chore: Use Jina optionaly

Author: GoodbyePlanet

Makefile

@@ -2,7 +2,7 @@ QDRANT_URL      := http://localhost:6333
 SEMCODE_URL     := http://localhost:8090
 
 .PHONY: qdrant-clean qdrant-dashboard index-code index-history \
-        docker-build-restart docker-up docker-logs docker-logs-semcode
+        docker-build-restart docker-build-restart-jina docker-up docker-up-jina docker-logs docker-logs-semcode
 
 qdrant-clean:
 	curl -sf -X DELETE $(QDRANT_URL)/collections/code_symbols && \
@@ -25,9 +25,15 @@ index-history:
 docker-build-restart:
 	docker compose down && docker compose up --build -d
 
+docker-build-restart-jina:
+	docker compose --profile jina down && docker compose --profile jina up --build -d
+
 docker-up:
 	docker compose up -d
 
+docker-up-jina:
+	docker compose --profile jina up -d
+
 docker-logs:
 	docker compose logs -f
 

README.md

@@ -64,16 +64,25 @@ code, etc.).
 ## Running
 
 ```bash
+# Jina embeddings (default — local TEI container)
+docker-compose --profile jina up
+# or: make docker-up-jina
+
+# Hosted provider (Voyage / OpenAI) or local Ollama — no TEI container
 docker-compose up
+# or: make docker-up
 ```
 
-This starts three services with health checks and persistent volumes:
+`jina-embeddings` only starts when the `jina` profile is active. For any other provider, omit
+`--profile jina` and set `EMBEDDINGS_PROVIDER` plus the relevant API key in `.env`.
+
+Services started with health checks and persistent volumes:
 
-| Service                   | Port                         | Volume                           | Purpose                |
-|---------------------------|------------------------------|----------------------------------|------------------------|
-| **Qdrant**                | `6333` (HTTP), `6334` (gRPC) | `qdrant_data`                    | Vector DB              |
-| **Jina Embeddings** (TEI) | `8087`                       | `embeddings_cache`               | Embedding model server |
-| **semcode MCP**           | `8090`                       | mounts `./config.yaml` read-only | MCP + HTTP server      |
+| Service                    | Profile | Port                         | Volume                           | Purpose                |
+|----------------------------|---------|------------------------------|----------------------------------|------------------------|
+| **Qdrant**                 | always  | `6333` (HTTP), `6334` (gRPC) | `qdrant_data`                    | Vector DB              |
+| **Jina Embeddings** (TEI)  | `jina`  | `8087`                       | `embeddings_cache`               | Embedding model server |
+| **semcode MCP**            | always  | `8090`                       | mounts `./config.yaml` read-only | MCP + HTTP server      |
 
 The MCP server starts with empty collections — trigger an initial index by calling the `reindex` MCP tool
 or `POST /reindex` (see below).
@@ -200,10 +209,9 @@ the existing Qdrant collection, the server fails fast at startup with a clear er
 offending collection. To switch, drop both collections (`code_symbols` and `git_commits`) via the
 Qdrant UI or API, then reindex. There is no automatic migration.
 
-**Hosted-only setup (no local TEI container):** in `docker-compose.yaml`, comment out the entire
-`jina-embeddings` service block, the `jina-embeddings` entry under `semcode.depends_on`, and the
-`JINA_URL` line under `semcode.environment`. Then set `EMBEDDINGS_PROVIDER` and the relevant API
-key in `.env`.
+**Hosted-only setup (no local TEI container):** set `EMBEDDINGS_PROVIDER` and the relevant API key
+in `.env`, then start without the `jina` profile (`docker-compose up` / `make docker-up`). The
+`jina-embeddings` container will not start.
 
 ## Qdrant collections
 

docker-compose.yaml

@@ -13,13 +13,11 @@ services:
       retries: 5
 
   # Jina Code V2 served by HuggingFace Text Embeddings Inference. Default
-  # embedding backend. To use a hosted provider instead (Voyage / OpenAI) or a
-  # local Ollama install, set EMBEDDINGS_PROVIDER + the relevant API key in
-  # .env, then comment out:
-  #   1. this entire `jina-embeddings` service
-  #   2. the `jina-embeddings` entry under `semcode.depends_on`
-  #   3. the JINA_URL line under `semcode.environment`
+  # embedding backend — only started when the `jina` profile is active.
+  # For hosted providers (Voyage / OpenAI) or a local Ollama install, omit
+  # --profile jina and set EMBEDDINGS_PROVIDER + the relevant API key in .env.
   jina-embeddings:
+    profiles: [jina]
     image: ghcr.io/huggingface/text-embeddings-inference:cpu-1.6
     platform: linux/amd64
     container_name: jina-embeddings
@@ -51,6 +49,7 @@ services:
         condition: service_healthy
       jina-embeddings:
         condition: service_healthy
+        required: false
     volumes:
       - ./config.yaml:/app/config.yaml:ro
       - fastembed_cache:/fastembed_cache