+- metadata: {}
+  model_id: ${env.INFERENCE_MODEL}
+  provider_id: my_rhoai_dev
+  provider_model_id: null
+- metadata:
+    embedding_dimension: 768
+  model_id: ${env.EMBEDDINGS_MODEL:=/.llama/data/distributions/ansible-chatbot/embeddings_model}
+  provider_id: inline_sentence-transformer
+  model_type: embedding
+shields: []
+vector_dbs:
+- metadata: {}
+  vector_db_id: "aap-product-docs-2_5"
+  embedding_model: ${env.EMBEDDINGS_MODEL:=/.llama/data/distributions/ansible-chatbot/embeddings_model}
+  embedding_dimension: 768
+  provider_id: "aap_faiss"
+datasets: []
+scoring_fns: []
+benchmarks: []
+tool_groups:
+  - toolgroup_id: builtin::rag
+    provider_id: rag-runtime
+  - toolgroup_id: mcp::aap-controller
+    provider_id: model-context-protocol
+    mcp_endpoint:
+      uri: http://localhost:8004/sse
+  - toolgroup_id: mcp::aap-lightspeed
+    provider_id: model-context-protocol
+    mcp_endpoint:
+      uri: http://localhost:8005/sse
+logging: null
+server:
+  port: 8321
+  tls_certfile: null
+  tls_keyfile: null
+  tls_cafile: null
+  auth: null
+  disable_ipv6: false
+external_providers_dir: ${env.EXTERNAL_PROVIDERS_DIR:=/.llama/providers.d}

ansible-chatbot-run.yaml

@@ -15,9 +15,9 @@ providers:
     provider_type: remote::vllm
     config:
       url: ${env.VLLM_URL}
-      max_tokens: ${env.VLLM_MAX_TOKENS:4096}
-      api_token: ${env.VLLM_API_TOKEN:fake}
-      tls_verify: ${env.VLLM_TLS_VERIFY:true}
+      max_tokens: ${env.VLLM_MAX_TOKENS:=4096}
+      api_token: ${env.VLLM_API_TOKEN:=fake}
+      tls_verify: ${env.VLLM_TLS_VERIFY:=true}
   - provider_id: inline_sentence-transformer
     provider_type: inline::sentence-transformers
     config: {}
@@ -28,7 +28,7 @@ providers:
       kvstore:
         type: sqlite
         namespace: null
-        db_path: ${env.VECTOR_DB_DIR:/.llama/data/distributions/ansible-chatbot}/aap_faiss_store.db
+        db_path: ${env.VECTOR_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/aap_faiss_store.db
   safety:
   - provider_id: llama-guard
     provider_type: inline::llama-guard
@@ -41,29 +41,29 @@ providers:
       persistence_store:
         type: sqlite
         namespace: null
-        db_path: ${env.PROVIDERS_DB_DIR:/.llama/data/distributions/ansible-chatbot}/agents_store.db
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/agents_store.db
       responses_store:
         type: sqlite
         namespace: null
-        db_path: ${env.PROVIDERS_DB_DIR:/.llama/data/distributions/ansible-chatbot}/responses_store.db
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/responses_store.db
       tools_filter:
         enabled: true
-        model_id: ${env.INFERENCE_MODEL_FILTER:}
+        model_id: ${env.INFERENCE_MODEL_FILTER:=}
   datasetio:
   - provider_id: localfs
     provider_type: inline::localfs
     config:
       kvstore:
         type: sqlite
         namespace: null
-        db_path: ${env.PROVIDERS_DB_DIR:/.llama/data/distributions/ansible-chatbot}/localfs_datasetio.db
+        db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/localfs_datasetio.db
   telemetry:
   - provider_id: meta-reference
     provider_type: inline::meta-reference
     config:
-      service_name: ${env.OTEL_SERVICE_NAME:}
-      sinks: ${env.TELEMETRY_SINKS:console,sqlite}
-      sqlite_db_path: ${env.PROVIDERS_DB_DIR:/.llama/data/distributions/ansible-chatbot}/trace_store.db
+      service_name: ${env.OTEL_SERVICE_NAME:=ansible-chatbot-stack}
+      sinks: ${env.TELEMETRY_SINKS:=console,sqlite}
+      sqlite_db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/trace_store.db
   tool_runtime:
   - provider_id: rag-runtime
     provider_type: inline::rag-runtime
@@ -74,22 +74,22 @@ providers:
 metadata_store:
   namespace: null
   type: sqlite
-  db_path: ${env.PROVIDERS_DB_DIR:/.llama/data/distributions/ansible-chatbot}/registry.db
+  db_path: ${env.PROVIDERS_DB_DIR:=/.llama/data/distributions/ansible-chatbot}/registry.db
 models:
 - metadata: {}
   model_id: ${env.INFERENCE_MODEL}
   provider_id: my_rhoai_dev
   provider_model_id: null
 - metadata:
     embedding_dimension: 768
-  model_id: ${env.EMBEDDINGS_MODEL:/.llama/data/distributions/ansible-chatbot/embeddings_model}
+  model_id: ${env.EMBEDDINGS_MODEL:=/.llama/data/distributions/ansible-chatbot/embeddings_model}
   provider_id: inline_sentence-transformer
   model_type: embedding
 shields: []
 vector_dbs:
 - metadata: {}
   vector_db_id: "aap-product-docs-2_5"
-  embedding_model: ${env.EMBEDDINGS_MODEL:/.llama/data/distributions/ansible-chatbot/embeddings_model}
+  embedding_model: ${env.EMBEDDINGS_MODEL:=/.llama/data/distributions/ansible-chatbot/embeddings_model}
   embedding_dimension: 768
   provider_id: "aap_faiss"
 datasets: []
@@ -106,4 +106,4 @@ server:
   tls_cafile: null
   auth: null
   disable_ipv6: false
-external_providers_dir: ${env.EXTERNAL_PROVIDERS_DIR:/.llama/providers.d}
+external_providers_dir: ${env.EXTERNAL_PROVIDERS_DIR:=/.llama/providers.d}

