Refactor/Cleanup

Dockerfile

@@ -8,14 +8,10 @@ RUN apt-get update \
 
 WORKDIR /opt/llmize
 
-ARG INSTALL_ENRICH=false
-RUN python3 -m pip install --no-cache-dir --break-system-packages "ollama>=0.5" \
- && if [ "$INSTALL_ENRICH" = "true" ]; then \
-        python3 -m pip install --no-cache-dir --break-system-packages tooluniverse "PyYAML>=6" ; \
-    fi
+RUN python3 -m pip install --no-cache-dir --break-system-packages "ollama>=0.5"
 
-# TEST_MODEL is baked in for offline runs; LLMIZE_MODEL is the runtime default.
-# Set BAKE_MODEL=true to also bake LLMIZE_MODEL for air-gapped use.
+# TEST_MODEL is baked in for offline runs, LLMIZE_MODEL is the runtime default.
+# Set BAKE_MODEL=true to also bake LLMIZE_MODEL for isolated use.
 ARG TEST_MODEL=smollm2:135m
 ARG LLMIZE_MODEL=gemma4
 ENV LLMIZE_MODEL=${LLMIZE_MODEL}

README.md

@@ -25,9 +25,8 @@ source .venv/bin/activate
 python3 -m pip install -r requirements.txt
 ```
 
-This installs `ollama` plus the optional enrich stack
-(`tooluniverse`, `PyYAML`). Enrich is optional, the pipeline can run without
-it but enrich gives more context.
+This installs `ollama` plus the optional enrich stack (`tooluniverse`, `PyYAML`).
+Enrich is optional, the pipeline can run withoutit but enrich gives more context.
 
 ### 3. Run the pipeline
 ```bash

check_env.py

@@ -73,32 +73,6 @@ def _ollama_server_and_models() -> list:
     return [server_ok, model_check]
 
 
-def _tooluniverse() -> Check:
-    if importlib.util.find_spec("tooluniverse") is None:
-        return Check(
-            "ToolUniverse (optional, --enrich)", WARN,
-            "not installed; gene enrichment will be skipped. "
-            "Install with: python3 -m pip install tooluniverse PyYAML",
-            required=False,
-        )
-    try:
-        from enrich import EntityEnricher
-        enricher = EntityEnricher()
-    except Exception as exc:
-        return Check(
-            "ToolUniverse (optional, --enrich)", WARN,
-            f"installed but failed to initialise ({type(exc).__name__}); enrichment will be skipped.",
-            required=False,
-        )
-    if enricher.available:
-        return Check("ToolUniverse (optional, --enrich)", OK, "ready (gene-lookup tool resolved)", required=False)
-    return Check(
-        "ToolUniverse (optional, --enrich)", WARN,
-        "installed but no gene-lookup tool matched. Set LLMIZE_GENE_TOOL to a valid tool name.",
-        required=False,
-    )
-
-
 def _descriptor_schema() -> Check:
     path = os.path.join(PROJECT_ROOT, "json_reduction", "descriptor_schema.json")
     if os.path.exists(path):
@@ -126,7 +100,6 @@ def _data_dir_writable() -> Check:
 def run_checks() -> list:
     checks = [_python_version(), _ollama_package()]
     checks += _ollama_server_and_models()
-    checks.append(_tooluniverse())
     checks.append(_descriptor_schema())
     checks.append(_data_dir_writable())
     return checks

docker/boot_ollama.sh

@@ -1,19 +1,12 @@
 #!/usr/bin/env bash
-# Start the in-container Ollama server, wait for it, and pull the model.
-#
-# Reusable so both the image ENTRYPOINT (docker run) and the Nextflow process
-# script can call it. The server is started with nohup + disown so it survives
-# this script exiting and is reachable by later commands in the same task.
-#
-# Honours:
-#   LLMIZE_MODEL  - model tag to pull/run (default: gemma4)
-#   OLLAMA_HOST   - server/client endpoint (default: 127.0.0.1:11434)
+# Start the in-container Ollama server and pull the model. nohup + disown let the
+# server survive this script exiting so later commands in the task can reach it.
+# Honours LLMIZE_MODEL (default gemma4) and OLLAMA_HOST (default 127.0.0.1:11434).
 set -euo pipefail
 
 MODEL="${LLMIZE_MODEL:-gemma4}"
 ENDPOINT="http://${OLLAMA_HOST:-127.0.0.1:11434}"
 
-# Start the server only if one isn't already answering.
 if ! curl -sf "${ENDPOINT}/api/tags" >/dev/null 2>&1; then
     echo "[boot] starting 'ollama serve'..."
     nohup ollama serve >/tmp/ollama.log 2>&1 &
@@ -33,9 +26,8 @@ for i in $(seq 1 60); do
     fi
 done
 
-# Skip the pull if the model is already present (baked into the image, or cached
-# on a mounted volume). This is what makes an offline / air-gapped image work:
-# 'ollama pull' would otherwise contact the registry even for an existing model.
+# Skip the pull when the model is already present (baked in or volume-cached) so the
+# image works offline; 'ollama pull' would otherwise contact the registry regardless.
 if ollama list 2>/dev/null | grep -qF "${MODEL}"; then
     echo "[boot] model '${MODEL}' already present; skipping pull (offline-safe)."
 else

docker/entrypoint.sh

@@ -5,7 +5,6 @@ set -euo pipefail
 
 source /opt/llmize/docker/boot_ollama.sh
 
-# Default to the env preflight if no command was given.
 if [ "$#" -eq 0 ]; then
     exec python3 /opt/llmize/pipeline.py --check
 fi