entrypoint.sh

@@ -23,4 +23,4 @@ else
   fi
 fi
 
-python3.12 /app-root/src/lightspeed_stack.py --config /.llama/distributions/ansible-chatbot/lightspeed-stack.yaml
+python3.12 /app-root/src/lightspeed_stack.py --config /.llama/distributions/ansible-chatbot/config/lightspeed-stack.yaml

lightspeed-stack-mcp.yaml

@@ -0,0 +1,23 @@
+name: Ansible Lightspeed Intelligent Assistant
+service:
+  host: 0.0.0.0
+  port: 8080
+  auth_enabled: false
+  workers: 1
+  color_log: true
+  access_log: true
+llama_stack:
+  use_as_library_client: true
+  library_client_config_path: /.llama/distributions/llama-stack/config/ansible-chatbot-run.yaml
+user_data_collection:
+  feedback_disabled: true
+  transcripts_disabled: true
+customization:
+  system_prompt_path: /.llama/distributions/ansible-chatbot/system-prompts/default.txt
+mcp_servers:
+- name: mcp::aap-controller
+  provider_id: model-context-protocol
+  url: http://localhost:8004/sse
+- name: mcp::aap-lightspeed
+  provider_id: model-context-protocol
+  url: http://localhost:8005/sse

lightspeed-stack.yaml

@@ -8,7 +8,7 @@ service:
   access_log: true
 llama_stack:
   use_as_library_client: true
-  library_client_config_path: /.llama/distributions/ansible-chatbot/ansible-chatbot-run.yaml
+  library_client_config_path: /.llama/distributions/llama-stack/config/ansible-chatbot-run.yaml
 user_data_collection:
   feedback_disabled: true
   transcripts_disabled: true

pyproject.toml

@@ -8,7 +8,8 @@ dependencies = [
     "aiosqlite~=0.21.0",
     "faiss-cpu~=1.11.0",
     "fire~=0.7.0",
-    "lightspeed-stack-providers~=0.1.9",
+    "lightspeed-stack-providers~=0.1.10",
+    "llama-stack==0.2.14",
     "mcp~=1.9.4",
     "numpy==2.2.6",
     "opentelemetry-api~=1.34.1",